1 · What the track record is
MARS (the Montex AlphaRail System) is a trading framework. The trader picks each trade by hand, and a set of quantitative rules decides how much can be risked. The Equilibrium Track Record is the public scoreboard for one live test of that framework:
- One fixed strategy, 1:1. Every trade has a hard stop and a take-profit at the same distance. A win earns about the amount that was risked; a loss costs about the same.
- Four trades a day, four days a week (16 trades a week), with the same risk on all four trades of a day.
- 78 weeks (18 months), starting Monday 16 November 2026, unless the trader stops it earlier.
- No dollars on the page. Growth is shown as a % return and results are shown in R (explained below). Account size, position sizes and account numbers are never published.
The page updates from the trader's journal. A main menu has seven sections, each answering one question: Home (how is it going?), Edge (is the edge real?), Risk (can it survive?), Performance (where does the edge come from?), Execution (is the plan followed?), Lab (what if?) and Records (show me everything), plus the Glossary. Hands-on pages let you filter every trade (Explore), run your own Monte Carlo test (Simulator), use the decision and statistical tools, and search, pin, compare and share with the everyday tools. Section 4 maps every page and explains the tools, section 5 explains charts 1–20, section 6 the 39 Analytics charts, and section 8 the trading plan behind every trade.
Three ideas that make every page easier to read
1R is the amount risked on a trade. A full win at 1:1 is about +1R and a full loss is about −1R. Measuring in R lets you compare trades of different sizes fairly, and it keeps dollar figures private.
Every R figure on the page is net: spread, commission and slippage are already taken out. That's why a "full win" usually shows a little under +1R and a full loss a little worse than −1R.
A trade that closes between −0.10R and +0.10R counts as breakeven (BE), not as a win or a loss. Tiny scratches don't move the win rate.
A computer runs the same plan (1:1, 60% win rate, same risk rules) thousands of times to show the range of normal outcomes. It's a ruler, not a target: it shows whether real results are inside, above or below what the plan should produce by luck alone.
2 · What to focus on: the 2-minute check
If you only have two minutes, read these five things on the Home page, in this order. They follow MARS's own order of authority: edge first, survival second, everything else after.
- Sample size and verdict. How many trades are counted? Under 30 trades, every status is "Provisional" and the verdict reads COLLECT DATA. That's normal, not a warning. The verdict stays provisional until 300 trades. Before that, treat every number as a first impression.
- EV (expectancy): is there an edge? The average net R per trade. The plan expects about +0.15R after costs. Above zero means the strategy makes money on average. This is the master metric; everything else is secondary.
- Max drawdown and the gate: can it survive? Drawdown is how far the account has fallen from its highest point. The gate (Growth, Recovery, Buffer, …) shows which safety level the drawdown has triggered. Deeper drawdown means smaller risk, automatically.
- Return vs the 1:1 benchmark: on schedule? Is the account's % return inside, above or below the shaded Monte Carlo band? Above the middle line with a similar or smaller drawdown is what MARS calls alpha.
- All-in cost: is cost eating the edge? The average cost per trade in R. At 1:1, cost matters a lot: a cost of 0.05R uses a third of a +0.15R edge.
The status chips
Most tiles and table rows carry a coloured chip. Each chip has a colour, a dot and a word, so you can read them without relying on colour.
| Chip | Meaning |
|---|---|
| Ahead | Better than the benchmark middle (the top quarter of normal outcomes, or above plan). |
| On pace | Inside the normal range for the plan. This is where a healthy test spends most of its time. |
| Watch | Slipping toward the low side. Not a problem yet, but worth following. |
| Behind | Below the normal range (the bottom quarter of outcomes, or under plan). |
| Provisional | Too few trades to judge. Shown on everything until 30 trades. |
The colour key
Besides the five status chips, the page colours the words that describe the account's state, so a glance tells you where things stand. The word is always written out; the colour and the small icon only help. Icons appear on the larger pills and become plain dots inside tables.
| What | The colours | Meaning |
|---|---|---|
| Gate Risk page, header | 🌱 Growth🔄 Recovery🛡️ Buffer⚠️ Floor🚨 Deep-Floor🚨 Ground-Floor🔒 System Lock | The safety level set by how far the account is below its peak. Green is near the peak; amber and red mean deeper drawdown and smaller risk. System Lock means no trading. Risk › Gates and tiers lists the exact levels. |
| Mode Header, Risk | 🐢 Conservative⚖️ Balanced🔥 Aggressive | How much risk the rules allow at most: amber is cautious, cyan is the live setting, red is the highest cap. |
| Tier T1 to T7 | T1T2T3T4T5T6T7 | The risk size for the day, cool to hot: blue is the smallest, magenta the largest. |
| Day colour | ⭐ Top Green🟢 Green🟡 Yellow🔴 Red | How a day of four trades went. It sets the next day's starting tier. |
| Results and direction | WinLossBELongShort | Win, loss or break-even (within ±0.10R); long or short. Numbers with a + show green and with a − show red. |
| Alerts and checks | ✅ Clear🚨 Triggered✅ PASS🚨 BREACH | Daily and weekly loss limits (Clear or Triggered) and the review checklist result. |
- One day or one week. At a 60% win rate, about 1 in 6 full trading days still ends with only 0 or 1 wins out of 4. That's normal (see chart 9).
- Losing streaks under about 6. Over 78 weeks a 60% strategy will almost certainly hit several runs of 5 or more losses in a row.
- Breakdowns with only a few trades (e.g. "Tuesday EV +0.6R" from 5 trades). Small slices are mostly luck.
- EV that stays below zero after 100 or more trades, especially if even the top of its 95% range is below zero.
- A win rate stuck below the break-even win rate (shown under Edge › Edge statistics and Performance › Cost).
- Drawdown passing the deeper gates (Floor at 19%, Deep-Floor at 27%) and staying there.
- Cost using up most of the gross edge (Cost status turns to a warning).
3 · The Home page, piece by piece
The banner
The banner at the top of every page shows the verdict and the reason for it, the risk mode, the current week of 78, the sample (how many trades so far), the schedule (where the account sits against the 1:1 benchmark fan), the current gate, the risk per trade for the next day and the cost status.
The verdict follows EV and its 95% range. Under 30 trades it reads COLLECT DATA. After that it moves through levels from strong to weak. It's guidance for review: no verdict level stops trading on its own. Only the System Lock (drawdown beyond 40%) forces risk to zero; otherwise the trader decides whether to continue.
The nine tiles
| Tile | What it means | What to compare it with |
|---|---|---|
| Return since start | Account growth in % since the first counted trade. | The benchmark band on the return chart. |
| EV (avg net R) | Average result per trade after costs. The master metric. | Plan ≈ +0.15R. Above 0 = edge. |
| Win rate | Share of trades that are wins (breakevens are not counted as wins). | Plan 60%. It must stay above the break-even win rate. |
| Max drawdown | The largest fall from an equity peak so far, in %. | The gates at 7 / 13 / 19 / 27 / 34 / 40%. |
| RAER | Risk-adjusted efficiency ratio: total profit ÷ total risk put on. How well risk turns into return. | Above 0 is good. Compare with the benchmark P25–P75. |
| R profit factor | R won on winning trades ÷ R lost on losing trades. | 1.50 at plan before costs. Below 1.0 = losing money. |
| Acceleration | This week's % return minus last week's. Is compounding speeding up or slowing down? | Swings a lot week to week. Read the trend, not one value. |
| All-in cost | Average total cost per trade (spread + commission + slippage), in R. | Smaller is better. Compare with the gross edge. |
| Trades | Trades counted, plus the number of planned slots that were skipped. | 16 planned per week. |
Five core metrics table
This is the heart of the page. Each of the five MARS core metrics (EV, drawdown, RAER, RAPF and equity) is shown three ways: You (the real result), the 1:1 benchmark P25 / P50 / P75 (the low-normal, middle and high-normal results of the simulation at the same number of trades), and two judgements, vs 1:1 and vs 7-Tier. The 7-Tier column is a second ruler built from the full MARS seven-tier risk schedule.
Reading P25 / P50 / P75: imagine 100 simulated traders running the same plan. P50 is the one in the middle. P25 did worse than 75 of them, and P75 did better than 75 of them. Landing between P25 and P75 is perfectly normal.
Next day line and verdict detail
The Next day line under the header (on every page) gives the next trading day and the risk per trade the rail has set. Its details link opens Risk › Tomorrow, which shows the rest: drawdown from peak, the gate, the last day's colour, last week's status, the tier, the pool (total risk for all four trades), a one-line reason, the daily and weekly loss warnings (warnings only; they never block a trade) and what each possible day would do to the day after.
Verdict detail explains the verdict: the win rate and EV with their 95% ranges, the break-even win rate, how many trades are needed to prove the EV, and whether trades are keeping pace with the plan. Section 7 explains each of these.
Road to week 78
A track from the start to the finish flag of the 78-week test, in % only (log scale, so early and late progress both show). Stations mark doubling, ×5, ×10, ×25 and ×50; the finish flag is the end of the plan-pace line on the return chart. The blue dot is the account's return now; the orange marker is where the plan pace says it should be this week. A chip says roughly how many weeks ahead of or behind plan pace the account is, and a bar shows how much of the 78 weeks has passed.
Replay the record
Press Play to watch the record build itself trade by trade: the cumulative R line grows, the red shading underneath shows how far it sits below its peak, and the strip below shows each trade's date, pair, result and the gate in force. Choose a speed, pause at any time, or drag the slider to jump to any trade. It's the quickest way to feel the rhythm of wins, losses and recoveries.
Achievements
80 badges in 11 families unlock automatically from the record. The Home page shows the three most recent and the three closest to unlocking; Records › Achievements shows them all. Process families come first and are drawn larger, because they are what the trader controls: Discipline (compliant trades in a row, zero-breach weeks, periods and seasons, A-grade trades and days), Resilience (new highs after 7% and 13% drawdowns, bounce-back days and weeks, a losing run traded fully by the rules) and Consistency (full 16-trade weeks, runs of them, a season with no missed day). Then Sample, Growth (multiples of the start, up to the ×110 goal), Streaks, Perfect days, Rail (each risk tier reached, the Growth gate held), Evidence (edge proven, edge confirmed on the sequential test, EV on plan at 100 and 300 trades, ahead of the benchmark), Cost and Integrity and seasons. A bronze, silver or gold ring marks the harder levels; locked badges are gray with a progress bar, earned ones show the date. No badge rewards trading more than the plan's 4 a day. Badges are for fun and motivation; they are not rules and change nothing in MARS.
4 · The menu: seven sections
The page opens on Home. A main menu has seven sections: on the website it runs down the left side of a desktop screen, and across the top on a narrower screen. Each section answers one question and holds a few pages, shown as a row of buttons above the page. After the sections, the menu lists the tools: ⌕ Search (or press /), ↺ Recent, Tour, ▶ Present, ⎙ Print, ✦ What's new, ★ My dashboard, Share card, the Glossary and a link to this guide. Each page has its own web address (for example #risk/outlook), so a link can point straight at one page; old links to the earlier tabs still open the right page.
| Section | Its question | Pages | Best used for |
|---|---|---|---|
| Home | How is it going? | Summary: since your last visit, banner, tiles, return vs benchmark, five core metrics, verdict detail, road to week 78, replay, achievements (section 3) · Five-metric report card. | The 2-minute check. |
| Edge | Is there an edge, and is it real? | Edge statistics (expectancy, edge-proof meter, cumulative R, monthly and weekly scorecards) · Proof charts (A1–A4) · Rolling windows (A5–A10) · Stability (change points, independence, streak odds) · Win-rate belief · Luck tests · Sequential test · Payoff map. | Answering "is the edge real, and how sure can we be?" |
| Risk | Can it survive? | Gates and tiers (gate ladder, rules of the test, daily record) · Tomorrow · Gate ladder and recovery (decision tools 4 and 5, with the loss runway) · Tier forecast · Growth vs risk (tool 6) · Equity and drawdown (A22–A28) · Pain and tail risk (with drawdown anatomy) · Rail value-added · Currency exposure · Day colours. | Checking risk was set by the rules, and how far today is from the next gate. |
| Performance | Where does the edge come from? | Explore (filters, saved views, calendar, Compare by, filter-trap meter) · Compare periods · Growth attribution · Breakdowns · Leaderboard · Benchmark rank · Volatility and news · Time and rhythm (A11–A16, time in trade) · Pairs and mix (A17–A21) · Cost (cost in R, cost heatmap and A29–A33) · Excursions and what-if (MFE / MAE, ghost exits, A38–A39). | Asking your own questions. Only trust slices with many trades. |
| Execution | Is the plan being followed? | Grades and compliance (A34–A37, from the per-trade reviews; section 8 explains the grading; with the checklist heatmap and focus follow-through) · Booster value · Fill quality · Breach cost (with the mistake library) · Tilt detector. | Seeing whether the record is a fair test of the plan. |
| Lab | What if? | Simulator (with MARS rail mode and A vs B) · Goal-seek · Next-week forecast · Stress tests · ETA board · Sample-size planner · Trading school (100 questions) · Play a week. | Building intuition. Research only: nothing here changes a rule. |
| Records | Show me everything. | Trade log (search, filters, trade cards, Download CSV) · Charts 1–20 (section 5) · Week stories · Timeline · Hall of fame and seasons · Achievements · Integrity (with the rule-change ledger). | Full transparency: every number traces back to these rows and charts. |
Every chart keeps its number wherever it appears. A chart can sit on its topic page and also in Records; it is the same chart.
Explore: filter, compare and the calendar
- Pick filters. Before the first counted trade the filter bar is shown greyed out with a gray Locked chip; it unlocks by itself once trades arrive. Each drop-down (Pair, Session, Timeframe, Direction, Weekday, Slot) starts at "All". The From and To boxes limit the dates. You can combine any of them, for example London + EURUSD + Tuesday.
- Read the count first. The line under the filters says how many trades match. Under 30 it shows a gray Provisional chip: the slice is too small to judge, however good or bad it looks.
- Read the eight tiles. Trades (with wins, losses and breakevens), win rate (breakevens left out), EV per trade, net R, R profit factor, average cost, worst slide (the deepest drop in cumulative R) and the longest losing run, all for the matching trades only.
- Compare. The "Compare by" chart splits the matching trades by any field (pair, session, timeframe, direction, weekday, slot, exit, tier or gate). Bars show EV per trade; the numbers on the right are trades and win rate.
- Use the calendar. One square per trading day, Monday to Friday, one row per cycle week. Green squares made R, red squares lost R, and darker means a bigger day. Every square also prints its net R and trade count, so colour is never the only clue. Dashed squares had no matching trades. Click a square to see that day's trades with its gate, tier, risk per trade and drawdown at the close; click again or press "Close day" to hide it.
- Sort the list. Click any column heading in "Matching trades" to sort by it; click again to reverse. "Reset filters" clears everything.
Simulator: the Monte Carlo playground
The Simulator plays thousands of imaginary 1:1 runs. In each run every trade is either a win (+1R minus cost) or a loss (−1R minus cost), drawn at random at the win rate you choose, with 16 trades a week. A run that falls 40% from its peak stops trading, like the System Lock.
| Control | What it does |
|---|---|
| Win rate | The share of trades that win, 40% to 75%. Plan: 60%. |
| Risk per trade | % of the account risked on every trade, 0.5% to 6%. It starts at the next day's real risk. |
| Cost per trade | Spread, fees and slippage, in R, taken off every trade. Plan: about 0.05R. |
| Weeks | How long each run lasts, up to the full 78 weeks. |
| Buttons | Run again draws a fresh set of runs. Plan settings resets to 60% and 0.05R. Use my measured stats copies the real win rate and cost, using whatever filters are set on Explore (so you can simulate "London only"). The run count (1,000 / 2,000 / 5,000) trades speed for smoothness. |
What comes out: EV per trade for your settings; the median outcome and the range that 90% of runs end inside; the share of runs that end above the start; the median worst drawdown; the share of runs that hit the System Lock; and Your rank: the share of simulated runs that the real account is beating at its latest week. Two charts follow: the fan (return by week on a log scale, so big gains and small losses both fit; shaded = middle 90% of runs, dashed lines = middle 50%, solid violet = the median run, blue = the real record) and worst drawdown per run (how deep runs fell from a peak at some point).
The decision tools
Seven tools turn the record into forward-looking questions. They live where their question belongs: four under Risk, three in the Lab. Until 30 trades are counted they use the plan's figures (60% win rate, 0.05R cost) and say so; from 30 trades they use the measured win rate and cost.
| Tool (where) | What it answers | How to read it |
|---|---|---|
| 1 · MARS rail mode Lab › Simulator | What do the gates and tiers do to the range of outcomes? Switch from Constant risk to MARS rail: each simulated day's risk then follows the gate cap, the last day's colour and the weekly status, as on the Gates page. | Compare the System Lock share and the worst drawdown between the two modes. A tile shows the average risk the rail actually used. |
| 2 · A vs B Lab › Simulator | What changes if one setting changes? Pin this run as A, change a slider or the risk mode, and a table compares A with the new run (B): EV, median, both tails, ends above start, worst drawdown, System Lock and average risk. A's median is drawn on the fan as a dotted line. | Read the B − A column. Change one thing at a time so the difference has one cause. |
| 3 · Goal-seek Lab › Goal-seek | What would it take? Pick a target (×1.25 to ×100) and a week, then solve for the win rate needed at a fixed risk, or the risk needed at a fixed win rate. The answer is checked by simulation. | About half of runs reach the target by design (the solver aims at the median run). A target that needs more than 60% or more than 6% risk is a warning, not a plan. |
| 4 · Gate distance ladder Risk › Gate ladder and recovery | How close is the next gate? From today's drawdown and risk, the chance of falling to each deeper gate within the next 16, 32 and 64 trades (one, two and four weeks), with the rail cutting risk on the way down. Checked at each day's close, as Equilibrium sets the gate. | High odds for Recovery are normal at full risk: two losses in a row reach it. Watch the Floor row and below. |
| 5 · Recovery clock Risk › Gate ladder and recovery | How long until a new equity high? Starting from today's drawdown (or a what-if, set with the slider): the median time back, the middle 80% of recovery times, and the chance of recovering within 4 and 13 weeks. | Deeper holes take much longer: a −20% fall needs +25% to recover, and the rail climbs back at lower risk. That is the price of survival. |
| 6 · Growth vs risk Risk › Growth vs risk | What does a fixed risk per trade do over 78 weeks, including the System Lock at −40%? Two charts: the median result by risk, and the chance of the lock by risk, plus the highest risk that keeps the lock at or under 1 in 20. | The textbook "growth-optimal" risk (Kelly) ignores the lock. With the lock included, fixed risk above a few percent locks most runs. That is why MARS cuts risk as drawdown deepens. |
| 7 · Next-week forecast Lab › Next-week forecast | Where is next week's net R likely to land? A cone for the next 16 trades (80% range and median), shown in R and in account % at today's risk, with the chance of a losing week. Below it, a scorecard checks each finished week against the range it would have had, using only the trades before that week. At the bottom, frozen forecasts saved before each week are scored without hindsight (Statistical depth III). | A well-calibrated forecast catches about 8 weeks in 10. Far fewer: the model is too confident. Nearly all: too cautious. Judge it after a quarter, not after two weeks. |
The statistical-depth tools
Ten more tools test whether what you see is real. Most need trades to work on: they show "fills in" messages until then, and a Provisional note while the sample is small.
| Tool (where) | What it answers | How to read it |
|---|---|---|
| Change-point detector Edge › Stability | Has the edge shifted? A line that adds up each trade's distance from the overall average (a CUSUM), for EV, win rate or cost per trade. Its peak or trough marks the most likely change; the value before and after is shown, with the chance a shift that strong is just luck (1,000 shuffles). See Statistical depth III. | A steady edge wanders around zero. Under 5% chance of luck is the usual bar, and a real shift still needs a reason. |
| Win/loss independence Edge › Stability | Does a loss make the next loss more likely? Win rate after a win and after a loss, a runs test, the lag-1 correlation and the longest streaks. | A clearly lower win rate after losses can point to tilt. Check it against the execution grades. |
| Win-rate belief Edge › Win-rate belief | What is the true win rate, probably? Updated trade by trade (Bayes), from the plan's 60% or from no opinion. Shows the best estimate, a 90% range, the chance it beats break-even and the chance it is at least 60%. | Watch the chance above break-even: 90%+ is strong evidence the win rate pays for its costs. |
| Coin-flip rival Edge › Luck tests | Could a trader with no edge, paying your costs, have done this well? 2,000 coin-flip rivals drawn against your cumulative R. | Beating 95% of rivals is the same bar as the EV's 95% range clearing zero. |
| Path luck Edge › Luck tests | How much of the worst slide was the order of the trades? Your real trades re-ordered 2,000 times; same total, different paths. | Plan for the bad-luck (1 in 20) slide, not your own so far. "Lucky order" means deeper slides are likely ahead. |
| Filter-trap meter Performance › Explore | Is the slice you filtered luck? Compares it with random groups of the same size, and tests the best slice in each field (pair, session, …) against shuffled data. | Trap level Low needs 30+ trades and under 5% chance. "Normal for the best of 6 slices" means the leader is what luck alone would give. |
| Slice leaderboard Performance › Leaderboard | Which slices really lead? EV per slice pulled toward the overall average in proportion to how little evidence it has (empirical-Bayes shrinkage). | "No real spread detected" means the differences between slices are noise. |
| Volatility regime Performance › Volatility and news | Results by how big the market's moves were at entry: ATR on the trade's timeframe against the pair's last 180 days in the same session (Very Low … Extreme). | Context, not a rule. Trades are tagged after the Sunday price-history archive covers them. |
| News and results Performance › Volatility and news | Results for trades opened within an hour of high-impact news for either currency, on a news day, or with no high-impact news, from the weekly calendar. News trades are marked on a cumulative R chart. | Needs many trades per group before it means anything. |
| Five-metric report card Home › Five-metric report card | The five core metrics week by week: EV and drawdown against the benchmark's middle 50%, RAER and RAPF, equity acceleration, plus Equilibrium's weekly statuses. | Look for the trend across weeks, not one week's colour. |
How risk is set: the gate and tier ladder
MARS lowers risk automatically as drawdown deepens. Drawdown is measured from the highest equity reached (the Equity Peak High). The gate caps the highest tier allowed; the tier then sets the pool of risk shared by the day's four trades. Across the whole ladder, risk per trade runs from about 1% to 6% of the account.
| Gate | Drawdown from peak | Highest tier | Posture |
|---|---|---|---|
| Growth | 0 – 7% | T7 | Expansion |
| Recovery | 7 – 13% | T6 | Stability |
| Buffer | 13 – 19% | T5 | Defense |
| Floor | 19 – 27% | T3 | Containment |
| Deep-Floor | 27 – 34% | T2 | Suppression |
| Ground-Floor | 34 – 40% | T1 | Recovery crawl |
| System Lock | beyond 40% | none | No trading (risk set to zero) |
Day colours
Each full day gets a colour from its average net R per trade. The colour sets the starting tier for the next day:
| Colour | Avg R per trade | Next-day base |
|---|---|---|
| Top Green | ≥ +0.35R | T5 |
| Green | ≥ +0.25R | T4 |
| Yellow | ≥ +0.05R | T2 |
| Red | below +0.05R | T1 |
Weekly status
Last week's average R per trade nudges this week's tier up or down:
| Status | Avg R per trade | Tier change |
|---|---|---|
| Above plan | ≥ +0.33R | +1 |
| On plan | ≥ +0.21R | 0 |
| Below plan | ≥ +0.06R | 0 |
| Weak positive | above 0 | −1 |
| Negative | 0 or below | −2 |
The tightest limit always wins: the gate cap, then the tier, then a caution lock after repeated red days. That is how the system "keeps the trader on the rail".
Fun, everyday use and trust: records, seasons, games and integrity
Nine lighter tools. Four make the record easier to follow (hall of fame, seasons, a guided tour, presentation mode), two teach what normal luck looks like (a quiz and a game), and three show that the record can be trusted (lessons, the rule-change ledger and the integrity checks).
| Tool (where) | What it answers | How to read it |
|---|---|---|
| Hall of fame Records › Hall of fame | The record board: best-graded trade, biggest and fastest winner, best day and week, longest winning, compliant and green-week streaks, perfect 4–0 days and the highest weekly return. Plus the top five trades and weeks. | Records, not evidence: a single best trade says nothing about the edge. |
| Seasons Records › Hall of fame and seasons | The 78 weeks as six seasons of 13 weeks, each with a league table of its weeks: 3 points for a green week, 1 for a flat week, 0 for a red week. | For fun. At a 60% win rate about 28% of full weeks still end red, so even a good season has red weeks. |
| Trading school Lab › Trading school | 100 general questions in five topics, ten new ones a round (see Process and coaching). Below them, a bonus set of seven questions about normal bad luck for a 60% strategy at 1:1: the longest losing streak, the share of red weeks, 0–4 days, how low the win rate can read after 100 trades, the deepest drawdown in R and the longest run of red weeks. | Guess first, then read the answer. Answers come from the plan and 2,000 simulated runs of the full cycle. |
| Play a week Lab › Play a week | Click through 16 random trades at the measured win rate and cost (the plan before 30 trades) on today's risk rail, then see where your week lands among 2,000 simulated weeks. | A game, not a forecast. It shows how wide a single week's range is. |
| Presentation mode ▶ Present in the menu | A full-screen slideshow of eight slides: the headline, the summary tiles, return vs the benchmark, cumulative R, the five core metrics, the next trading day, the hall of fame and the integrity checks. Arrow keys or the buttons move through it; Auto turns a slide every 9 seconds; Esc closes it. | The Colour-blind switch (in the header next to Light / Dark, and in the slideshow; its two dots show the current good and bad colours) switches every chart to colours that stay distinct for colour-blind readers: blue for gains, orange-red for losses. The choice is remembered on this device. |
| Guided tour Tour in the menu | Sixteen short steps in four chapters: the 2-minute check (section 2: sample and verdict, the header line, EV, drawdown and the gate, return vs the benchmark, cost, the status chips), reading any page (the Today line and the four card buttons, display settings), one stop per menu (Edge, Risk, Performance, Execution, Lab) and getting around (Records, Search, Recent, What's new, Print). Chapter buttons skip ahead; each step highlights the piece it explains and links to this guide. | Offered once on a first visit; restart it any time from Tour. |
| Lesson of the week Calendar, Week stories, Timeline | One sentence a week: the first focus item from that week's weekly review. It appears under the week on the calendar (Performance › Explore), at the end of the week's story and on the timeline. | Written by the reviews, published as written. Practice weeks are not shown. |
| Rule-change ledger Records › Integrity (rule-change ledger) | A dated list of every change to the trading rules, settings, grading or this page, with "rules frozen since" and the days and trades since the last rule change. Rules cannot change before 30 trades. | A long frozen stretch is good: a strategy that keeps changing cannot be measured. |
| Integrity Records › Integrity; chip in the header | Five checks: when the figures were last refreshed; the latest daily Equilibrium health check (PASS / WARN / FAIL); the closed trades in the broker statement against the trades on this page; and a fingerprint (SHA-256) of the trade log. Each update stores the fingerprint of all trades so far and re-checks every earlier one; your browser also recomputes today's. | ✓ Unchanged and Matches mean no published trade has been edited since. A ✗ means an earlier row changed after it was published; the rule-change ledger should say why. |
Quality: empty pages, sample sizes, self-test and phones
Four improvements that work across every page. They change how the page reads, not what it measures.
| Feature (where) | What it does | How to read it |
|---|---|---|
| What unlocks when Home › Summary (until 30 trades) | Before the first trade, empty numbers say From 16 Nov instead of a dash, tables that would be all dashes say what they will show, and a Home card lists the order things appear: trade 1, 2, 5 reviewed, 8, 10 trades, week 1 done, 20, 30, 32, 100 and 300 trades, each with how many trades are still to go. | Empty is on purpose, not broken. Blank values (—) left in tables next to plan or benchmark figures are muted and fill in from the first trade. |
| Sample-size labels Home tiles · Explore · every slice table and bar chart | Every headline number carries its trade count (n 12). One rule everywhere: under 10 trades is faded, 10–29 is lighter, 30 or more is shown normally. A short key sits under each table that has faded rows. | Faded = mostly luck; lighter = provisional. Don't compare a pair, session or weekday until its row is at full strength. |
| Self-test Records › Integrity; header chip | The page re-checks its own numbers every time it opens: the trade count agrees everywhere, cumulative R adds up trade by trade, every win, loss and breakeven label fits the ±0.10R band, the win / loss / BE counts add up, and the EV and win-rate tiles match the trade log (hard checks); every breakdown, the weekly table and the daily rows each cover every trade (soft checks). The update script runs the same checks and refuses to publish a page that fails a hard check. | All green = the figures agree with each other. A red ✗ also turns the header chip to ! Integrity: see checks. It checks consistency, not whether the trading is good. |
| Phone layout Phones (640 px wide or less) | A menu bar at the bottom (Home, Edge, Risk, Performance, More). More opens every other section and tool. Sub-pages sit in one sliding row with Section · 3 of 8; swipe left or right on the page to move between them. Buttons are larger. | Swipes on tables, charts and sliders scroll or use the chart instead of changing page. Desktop is unchanged. |
Everyday use II: simple view, achievements and the weekly update
Three tools for readers who want less, or want to share.
| Tool (where) | What it does | How to read it |
|---|---|---|
| Simple view ◑ Full view / ◐ Simple view in the menu; also on the "New here?" card | One switch that keeps the ten pages a beginner needs: Summary, Five-metric report card, Tomorrow, Explore, Trade log, Week stories, Timeline, Hall of fame and seasons, Achievements and Integrity (plus Glossary, Share card and My dashboard). Edge, Execution and Lab are hidden until you switch back, and Home is shorter: the unlock list becomes one line, and Verdict detail, the status legend and Replay are left out. | Nothing is removed from the record; the other pages are one tap away. Remembered on this device. |
| Achievements Records › Achievements; Home strip | 80 badges in 11 families, process families first (see section 3). Tiles show badges earned, process badges earned, the next badge and gold earned; a filter shows one family at a time. | A motivation board, not evidence. The Evidence family is the only one tied to whether the edge is real. |
| Copy this week's update Share card page · "Copy update" under each week story | A ready-written message about any week: result, wins and losses, win rate, EV and return since the start, the gate, and a link to the record. Short (4 lines) or long (adds two story sentences, the lesson of the week and a provisional / not-advice line). One button copies it. | Written by fixed rules, R and % only, never dollars. Paste it into a text, WhatsApp, a post or an email. |
Decision tools III: loss runway, the day in progress, tier forecast and cost margin
Four tools that start from today's position. They use the page's simplified rail and the plan figures (60%, 0.05R) until 30 trades, then the measured ones.
| Tool (where) | What it answers | How to read it |
|---|---|---|
| Loss runway Risk › Gate ladder and recovery (column "Losses in a row") | How many straight full losses from today it would take to reach each gate, down to System Lock, with the rail cutting risk after every losing day. Next to it: the chance of reaching each gate within 16, 32 and 64 trades. | The deeper gates need far more losses than at today's risk, because the rail shrinks risk on the way down. A few losses to the next gate is normal at full risk. |
| During the day Risk › Tomorrow (under the tree) | For every part-day score (1–0, 0–1, 2–0, 1–1, 0–2 …), the chance the day still finishes in profit and which next-day tiers are still possible. | A 4-trade day needs 3 wins to finish in profit at 1:1 with costs. The day's result moves tomorrow's tier; it never changes the plan for the rest of today. |
| Tier forecast Risk › Tier forecast | Where the next-day tier (T1–T5) and the gate are likely to be after 4, 8 and 13 more weeks, from 2,000 simulated paths; the most likely tier, the chance of T4 or higher, and the median risk per trade. | The tier follows the day colour, so T1 after any red day is normal. Watch the gate table: most paths should stay in Growth or Recovery. |
| Cost safety margin Performance › Cost (top); short version on Edge › Edge statistics | The break-even win rate at today's cost, the cushion in points above it, the cost per trade that would erase the edge, and the break-even win rate for growth at today's risk. A line shows EV falling as cost rises. | At 1:1, every 0.02R of extra cost needs one more point of win rate. A cushion of 5+ points and an edge-erasing cost several times today's are healthy. |
Everyday tools III: reading, checking and saving any page
Tools that work the same way on every page. Most of them are buttons at the top right of each card or in the menu.
| Tool (where) | What it does | Tip |
|---|---|---|
| Today line Top of every page | One plain sentence that says what the page's numbers mean right now, and how much to trust them at this sample size. | Read it first; the cards below give the detail. |
| ⓘ How it's calculated Every card | Opens a short note under the card title: the formula in words, the inputs (how many trades, when the page was built), the sample-size warning and a link to the right section of this guide. | Use it before quoting a number to someone else. |
| ▦ View as table Every card with a chart | Shows the numbers behind each chart as a table, with a Copy table button that pastes cleanly into a spreadsheet. | Handy for checking one exact value or doing your own maths. |
| Range buttons and zoom Charts with weeks, trades or days along the bottom | 4 wk / 13 wk / All (or Last 50 / Last 200, or 20 / 65 days) set the range of every such chart on the page at once. With a mouse, drag across a chart to zoom into that stretch; Reset zoom goes back. | Short ranges are noisy: use them to look closely, not to judge the edge. |
| Exact values Every chart | Hover with a mouse, tap on a phone (the readout stays until you tap elsewhere), or focus a chart with Tab and use the left and right arrow keys. | Screen readers also get a one-sentence summary of each chart. |
| Text size and display Aa, Colour-blind and Light / Dark in the header | Aa steps the text through 100%, 112% and 125%. The page follows the device's reduce-motion setting, and a Skip to the page content button appears for keyboard users. | All choices are remembered on this device only. |
| ⎙ Print Menu | Prints the current page, or saves it as a PDF, in light colours with every chart drawn and the buttons left out. | Open the page you want first; each page prints on its own. |
| New-data dots and ↺ Recent Menu | A small gold dot marks menu sections and pages whose numbers changed since you last opened them. ↺ Recent lists the last eight pages you viewed. | Pages you have never opened show no dot. |
| ✦ What's new Menu | The page's versions in plain words, newest first; a dot shows when there is a version you haven't seen. Trading-rule changes stay in Records › Integrity. | If one card ever fails with new data, it shows a short notice and the rest of the page keeps working. |
Process and coaching: mistakes, the checklist and focus follow-through
Three cards built from the per-trade reviews (section 8) and the weekly reviews. They fill in from the first reviewed trade (16 November) and the first weekly review (after week 1).
| Tool (where) | What it answers | How to read it |
|---|---|---|
| Mistake library Execution › Breach cost (bottom) | Which of the 14 mandatory items get missed, ranked by how often: trades, share of trades, the EV of those trades against fully compliant trades, a rough R cost and when it was last seen. Items never missed are listed too. | The top row is the habit to work on. The R cost is a rough price of the habit, not proof that the item caused the loss; under 10 trades a row is mostly luck. |
| Checklist heatmap Execution › Grades and compliance | The 14 items by cycle week (last 12 weeks): red = breached on at least one trade, green = passed wherever it was visible, grey = not visible on the screenshots, with breaches out of trades reviewed. A last column gives each item's "not visible" rate. | A red row that keeps coming back is a recurring mistake. A high "not visible" rate is a screenshot habit to fix: an item that can't be seen can't be checked. |
| Focus follow-through Execution › Grades and compliance | Every focus item from each weekly review, checked against the following week. A focus that names a checklist item (by number or its words, for example "trigger panel" = item 4) is measured by that item's breach rate; anything else by the week's compliance rate and average quality grade. | Improved or Held = the focus worked. One week is a small sample: look for the same focus coming back as Worse more than once. |
| Trading school Lab › Trading school | 100 multiple-choice questions in five topics (variance and probability, risk and position sizing, expectancy and statistics, trading psychology, costs and market sense). Each round deals 10 questions not seen yet; every answer shows a short explanation. Pick one topic or mix them all. Below it, a bonus set of seven questions on this plan's own variance. | General education, not advice. Progress is kept on the visitor's device only; no question repeats until all 100 have been seen. |
Statistical depth III: shifts, streaks, fills, dips and honest forecasts
Seven tools that test whether what you see is real. Most need many trades; under 30 they carry a Provisional note, and none of them changes a rule.
| Tool (where) | What it answers | How to read it |
|---|---|---|
| Change-point detector Edge › Stability | Has the EV, the win rate or the cost per trade shifted? Pick a measure; a CUSUM line marks the most likely change, with the value before and after and the chance a shift that strong is luck. A table and one plain line sum up all three. | "No clear shift" on all three is the normal state. A likely shift (under 5% luck) needs a reason, such as a market regime or a broker cost change, before it means anything. |
| Streak odds Edge › Stability (below the independence test) | How normal is a losing run? For runs of 3 to 10 losses: the chance of at least one in the next 100 trades and over the full 78 weeks, at the measured win rate (the plan's 60% until 30 wins and losses). The current and longest runs get the same reading. | Expected or Normal = arithmetic, not a broken edge: at 60%, a 5-loss run happens in about half of all 100-trade stretches. Rare = check the grades and the tilt detector, not the rules. |
| Time in trade Performance › Time and rhythm (under A14) | Results by holding time (under 15 min, 15–60 min, 1–4 hours, 4 hours +): trades, win rate, EV and net R per bucket, the median hold of winners and losers, and fast (under 1 hour) against slow with the chance the gap is luck. | At a 1:1 hard stop, holding time reflects how fast the market moved, not a choice. A bucket with under 30 trades means little. |
| Fill quality Execution › Fill quality | Before cost, a clean target fill reads +1R and a clean stop −1R. How close winners and losers land, how many winners closed early (below +0.9R) and how many losers slipped past −1.1R, any drift between the first and second half, and the same by exit type. | 80% or more of fills within ±0.05R of the bracket is clean. Early winners point to manual exits; deep losers to slippage or gaps (a broker or news question). |
| Drawdown anatomy Risk › Pain and tail risk (bottom) | Every dip of 2R or more: start, trough, depth in R and in account %, trades down, trades to recover, win rate in the dip, and its main cause. The cause split shows how much more the dip lost than an average stretch: win rate (losses bunched), cost (a costlier patch) or sizing (risk above average, which deepens the account dip without changing R). | Most dips at 1:1 are win-rate dips: losses bunching by chance. A dip caused by cost or sizing is worth a closer look. |
| Benchmark rank Performance › Benchmark rank | Each finished week: the share of simulated 1:1 runs the account was beating, with zones (bottom 10%, median, top 10%), the average rank, weeks above the median, the best and worst week and the trend over the last 8 weeks. Same estimate as A27. | The fan is the ruler, not a target. Hovering around 50% is on plan; many weeks in the bottom 10% means the record is falling behind what the plan's odds allow. |
| Frozen forecasts Lab › Next-week forecast (bottom) | Before each week starts, the page builder saves four forecasts and never changes them: the week's net R (80% range), the chance of a losing week, the worst week of the next four and the account change over the next four. Each is scored once its weeks are over. | About 8 ranges in 10 should land inside. The losing-week calls get a Brier score: under 0.25 beats a coin flip. Because nothing is rebuilt, hindsight cannot flatter these. |
More statistical depth: proof, discipline and hidden risk
Nine more tools. Three test the edge formally, three measure discipline from the trade reviews, and three look for risk and cost that the totals hide. Most need many trades; under 30 they carry a Provisional note.
| Tool (where) | What it answers | How to read it |
|---|---|---|
| Sequential test Edge › Sequential test | Cumulative R between two straight lines (Wald's sequential test). Crossing the top line accepts "the edge is about +0.15R a trade" (or +0.10R / +0.20R); crossing the bottom line accepts "no edge". Both mistakes are held to 5%. | A reading aid for the judgment-based stop rule, not a rule. Wandering between the lines for a long time means the edge, if any, is small. |
| Payoff map Edge › Payoff map | Win rate across, payoff ratio (average win ÷ average loss) up, with lines of equal edge and the break-even line. Your dot sits with a cloud of 400 re-draws of your trades, beside the plan. | The share of the cloud above break-even is the chance the edge is real. At 1:1 the payoff stays near 0.90, so the win rate does all the work. |
| Booster value Execution › Booster value | For each optional booster (quick 55 cross, scorecard 5+, zone pullback, strong HTF trend, stop beyond structure, triangle, room 2R+, golden candle / divergence, price beyond the 987): win rate and EV with and without it, and results by number of boosters. | Boosters are tagged from the trade reviews' notes. Gaps on a few trades are noise. |
| Breach cost meter Execution › Breach cost | Discipline in R: EV of compliant trades against trades with a plan breach, the cost per breach with a 90% range, and the total R lost to breaches. | A positive cost per breach that stays positive as trades add up is the strongest argument for the checklist. |
| Tilt detector Execution › Tilt detector | The trade after a win against the trade after a loss (and after two losses the same day): win rate, EV, minutes before the next entry and the next trade's grade. Plus results and grades by slot T1–T4. | Watch for a lower win rate or grade after losses, or a much shorter wait (rushing back in). |
| Sample-size planner Lab › Sample-size planner | How many trades it takes to detect a gap you care about between two groups, or to prove the edge above zero; and, for any two groups already traded, the smallest gap the current trades can detect. | Halving the gap quadruples the trades needed. Most slice differences will stay hints for the whole 78 weeks. |
| Currency exposure Risk › Currency exposure | How often trades overlapped, the most open at once, the biggest same-currency stack, results alone vs overlapping vs stacked, and long / short counts and EV by currency. | Stacked trades double the risk on one currency. Equilibrium's EXPOSURE sheet warns about them on the day. |
| Cost heatmap Performance › Cost | Average all-in cost in R by pair and New York time of entry (3-hour blocks), with the cheapest and dearest cells. | Rollover hours (17–19 NY) are usually dearest. Small stops also raise cost in R. |
| Day colours Risk › Day colours | How often each day colour follows each other, against the all-days mix, and the long-run average base tier this implies. | At 1:1 a full day is either Top Green (4–0, 3–1) or Red (2–2 or worse). If rows differ a lot from the all-days row, days are not independent. |
More decision tools: tomorrow, stress, pain and value
Seven more tools look at risk from the trader's chair: what tomorrow can do, how bad things could get, and whether the rail itself is paying off. All start from today's real position and use the same simplified rail as the decision tools above.
| Tool (where) | What it answers | How to read it |
|---|---|---|
| Tomorrow Risk › Tomorrow | Tomorrow's gate, tier and risk per trade, and a tree of the five possible days (4–0, 3–1, 2–2, 1–3, 0–4): the chance of each, the account change, and the gate, day colour, tier and risk it hands to the day after. Also how many R (and straight losses or wins) separate today from the next gate down, the gate above, or a new equity high. | A 2–2 day is already Red (average below +0.05R a trade), so the next day drops to T1. That is the rail working, not a failure. |
| Stress tests Lab › Stress tests | Bad scenarios run from today: 10 losses in a row, a 50% win rate for 8 weeks, a 45% month, cost doubled for a quarter, with normal conditions alongside. Shows the result, the bad case (1 in 10), the deepest drawdown and gate, and the System Lock chance. | Look at the gate column: the rail is built so most bad scenarios end in the lower gates, not the lock. |
| Pain panel Risk › Pain and tail risk | Drawdown the way an investor feels it: share of days below the high, ulcer index (typical depth, weighted toward deep dips), average drawdown, how long spells under water last, and an underwater chart with the first gates marked. | Each tile shows the benchmark's middle half for the same number of days. Inside or below it is normal for this plan. |
| Tail risk Risk › Pain and tail risk | The worst day and week, the average of the worst 5% (expected shortfall), and the worst week to expect in the next 13, for the record and for the benchmark. | Plan for the "1 in 10" worst week, not the typical one. |
| ETA board Lab › ETA board | When the 30, 100 and 300-trade milestones (and their milestone reviews) should arrive at the full schedule and at the current pace, and the chance and likely week of reaching ×2, ×5, ×10 and the end of the plan-pace line by week 78. | Growth dates swing by months with a few points of win rate. Treat them as ranges. |
| Rail value-added Risk › Rail value-added | The real record beside the same trades at one flat risk (by default the same average risk, the fair test): return, max drawdown, return ÷ drawdown, ulcer index. A luck check replays the trades in 1,000 random orders. A 78-week simulation shows what to expect from the rail against flat risk. | The rail's main job is survival: expect it to cut the System Lock chance sharply, sometimes at the cost of some return. |
| Growth attribution Performance › Growth attribution | Splits growth into gross edge, cost, sizing (risking more on winners or losers) and compounding, in growth points that add up exactly. A second chart explains the gap to the plan-pace line: trades, edge, cost, sizing and compounding. | Edge and cost tell you about the strategy; trades and sizing about execution and the rail. |
The everyday tools
Eight tools make the page quicker to use day to day. None of them changes a number; they help you find, keep, compare and share what is already there.
| Tool (where) | What it does | Good to know |
|---|---|---|
| Search ⌕ Search in the menu, or press / | Jumps to any page, chart or card, glossary term, week, day, trade or timeline event. Type a few letters, use ↑ ↓ and press Enter; the page opens and the card you picked is highlighted. | Ctrl+K also opens it. Try "drawdown", "week 3", "EURUSD buy" or "gate". |
| My dashboard ★ My dashboard in the menu | Your own page. Tap the ☆ at the top right of any card to pin it; the dashboard shows your pinned cards in the order you choose (↑ ↓), each with a link back to its page. | Pins are kept in your browser on your device only. Good first pins: Five core metrics, the edge-proof meter, the gate ladder. |
| Saved views Performance › Explore | Name the current Explore filters (for example "London M5 buys") and recall them later from the Saved views list. | A saved value not yet in the record (a pair not traded yet) is left as All, and the page says so. |
| Compare periods Performance › Compare periods | Any two weeks, months or date ranges side by side. Each tile shows B large, A underneath and the change (green = better, red = worse); a chart lays both periods' cumulative R over each other, and a table splits them by session, pair or any other field. | One week against another is mostly noise. Under 30 trades in a period the page marks it Provisional. |
| Since your last visit Home › Summary (top strip) | What changed since you last opened the page: new trades and their net R, EV, win rate, return and drawdown then → now, gate and verdict changes, new badges. | Measured against your last visit, not against the plan. Kept in your browser only. |
| Week stories Records › Week stories | Each week in 3–4 plain sentences written by fixed rules: result and win rate, best and hardest day, the session that carried the week, then a record, a streak, heavy costs or what the risk rail did. | Rule-based description, not opinion and not a verdict. |
| Share card and chart images Share card in the menu · ⤓ on chart cards | The share card puts one week on one image (result, win rate, EV and return to date, two sentences of its story, every trade as a bar). The ⤓ button on any card with a chart saves that chart, its title and legend as a PNG. | R and % only, never dollars. Saving works when the page is open in the Claude viewer, which asks you first; elsewhere take a screenshot. |
| Timeline Records › Timeline | Go-live, trade milestones (1, 30, 100 and 300 trades, when a milestone review is due), every gate change, badges and the trader's own short notes, newest first. The same notes, milestones and gate changes appear as ▼ markers on the return and cumulative R charts; hover a marker's column to read it. | Notes are the trader's words, published as written. Everything else comes from the record. |
5 · The 20 charts, explained
Charts 1–20 sit together under Records › Charts 1–20, in five groups (several also appear on their topic pages). For each one: what it analyses, what a good reading looks like, and what a warning looks like. Results are always net of costs.
Growth and edge over time
The account's % return week by week, drawn over the Monte Carlo fan for a 1:1 strategy at plan. It's zoomed in on the weeks so far.
Every trade's net R added up in order. This is the purest view of the edge, because it ignores position size and compounding.
Average net R of the most recent 20 trades, recalculated after each trade. It shows whether the edge is holding up recently, not just over the whole test. The plan line is about +0.15R.
Cumulative R of the real trades beside three "ghost" versions of the same entries: what would have happened with fixed 1:1, 1:2 and 1:3 targets. The ghosts come from a price logger that keeps watching each trade after it closes.
Risk and survival
How far below its highest point the account was at the end of each day. Zero means a new high. Dashed lines mark the gates where risk steps down (7, 13, 19, 27, 34, 40%).
The % of the account risked on each trade, day by day, as set by the gate, the tier and the last day's colour. It shows the risk rail working.
Consistency
One bar per week: the total net R made (green) or lost (red) that week.
Each week's win rate against the 60% plan assumption. Breakevens (within ±0.10R) don't count as wins.
For full 4-trade days, the share of days with 0, 1, 2, 3 and 4 wins, against the binomial theory above. It tests whether days behave the way a 60% coin would.
How many trading days ended Top Green, Green, Yellow or Red under the MARS day-colour rule (section 4). The colours drive the next day's risk.
Trade quality and exits
A histogram: how many trades closed at each net R value. With a 1:1 hard stop and target, results should cluster in two heaps near +1R and −1R.
The EV per trade for the actual exit beside each ghost target (1:1, 1:2, 1:3). Chart 4 shows the running total; this one shows the bottom line per trade.
One dot per trade. Up = the best move in the trade's favour (MFE). Right = the worst move against it (MAE), both in R. Dots past the green line reached +1R; dots past the red line hit the stop.
Where the edge comes from
Charts 14 to 19 slice the same trades in different ways. Each bar shows the slice's EV, and hovering on the live page shows trades, win rate and net R. Rule of thumb: ignore any slice with fewer than about 30 trades.
Results split by trading session (Asian, London, London–New York overlap, New York and Rollover), all in New York time.
The share of trades taken in each session: where the trader actually spends time, as opposed to how well each session pays.
Results by day of the week (four trading days are planned each week).
Results by the chart timeframe the trade was taken on (e.g. M5, M15, H1).
Buys vs sells, and trades closed at the take-profit, at the stop, or by hand.
Results for the 12 most-traded instruments (currency pairs, gold, indices and so on).
Cost
The average cost per trade in R, split into fees, spread and slippage, beside the gross edge (EV before costs). It shows how much of the edge the market and broker take.
6 · The Analytics lab: 39 more charts
The Analytics lab adds 39 numbered charts (A1–A39) in eight groups. Each group has its own page in the menu (for example Edge › Proof charts, or Risk › Equity and drawdown); ⌕ Search finds any chart by its number. Every chart uses all counted trades; the Explore filters do not apply here. Hover over or tap a chart for exact numbers. Four of them are the headline tools: the edge-proof meter (A1, also at the top of Edge › Edge statistics), the rolling charts (A5–A10), the execution grades (A34–A37) and the stop and target what-if (A38–A39, also at the top of Performance › Excursions and what-if).
Edge proof
Is the edge real, or could it be luck?
A dial showing the chance that the true EV is above zero, worked out from how spread out the trade results are (a t-test). Beside it: EV, its 95% range, the t statistic, how many trades the current numbers need for proof, and progress toward that. The same meter opens Edge › Edge statistics.
Running EV from trade 10 on, inside its 95% range. The range is wide at first and narrows as trades add up.
The real trades re-drawn at random 2,000 times; each re-draw gives one possible EV. The bars show how far luck alone could move the EV.
The orange line is how many trades the current EV and spread need for 95% proof; the dashed line is trades done. Where dashed passes orange, the edge is proven if it holds.
Rolling windows
Pick a window of 10, 20 or 50 trades; every chart in this group uses it.
Average net R of the latest window of trades, with its uncertainty band.
Win rate of the latest window (breakevens left out), with a 95% range. Dashed lines: plan 60% and the break-even win rate at the current cost.
R won ÷ R lost in each window, capped at 5. Above 1 makes money; 1.5 is the plan before costs.
How spread out results are. At 1:1 with a 60% win rate the spread is about 0.98R.
Every run of results in order: bars up = wins in a row, down = losses in a row, small gray = breakevens.
How many losing streaks of each length happened, next to what pure chance predicts for this many trades at the measured win rate (the plan's 60% until 30 wins and losses). The odds for each length are on Edge › Stability › Streak odds.
Time and rhythm
When does the edge show up?
A 24-hour dial in New York time. Each wedge is an hour: longer = more trades, green made R, red lost R, darker = stronger EV.
A grid of average net R per trade for every weekday and session pair, with trade counts.
Total R by trade slot (T1 = first trade of the day) and weekday.
Every trade as a dot: minutes held (log scale) against net R, coloured by session. The Time in trade card below it groups the same trades into holding-time buckets.
Each calendar month as a step up (green) or down (red); the last bar is the running total.
Average cumulative R through the four slots of a day, one line per weekday.
Pairs and mix
Which markets carry the edge, and in which direction?
Each bubble is a pair: right = more trades, up = higher EV, bigger = larger total R.
EV of long and short trades on each of the 10 most-traded pairs.
Win rate for each pairing of trigger timeframe and session. Colour shows distance from the 60% plan.
Running net R for the five most-traded pairs.
A radar with six measures against plan: win rate, EV, profit factor, payoff, green weeks and cost efficiency. The bold ring is the plan; outside it is better.
Risk and equity
How the risk rail behaved and how deep the dips went.
Growth (line) and drawdown from peak (shaded) on one chart. Growth is rebuilt from each day's net R × risk, so it can differ slightly from the account.
Bars = each day's net R (left axis); line = risk per trade that day (right axis).
One coloured strip per trading day showing the drawdown gate, above the drawdown curve.
Each dip in cumulative R as a bubble: right = how many trades it lasted, down = how deep. Teal recovered; red is still open. Each dip of 2R or more is listed with its cause in Drawdown anatomy (Risk › Pain and tail risk).
Bars = days spent at each risk tier; line = average day net R at that tier.
Each week: the share of simulated 1:1 runs the account is beating, estimated from the fan. 50% is the median. Performance › Benchmark rank shows the same line with zones, a trend and a week-by-week table.
Thin lines = the same trades replayed in random order; the shaded band holds the middle 90% of 300 replays; the thick line is what happened.
Trade quality and cost
What a typical trade looks like and where the EV comes from.
Each dot is one trade, spread sideways so they don't overlap; the black bar is the session's average R.
20-trade averages of the edge before cost (gross), the cost per trade and what is left (net EV).
For winners: how far price kept going in your favour (MFE) beyond the R actually banked.
How far trades went against you (MAE) before closing, as a share of winners and of losers.
A waterfall per trade: what winners add, what losers take, breakevens and cost, ending at net EV.
Execution grades
From the per-trade reviews against the Sniper Pro method. Quality is graded A+ to D-; compliance is a separate pass / breach line and is never averaged in.
Each trade's grade with a 10-trade average line; B is a solid professional standard.
Bars = share of trades that week with full compliance; line = the week's average grade.
A donut of how grades spread across the A, B, C and D families.
Each trade's grade against its net R; red dots are compliance breaches.
Stop and target what-if
Research only: MARS keeps the 1:1 hard stop.
Two sliders replay the logged trades with a different stop (0.5× to 2× the real one) and target (0.5R to 3R). Results are re-scaled to R of the new stop. At 1.00× and 1.00R it matches the real results. The same tool opens Performance › Excursions and what-if.
EV for every stop (rows) and target (columns) combination, replayed on all logged trades.
7 · The advanced analytics, in plain words
These figures appear on the Edge statistics, Cost and Excursions pages and in the verdict detail on Home. You don't need them all. Each one answers a specific question.
Is the edge real? (Edge › Edge statistics)
The likely range of the true win rate given the trades so far. With few trades the range is wide; it narrows as trades add up.
The win rate needed just to cover costs at 1:1. It's about (1 + cost in R) ÷ 2.
The likely range of the true average R per trade. If even the low end is above zero, the edge is unlikely to be luck. If even the high end is below zero, the strategy is very likely losing.
How spread out the results are. At 1:1 most results are near +1R or −1R, so SD sits close to 1. A higher SD means a bumpier ride.
How many "noise units" the EV is above zero. Around 2 or more means the result is unlikely to be chance alone.
How many trades it takes before an EV this size would be statistically convincing.
Edge per unit of bumpiness. It compares strategies with different volatility on a level field.
EV ÷ SD × √(number of trades, capped at 100). A popular single-number quality score: roughly 1.6–2 is fair, 2–3 good, above 3 excellent.
The typical size of a win and of a loss in R. At 1:1 both should be close to 1R. If losers are clearly bigger, look at stops and slippage.
The longest streaks so far. At 60%, a run of 5–6 losses will very likely happen in a long test. It's a normal part of the distribution.
The recent edge. The same idea as chart 3, shown as one number.
Exits at TP / SL / by hand, true 1:1 plans, over- and under-risk trades, days with more than 4 trades and skipped slots. These measure discipline, not luck.
Is risk being used well? (core metrics)
Total profit ÷ total amount put at risk. It answers: for every unit of risk deployed, how much came back? Because MARS varies its risk, RAER shows whether bigger bets were placed at the right times.
The R profit factor (R won ÷ R lost), adjusted for the average risk level used. It answers whether profit quality is real or just a product of risk size.
This week's % return minus last week's. Positive means compounding sped up. It's naturally jumpy, so it's judged against the benchmark's normal range, not on its own.
The deepest fall from a peak, in %. It's the survival metric: an edge is only useful if the account lives long enough to collect it.
What does trading cost? (Performance › Cost)
Fees = broker commission. Spread = the gap between buy and sell prices. Slippage = getting filled at a slightly worse price than the stop or target. All are shown per trade, in R.
All-in cost adds the three together. Gross edge is the EV before costs. Share of edge used by cost = cost ÷ gross edge. Lower is better.
How did trades move? (Performance › Excursions)
The furthest a trade moved in profit before closing, in R.
The furthest a trade moved against it before closing, in R.
Winners that came within 0.2R of the stop before winning. A high share means entries are often "just survived".
Losing trades that were nearly a full win first. A high share suggests a target or trade-management question.
The same entries replayed with fixed 1:1, 1:2 and 1:3 targets, watched by a price logger. They show what a different exit would have earned, without changing the live plan.
The share of trades that have logger data. Excursion and ghost figures only describe the covered trades.
8 · The trading plan behind the record
Every number on the track record comes from one written trading plan. This section explains it in plain words: what the trader is allowed to do, how a trade is found, how much is risked, and how every trade is graded afterwards. Knowing the plan tells you what the numbers should look like if it is followed.
The plan on one page
| Rule | What it means |
|---|---|
| One strategy, 1:1 | Every trade has a hard stop loss and a take-profit at the same distance from entry. No moving the stop, no letting winners run past the target. A clean win ≈ +1R, a clean loss ≈ −1R. |
| Four trades a day | Up to four trades a day (slots T1–T4), four days a week: 16 trades a week. All four trades of a day carry the same risk. |
| Risk comes from the rail, not from feel | The next day's risk is set by the gate and tier ladder (section 4) after each day closes. The trader does not choose a bigger or smaller size by mood. |
| Markets | Forex pairs, on a 5-minute or 15-minute trigger chart, checked against higher timeframes up to the 4-hour or daily chart. |
| Length | 78 weeks from Monday 16 November 2026, or until the trader judges the test is no longer worth continuing. |
| No rule changes before 30 trades | The rules are frozen for the first 30 trades so early luck, good or bad, can't be used to "tune" the plan. |
| Plan assumption | A 60% win rate and about +0.15R per trade after costs. Both are assumptions until measured on live trades; the track record is how they get measured. |
How a trade is found: the entry routine
Execution is discretionary (a person decides), but every entry must pass the same sequence. In short: wait for price to stretch to an extreme, confirm the bigger trend agrees, and enter only when price turns back in the trend's direction.
- The alert. An indicator flags that price stretched beyond its normal range (outside the outer Bollinger band, with RSI overbought or oversold) and has started to come back inside. The alert only calls the trader to the chart. It is never an entry by itself.
- Top-down trend check. The trader looks up the timeframes. A 5-minute entry needs the 5-minute to 1-hour trends lined up with the trade; a 15-minute entry needs 15-minute to 1-hour. Trading with the larger trend, buying pullbacks in uptrends and selling rallies in downtrends, is the core idea.
- Price reaches the extreme band. Price must at least touch the inner line of the extreme band on the trading software: the lower band for buys, the upper band for sells.
- Trend panel agrees. The entry timeframe must show the trend in the trade's direction and in a pullback; the higher timeframe must point the same way with at least "Strong" strength.
- The trigger: a close across the 55 moving average. After the band touch, a candle must close on the trade side of the 55-period moving average. This is the main "go" signal. The sooner it follows the band touch, the better the trade.
- Confirmations line up. Momentum, candle colour and direction arrows, a volume reading, the 233 moving average, structure and trend-matrix readings must all point the trade's way (the full list is below).
- Place the trade at 1:1. Stop just beyond the recent swing (below the swing low for a buy, above the swing high for a sell), target the same distance away. The trader checks that the stop is big enough that costs don't eat a large share of 1R.
The 14-point checklist
Fourteen items are mandatory. If any one of them is visibly missing when the trade was taken, the trade is flagged as a plan breach, win or lose.
1 · Extreme-reversal alert printed. 2 · Multi-timeframe trend table aligned, with the average in the trade's direction. 3 · Price touched the inner extreme band. 4 · Trend panel: entry timeframe pulling back in the trade's direction, higher timeframe the same way and at least Strong.
5 · A candle closed across the 55 moving average in the trade's direction (closing on the line doesn't count).
6 · Oscillator line in the trade's colour. 7 · Oscillator trend bar in the trade's colour. 8 · Candle arrows in the trade's direction. 9 · Candle fill in the trade's direction. 10 · A volume reading printed on or after the band touch.
11 · Price on the trade side of the 233 moving average. 12 · Candle colour on the entry template matches the trade. 13 · Structure histogram in the trade's colour. 14 · Trend matrix: direction and momentum both on the trade side of zero, with momentum sloping the trade's way.
Many optional "boosters" raise the quality of a setup without being required: a quick 55 cross after the band touch, a high indicator scorecard, a pullback into the 38.2–61.8% zone of the last swing, room of 2R or more to the next level, a key price zone, and higher timeframes (4-hour, daily) also aligned. A few caution signs, such as a failed higher-trend signal printing near the entry, lower the grade but are never a breach.
The trader's own discretion notes
From history tests, not hard rules: be sceptical of New York-session entries, avoid trend-continuation entries around the daily rollover (17:00–19:00 New York time), avoid two open trades in the same currency at once, and avoid very small stops, because at 1:1 the cost of a tiny stop can take a large share of the win.
How every trade is graded
After each day every trade gets a written review from its before and after chart screenshots. Each trade ends with two separate results that are never blended:
How good the setup and execution were: each chart graded on what it shows, plus a cost check (was the stop big enough for its cost at 1:1?). The grades average into one letter.
Did the trade meet all 14 mandatory items? A breach names the missing item. It is a flag, not a score, so it is never averaged away.
Execution is judged against the plan, not the outcome: a compliant A-grade trade that loses is good trading, and a breached trade that wins is still a breach. Daily reviews roll up into weekly, period, quarterly, annual and full-run reviews, so the paper trail runs from the first trade to week 78.
The daily routine
| When | What happens |
|---|---|
| Sunday | Pull the week's economic news calendar and update volatility baselines for every pair. |
| Before trading | Read the day's risk per trade from the rail. Check the news days. |
| During the day | Take up to four trades that pass the checklist, saving before and after screenshots of each. |
| End of day | Save the broker statement, refresh the journal, run a health check on the numbers, sort the screenshots, review and grade every trade, and set the next day's risk. The track record updates from the same journal. |
- Results should cluster near +1R and −1R. Anything else (large wins, deep losses) means a stop or target was changed, and shows in the plan checks.
- Risk changes by rule only. The daily record under Risk › Gates and tiers should match the ladder every day.
- Losing streaks are expected. A 60% plan still loses 40% of the time, and losses come in runs.
- Watch compliance as closely as profit. A record is only a fair test of the plan if the plan was actually followed.
9 · Planning, what-if and fun tools
Pages 7 and 8 of the build plan (11 November 2026) added nineteen tools. They answer four kinds of question: is this week and the whole cycle on track?, what if the rules were different?, how did a week or a season actually unfold? and, for fun, can you beat chance at guessing? None of them changes how MARS trades: the risk rail, the 1:1 stop and the four trades a day stay fixed for the 78-week test, and no rule changes are made before 30 trades.
| Question | Tool (where) | Starts working |
|---|---|---|
| Is this week normal? | Is this week normal? (Home › Pace and finish line) | First trade |
| Is the 110-fold goal on track? | Required pace (Home › Pace and finish line) | Week 1 |
| Where will the cycle end? | Finish-line projection (Home › Pace and finish line) | Now (plan figures); 10 trades for ranges |
| One-line mood of the record | Record weather (Home › Summary) | First trade |
| Warn me when something slips | Watch levels (Home › Watch levels) | Now |
| Is the edge fading? | Edge decay check (Edge › Edge decay) | 40 trades |
| What risk does the edge justify? | Kelly check (Risk › Kelly check) | Now (plan); 30 trades measured |
| Was it worth the effort? | Yardsticks (Performance › Yardsticks) | Week 1 |
| What if a slice were skipped? | Skip-it what-if (Lab) | First trade; meaningful at 100+ |
| Is the 4th trade worth it? | Fourth-trade value (Lab) | First day with 4 trades |
| More or less risk? | Mode comparison (Lab) | First trade |
| Did a rule change help? | Rule-change impact (Records) | 30 trades each side |
| Show me one trade in context | Trade card (Records › Trade log, click a row) | First trade |
| Relive a week | Replay a week (Records) | First trade |
| Print or file one week | Week report (Records) | First trade |
| Look back on a season | Season recap (Records › Hall of fame and seasons) | First trade of the season |
| Guess next week | Predict the week (Lab) | Now |
| A game for the week | Week bingo (Lab) | First trade |
| How predictable is one trade? | Guess the trade (Lab) | 5 trades |
Two ideas from the plan were dropped because the page already had them: a recovery planner (Risk › Gate ladder and recovery has the recovery clock) and an edge clock (Analytics A11, the trading clock). A third, compare two periods, is Performance › Compare periods.
Is the cycle on track?
Is this week normal?
Home › Pace and finish line
This week's net R so far, set against the range that 8 weeks in 10 land in. It also shows the normal range for a full 16-trade week and where the week is likely to finish from here.
How it is worked out. Each trade either wins 1R minus cost or loses 1R plus cost. With a win rate p and k trades, the number of wins follows a binomial distribution, so the 10th, 50th and 90th percentiles of the week's R can be worked out exactly; no simulation is needed. The win rate and cost are the plan's (60%, 0.05R) until 30 trades are counted, then the measured ones.
How to read it. Compare the first two tiles. 'Inside the normal range' is the usual answer. 'Likely finish' adds the range of the trades still to come to what is already banked.
Good sign: most weeks inside the range; red weeks about 1 in 4, which is normal at 60%. Warning sign: several weeks in a row below the range. One week outside it is luck; a run of them is worth a look at Edge › Edge decay.
Example. After 8 trades the week is at −1.3R. At 60% the normal range for 8 trades is about −2.4R to +5.6R, so this is inside the range: a soft start, not a problem. With 8 trades left, the likely finish is roughly −3.7R to +4.3R.
Required pace
Home › Pace and finish line
The weekly growth the account needs from today to grow the account about 110-fold (the goal) by the end of week 78, next to the average weekly growth so far and the benchmark's median pace.
How it is worked out. Needed = (goal ÷ today's account multiple) ^ (1 ÷ weeks left) − 1. Achieved = (today's multiple) ^ (1 ÷ weeks played) − 1. The plan pace comes from the benchmark's median path.
How to read it. If 'achieved' is at or above 'needed', the goal is on schedule. The gap matters more late in the cycle, because fewer weeks are left to make it up.
Good sign: achieved close to or above the plan pace. Warning sign: needed climbing far above the plan pace. That says the goal is slipping; it is not a reason to raise risk, because the rail sets risk. Lab › Goal-seek shows what win rate or risk the goal would take.
Example. After 20 weeks the account is ×4. The goal needs ×110 ÷ 4 = ×27.5 in the remaining 58 weeks, about +5.9% a week. If the pace so far is +7.2% a week, the goal is ahead of schedule.
Finish-line projection
Home › Pace and finish line
Where the core metrics are likely to stand at the end of week 78: EV per trade, the maximum drawdown still to come, the return, the profit factor, and the chance of reaching the 110-fold goal.
How it is worked out. The return, drawdown and goal chance come from 1,000 runs of the MARS rail from today's account and drawdown to the end of week 78, using the plan or measured win rate and cost. The EV and profit-factor ranges come from resampling the trades so far 1,000 times (a bootstrap), from 10 trades.
How to read it. Each row gives today's value, a middle-80% range at week 78, and a good sign to compare with. A wide range is honest: it means the outcome is still very open.
Good sign: the EV range above zero; a median return near or above the plan; a typical drawdown well short of the Floor gate (19%). Warning sign: an EV range that straddles zero after 100+ trades, or a 'bad case' drawdown near 34% or deeper.
Example. With 120 trades at +0.12R a trade, the EV range might be +0.03R to +0.21R: the edge is likely real but its size is uncertain. The goal chance might read 46%: close to a coin flip, which is what an ambitious goal looks like.
Record weather
Home › Summary
A light-hearted one-line summary, from Sunny to Stormy, with the chances of a green day, a 4–0 day and a green week.
How it is worked out. The weather score adds up three things: the EV of the last 30 trades, today's drawdown from the peak, and any current losing run of 4 or more. The day and week chances are exact binomial probabilities at the plan or measured win rate: a green day needs 3 or 4 wins out of 4, because 2–2 loses the cost.
How to read it. Use it as a mood check before the detail. It summarises numbers found elsewhere (Edge, Risk); it is not a forecast of any single trade.
Good sign: Sunny or Mostly sunny most of the time. Warning sign: Showers or Stormy for several weeks. Then read Risk › Gates and tiers and Edge › Edge decay.
Example. At 60% the chance of a green day is about 48% and of a 4–0 day about 13%: roughly one 4–0 day every two weeks.
Watch levels
Home › Watch levels
Your own alarm lines: drawdown from the peak at or past a level, EV at or below a level, and a losing run of at least some length. When one is crossed, a banner shows at the top of Home › Summary on your next visit.
How it is worked out. The levels are saved in your browser only (nothing is sent anywhere) and checked against the page's numbers each time it loads. EV is checked from 10 trades.
How to read it. Set levels that would make you want to look closer, for example a 13% drawdown (the Buffer gate) or 6 losses in a row.
Good sign: no banner. Warning sign: a banner. It is a prompt to read the detail, not a verdict.
Example. Set drawdown 13% and a run of 6. A week later the drawdown reaches 13.4%: the banner reads 'Drawdown 13.4% is past your 13% level'.
Is the edge holding up, and is risk right?
Edge decay check
Edge › Edge decay
Compares the EV of the first half of the record with the recent half, with a test of whether a drop is real or just noise.
How it is worked out. A Welch t-test on the two halves: t = (recent EV − early EV) ÷ the combined standard error. It starts at 40 trades so that each half has at least 20. A rolling 30-trade EV is shown as well.
How to read it. The verdict chip reads Steady, Improving, Possibly fading (t below −1) or Likely fading (t below −2). The 'chance by luck' figure says how often a drop this big would appear with no real change.
Good sign: Steady or Improving. Warning sign: Likely fading after 100+ trades. Markets change, so an edge can fade; it can also look like it fades through plain luck. It is a reason for a careful review at a milestone, not for an instant rule change.
Example. Early EV +0.18R over 60 trades, recent EV +0.06R over 60: t = −1.3, about a 10% chance by luck. Possibly fading: worth watching, nothing more.
Kelly check
Risk › Kelly check
The risk per trade the measured edge would justify by the Kelly rule, at full and half Kelly, next to the risk the rail is using now.
How it is worked out. For a 1:1 trade with cost c: Kelly f = (p·(1 − c) − (1 − p)·(1 + c)) ÷ ((1 − c)·(1 + c)). The range uses the 95% Wilson range of the win rate. Plan figures are used until 30 trades.
How to read it. Full Kelly maximises long-run growth but brings deep drawdowns; most professionals use half Kelly or less. A small error in the win rate makes Kelly far too big, which is why the range matters.
Good sign: the rail's risk at or below half Kelly. Warning sign: the rail's risk above half Kelly while the win-rate range is still wide. That is information only: the rail sets risk.
Example. At 60% and 0.05R cost, full Kelly is about 15% and half Kelly about 7.5%. The rail's 5.3% sits below half Kelly: conservative, by design.
Yardsticks
Performance › Yardsticks
The record next to simple alternatives over the same weeks: a savings account at 4.5% a year, doing nothing, and (once the data is added) holding EURUSD.
How it is worked out. Savings = 1.045 ^ (weeks ÷ 52) − 1. Doing nothing = 0%. EURUSD will be the change from the week-1 open, once the build supplies weekly EURUSD prices.
How to read it. The Monte Carlo benchmark answers 'is the edge real?'. The yardsticks answer 'was it worth the effort and the risk?'.
Good sign: a return far above the savings line, with a drawdown you could live with. Warning sign: a return close to the savings line: all the effort and drawdown for little extra.
Example. After 26 weeks the savings account would show about +2.2%. A record at +180% is far ahead; one at +3% would barely beat it.
What if the rules were different?
These three tools replay the actual trades in their real order, day by day, through a simplified MARS rail (gates, day colour, week status and the tier cap; the caution lock and manual tiers are left out), so their returns can differ a little from Equilibrium's own figures. They are research for the milestone reviews. Changing a rule because one of them looks better in hindsight is the classic mistake they warn about.
Skip-it what-if
Lab › Skip-it what-if
The record re-run without one session, pair, slot, weekday or timeframe, next to the actual record and to that slice on its own.
How it is worked out. The chosen slice is removed and the remaining trades go through the same rail. Columns: trades, net R, EV, win rate, the worst slide in R, the return on the rail and the maximum drawdown.
How to read it. Look at both the R and the return: removing trades lowers the number of chances to compound, so a better EV does not always mean a better return.
Good sign: no slice whose removal changes much: the edge is spread out. Warning sign: a slice that 'should be cut' on fewer than 100 trades. Picking the worst slice after the fact always looks clever (see the Filter-trap meter on Performance › Explore).
Example. Without the Rollover session, EV rises from +0.11R to +0.14R, but the Rollover slice has only 24 trades: most likely luck. Note it for the 100-trade review.
Fourth-trade value
Lab › Fourth-trade value
What the 4th trade of the day adds, and what a 'stop after two losses in a day' rule would have done.
How it is worked out. Three versions of the record through the rail: the actual (up to 4 a day), only the first 3 trades of each day, and a day that stops after its second loss. The 4th trades are also shown on their own.
How to read it. If the 4th trades have a positive EV, they add growth even though they add variance. A stop-after-two-losses rule cuts some bad days but also skips the wins that would have followed.
Good sign: 4th trades with a positive EV over 100+ trades. Warning sign: 4th trades clearly negative over 100+ trades, which would suggest fatigue or forced setups. That is a question for a milestone review.
Example. Over 200 trades the 4th trades show +0.04R against +0.13R for the first three: a positive but weaker edge. Dropping them would cut the return by about a sixth.
Mode comparison
Lab › Mode comparison
The same trades under other risk settings: Conservative (tier cap T3), Balanced (cap T5, the live setting), Aggressive (cap T7), and flat 1% or 3% a trade.
How it is worked out. Each version replays the trades with the same gates and tiers but a different cap, or a fixed risk. Columns: return, maximum drawdown, the average risk a trade, and whether System Lock was hit.
How to read it. This is the trade-off in one table: a higher cap means more growth in good runs and deeper holes in bad ones. Past order only: the Simulator shows the spread over many possible orders.
Good sign: Balanced close to Aggressive's return with a clearly shallower drawdown. Warning sign: Aggressive far ahead with a similar drawdown over 300+ trades. Even then, the mode is not changed before 30 trades, and a change needs a proper test.
Example. Balanced +140% with a 16% max drawdown; Aggressive +190% with 27%; flat 1% +35% with 6%.
Rule-change impact
Records › Rule-change impact
For each entry in the rule-change ledger: trades and EV before the change against on or after it, and the change in win rate.
How it is worked out. The record is split at the change date. Changes made before go-live (16 November 2026) have no 'before' trades.
How to read it. A before/after gap only means something with 30+ trades on each side, and even then other things changed at the same time (markets, season, the trader's experience).
Good sign: 'Comparable' samples with no big swing after a change. Warning sign: drawing conclusions from 'Too few trades on one side'.
Example. A cost change in week 10: 144 trades before at +0.10R, 96 after at +0.13R, win rate +2 points. Promising, but not proof.
Looking back: a trade, a week, a season
Trade card
Records › Trade log (click any row)
The full card for one trade: date and time, session, timeframe, slot, minutes held, exit, net R, all-in cost, tier and gate, the day's risk, the plan check and the quality grade. Edition 19 adds the two trades before and after it, the review checklist line and ◀ Previous / Next ▶ buttons.
How it is worked out. The checklist line comes from the trade review: how many of the 14 mandatory items passed, which item was breached, and how many were not visible on the chart.
How to read it. Use Previous / Next to walk through a day or a streak trade by trade.
Good sign: compliant trades with a clean checklist, whatever their result. Warning sign: repeated breaches of the same item (also on Execution › Breach cost and the mistake library).
Example. Trade #37, a −1.03R loss: 14 of 14 items passed, no breach. A good trade that lost, which is normal 4 times in 10.
Replay a week
Records › Replay a week
Step through one week trade by trade with ◀ ▶ or Play: the account line for the week, each result, the gate, the tier and the risk the rail used.
How it is worked out. A per-trade trace of the simplified rail: each day's risk is set by the gate (drawdown from the peak), the previous day's colour and the week's status, capped at T5.
How to read it. Watch how the tier steps down after a red day and back up after green ones: the rail working in miniature.
Good sign: a bad day followed by a smaller risk the next day. Warning sign: none: this is for understanding, not judging.
Example. Week 6: a 1–3 Monday drops the tier for Tuesday; a 4–0 Wednesday lifts it again for Thursday.
Week report
Records › Week report
One week on one page: the four headline numbers, the week's story, the lesson of the week from the weekly review, a bar chart of every trade and the trade table. '⎙ Save this week to print' saves a standalone file that opens the print dialog (choose Save as PDF for a PDF).
How it is worked out. The story is the same rule-based text as Records › Week stories. The lesson is the first focus item of that week's weekly review.
How to read it. Use it to file each week with your journal or to share a single week.
Good sign: a story and lesson that match how the week felt. Warning sign: none.
Example. Week 3 report: +4.2R, 10–6, EV +0.26R; the story names Thursday's 4–0 as the best day, and the lesson is 'wait for the candle to close across the 55'.
Season recap
Records › Hall of fame and seasons
A six-slide show for any 13-week season, opened in Present: the season's results, net R by week, the best week and best trade, the lessons of its weeks, execution (graded trades and compliance), and what comes next.
How it is worked out. Seasons are weeks 1–13, 14–26 and so on, as in the Quarter seasons table. Esc or ✕ Close returns to the page.
How to read it. Good for a quarterly look back, or to show someone how a season went.
Good sign: more green weeks than red, and compliance near 100%. Warning sign: a season carried by one big week: read the weekly bars.
Example. Season 1: 13 weeks, +18.5R, EV +0.09R, 9–4 green–red weeks; best week 7 (+6.1R).
Just for fun
These three are games, kept in your browser only. Each makes a serious point about chance: single trades and single weeks are hard to predict even when the edge is real.
Predict the week
Lab › Predict the week
Lock in a guess for next week's net R before it starts; after the week, see how close you were. Within 1R is a bullseye, within 3R is close.
How it is worked out. The guess is saved in your browser. A week's normal range at the plan is roughly −3R to +7R.
How to read it. Most guesses miss by several R. That is the lesson: a single week is noisy.
Good sign: a few bullseyes over a season. Warning sign: none.
Example. You guess +3R; the week ends at +1.6R: 'Close'.
Week bingo
Lab › Week bingo
A 4×4 card of things that may happen in the latest week, such as a 4–0 day, 3 or 5 wins in a row, a new equity high, back-to-back green days or all 16 trades taken. Squares fill in from the real trades; full rows, columns and diagonals count as bingo lines.
How it is worked out. Each square is a rule checked against the week's trades.
How to read it. Just for fun: no square is a goal, and chasing one would break the plan.
Good sign: a few lines on a good week. Warning sign: none.
Example. Week 4: 9 of 16 squares and 1 bingo line.
Guess the trade
Lab › Guess the trade
One past trade at random, with its pair, direction, weekday, time, session, timeframe and slot. Guess whether it won, then see the result and your running score.
How it is worked out. The trade is drawn at random from the record; the score is kept in your browser. The note compares your score with always guessing 'win', which scores about the win rate.
How to read it. If you can't beat 'always win' by much, that shows how little a setup's surface details predict one result.
Good sign: a score near the win rate. Warning sign: none.
Example. After 20 guesses you have 12 right (60%), the same as always guessing 'win'.
10 · Common questions
Does the page keep anything about me?
Only conveniences, and only in your own browser: your pinned cards, your saved Explore views, your light / dark choice and what the page looked like on your last visit (for the "since your last visit" strip). None of it leaves your device or reaches the trader.
Why are results in R instead of dollars?
R makes trades comparable when risk changes from day to day. It also keeps the account size private. A +10R month means the same thing whether the account is small or large.
The account is down this week. Is the system failing?
Probably not. Check three things: is the sample still small (under 30, or even 100, trades)? Is the account still inside the benchmark band (chart 1)? Is the cumulative R (chart 2) still rising over the longer run? Losing weeks are built into any 60% strategy.
What counts as "alpha" here?
Beating the Monte Carlo benchmark for the same plan without a worse drawdown, worse efficiency or worse rule compliance. Higher return from simply taking more risk doesn't count.
When is the verdict trustworthy?
After 30 trades the statuses start to mean something. After about 170 trades a +0.15R edge becomes statistically clear. The verdict is still called provisional until 300 trades.
Does a bad verdict stop the trading?
No. The verdict informs reviews. The only automatic stop is the System Lock at a 40% drawdown, which sets risk to zero. Beyond that, the trader judges whether the test is still worth continuing.
Explore shows a great result for one pair. Should the trader only trade that?
Not on its own. Check the trade count and the Provisional chip first. Small slices swing wildly by luck, and choosing the best of many slices after the fact almost always overstates it. A slice earns attention only with many trades and a result that holds as more arrive.
Why don't the Simulator's numbers match the benchmark on Home?
They answer different questions. The benchmark on Home simulates the full MARS rules for the official ruler. The Simulator's Constant risk mode uses one fixed risk and no gates so you can see each setting's effect clearly; its MARS rail mode is closer to the real rules but simplified (no caution lock). Each click of "Run again" also draws new random runs, so its numbers move a little every time.
The finish-line projection says the goal chance is under 50%. Is MARS failing?
No. Growing an account about 110-fold in 78 weeks is an ambitious goal: even at the plan's 60% win rate, about half of the benchmark paths fall short. MARS is judged first on EV and drawdown. A real edge that misses the goal is still a success; a goal reached by luck without an edge is not.
Mode comparison shows Aggressive would have made more. Why not switch?
Because one past order of trades is one draw from many. Aggressive also digs deeper holes, and the next bad run may come early. No rule or mode changes are made before 30 trades, and any change needs the 100- or 300-trade evidence and a proper test, not one replay.
Skip-it shows the record is better without one session. Should that session stop?
Not on this evidence alone. Removing the worst slice after the fact always improves the past. The slice needs 100+ trades, a reason that makes sense before looking at results, and confirmation at a milestone review.
Which decision tool should I open first?
For the day ahead, the gate distance ladder (Risk). After a bad patch, the recovery clock. To understand why MARS sizes the way it does, Growth vs risk, then the Simulator in Constant risk vs MARS rail mode with A vs B. For the coming week, the next-week forecast.
The what-if shows a better stop or target. Should MARS switch?
No. The what-if is rough by design: it counts trades that touched both levels as losses, and moves after the real exit may not be recorded. It is for research questions only. MARS keeps the 1:1 hard stop for the 78-week test, and any change would need a proper test first.
What does the edge-proof meter need to say "proven"?
The whole 95% range of the EV above zero. With a +0.15R edge and normal 1:1 results that usually takes roughly 170 trades or more. Before 30 trades it always shows Provisional.
Why do some numbers show a dash?
A dash means "not enough data yet". For example, the ghost figures need trades with logger data, and the weekly scorecard starts with week 1.
Quick reference card
| Question | Where to look | Healthy reading |
|---|---|---|
| Is there an edge? | EV tile · chart 2 · Edge › Edge statistics | EV > 0, near +0.15R; cumulative R rising |
| Is it real or luck? | EV 95% range · t statistic | Low end of range > 0; t ≥ 2 |
| Can it survive? | Max drawdown · chart 5 · gate | Growth or Recovery gate; dips recover |
| Is risk used well? | RAER · RAPF · chart 6 | Inside or above the benchmark P25–P75 |
| On schedule? | Chart 1 · Five core metrics | Inside or above the violet band |
| Is cost eating the edge? | All-in cost · chart 20 | Cost well under a third of the gross edge |
| What changed since I last looked? | Home › Since your last visit | New trades listed; EV and drawdown moving the right way |
| How did one week go? | Records › Week stories · Share card | Read the story, judge the trend across weeks |
| What can tomorrow do? | Risk › Tomorrow | Most of the chance on the green branches; the next gate several losses away |
| Is the rail worth it? | Risk › Rail value-added | Shallower drawdown and far lower lock chance than flat risk |
| Is the edge proven yet? | Edge › Sequential test · Payoff map | Cumulative R near or above the confirm line; most of the cloud above break-even |
| What do breaches cost? | Execution › Breach cost | High compliance; breached trades no better than compliant ones |
| Can the record be trusted? | Records › Integrity (with the rule-change ledger) | Health check passed, statement matches, fingerprints unchanged, rules frozen |
| What appears next, and when? | Home › What unlocks when | Ticks moving down the list; nothing blank without a reason |
| Is this slice big enough to read? | The n label · faded rows | 30+ trades before comparing; faded rows are mostly luck |
| Do the page's numbers agree? | Records › Integrity › Self-test | All checks green; header chip shows ✓ |
| How many losses to the next gate? | Risk › Gate ladder and recovery | Several losses to the next gate; many more to System Lock |
| How much can costs rise? | Performance › Cost › Cost safety margin | Cushion of 5+ points; edge-erasing cost far above today's |
| Just the basics, please | ◐ Simple view (menu) | Ten pages; switch back any time |
| Tell someone how the week went | Share card › Copy this week's update | Short text with the link; R and % only |
| What does normal bad luck look like? | Lab › Trading school · Play a week · Edge › Stability (streak odds) | Losing streaks of 5–7 and roughly 1 red week in 4 are normal at 60% |
| Where is the edge strongest? | Explore · Compare by | Differences that hold with 100+ trades per slice |
| How lucky or unlucky so far? | Simulator · Your rank | Anywhere from about 25% to 75% is normal |
| Is the edge proven yet? | Edge-proof meter (A1) · A2 | Needle in green, 95% range above 0 |
| Is execution improving? | Analytics A34–A35 | Grade average at B or better; full compliance |
| Ahead of schedule? | Road to week 78 (Home) | Blue dot at or past the orange plan marker |
| How close is the next gate? | Risk › Gate ladder (tool 4) | Floor and deeper gates unlikely within 32 trades |
| How long to recover a dip? | Risk › Recovery clock (tool 5) | Median a few weeks; low System Lock share |
| Does the rail earn its keep? | Lab › Simulator, A vs B (tools 1–2) | Rail: far lower System Lock share than constant risk |
| Is the trader disciplined? | Plan checks · chart 18 · Trade log | Exits at TP / SL; no over-risk trades |
| Is a losing run normal? | Edge › Stability · Streak odds | Expected or Normal; no shift in the change-point table |
| Is this week normal? | Home › Pace and finish line | Inside the normal range; red weeks about 1 in 4 |
| Is the goal on track? | Home › Pace and finish line (Required pace, Finish line) | Achieved pace at or above needed; EV range above 0 |
| Is the edge fading? | Edge › Edge decay | Steady or Improving; t above −1 |
| Is risk sensible? | Risk › Kelly check · Lab › Mode comparison | Rail's risk at or below half Kelly |
| What if a rule were different? | Lab › Skip-it · Fourth-trade value | No change on fewer than 100 trades |
| File or print a week | Records › Week report · Replay a week | Story and lesson match the week |
| Are fills clean? | Execution › Fill quality | 80%+ within ±0.05R of ±1R; no drift |
| How is this number calculated? | ⓘ on any card · ▦ for the table | Formula in words, the inputs and the sample size |
| Which mistakes repeat? | Execution › Breach cost (mistake library) · checklist heatmap | No item breached on more than 1 trade in 20; low not-visible rates |
| Beating the benchmark? | Performance › Benchmark rank · Frozen forecasts | Rank near or above 50%; about 8 in 10 forecasts inside |
Important. This guide and the MARS Equilibrium Track Record are published for education and transparency only. They're not investment advice, a recommendation to trade, or an offer of any financial product. Trading leveraged products such as forex and CFDs carries a high risk of loss, and you can lose more than you expect. Past results, simulated benchmarks and statistical ranges do not guarantee future results. The 60% win rate is a planning assumption until it's measured on live trades.
MARS · The Montex AlphaRail System · Aura Logic Systems · "Turning Expectancy into Alpha."
Edition 20 · 11 November 2026 (menu wording: on the website the menu runs down the left side of a desktop screen; adds the colour key). Edition 19 · 11 November 2026 (adds section 9, the planning, what-if and fun tools: pace and finish line, record weather, watch levels, edge decay, Kelly check, yardsticks, skip-it, fourth-trade value, mode comparison, rule-change impact, the trade card, replay a week, week report, season recap, predict the week, week bingo and guess the trade; Edition 18 added the 16-step tour, the Today line, ⓘ and ▦ on every card, chart ranges and zoom, text size, print, new-data dots, Recent pages and What's new; Edition 17 added the trading school and the process-and-coaching cards). A printable PDF edition of this guide is available from the publisher on request.