Rules before results
Before the first run the Lab saves the strategy description, its SHA-256 hash and the date. The report always shows them — changing one number makes a new strategy, not a fix.
RBR Strategy Lab tells you whether your strategy survives out of sample
You describe your strategy from ready-made building blocks before you see a result. Strategy Lab freezes it, tests it on real ticks with your account's costs, compares it with random entries and with holding the market, and shows the last 30% of the data only once. Most strategies will hear “does not pass” — and that is exactly what the tool is for.
We are refining this version — the file will be back after testing.
Full version: €149 · 12 months · up to 3 MT5 accounts
Lifetime licence: €379 once, no expiry date · Renewal for another 12 months: €99, counted from the end of your current licence
Purchase opens after approval by the payment provider.
Try the free version before buying – no time limit. After the key is delivered, refunds only for defects. Terms of sale and licence · Refund policy
What it does differently from an ordinary tester
One out-of-sample check (full version)
The last 30% of the period is the check. Its result for a given strategy is shown once; later runs show the same, saved result. You cannot tune a strategy while looking at the check.
Attempt counter
Every new version on the same symbol and period is another attempt. The Lab counts them and subtracts from the average how much the best of M attempts looks better by luck alone.
Costs from ticks and your account
Entries and exits at the real ASK and BID prices of every test minute. Commission and broker markup are measured from your account history; swap comes from the symbol.
Two benchmarks
Random entries with the same SL/TP geometry and the same costs, and holding the market over the same time at the same risk. The strategy has to beat both, not just zero.
A bot from the same rules (full version)
If the strategy passes, the Lab writes the code of an MT5 bot with exactly these rules. On start the bot checks the rules' hash and refuses to run if anyone changed them.
How to use it
- Download the .ex5 file. In MetaTrader choose File → Open Data Folder, copy the file to MQL5\Experts and click Refresh in the Navigator.
- Drag the Lab onto any chart of your account once. It reads the trade history, measures commission and broker markup, saves them for the tester and stops — it trades nothing. On an account without trades you enter the commission in the parameters.
- Open the Strategy Tester (Ctrl+R), choose the Lab, symbol, timeframe, period and the “Every tick based on real ticks” model. Describe the strategy in the parameters: direction, entry condition, up to two filters and the exit. Start the test.
- The Lab writes the path to the HTML report in the tester journal; usually %APPDATA%\MetaQuotes\Terminal\Common\Files\RBR_LAB.
What you build a strategy from
Entry (1 or 2 conditions)
SMA/EMA crossover, RSI crossing a level, Donchian channel breakout, close outside the Bollinger band, MACD crossing its signal, a large candle (range ≥ k × ATR) or any custom indicator from a buffer. Signal on the closed candle, entry at the next candle's open.
Filters (0–2)
session hours, days of the week, trend from a higher timeframe (closed candles only), volatility level.
Exit
a stop is mandatory (k × ATR or points), a target in R or none, exit after N candles, exit on the opposite signal.
The set is small on purpose. The report counts free parameters and warns above three — each extra one makes an accidental fit to the past easier. No trailing stop and no move to break-even: in our measurements such exits cut off the best trades.
8 criteria — fixed in code before you see a result
- at least 100 trades in the learning part;
- profit factor ≥ 1.25;
- better than random entries with the same geometry and costs (p < 0.05);
- out-of-sample check averaging above zero (full version);
- still above zero on average without the 5 best trades;
- at least 60% of years positive (when the test covers ≥ 3 years);
- better than holding the market over the same time at the same risk;
- average after correcting for the number of attempts above zero.
Verdict: PASSES, TRACE or DOES NOT PASS — with every reason and value. The thresholds are not a setting: changing them makes a new product version. Why criterion 8: on a market with no edge at all, the best of 100 random strategies passed the other criteria 21% of the time; with the correction, about 4%.
Walk-forward (full version)
You can tune at most two parameters, on a grid of up to 50 combinations. The Lab then judges the tuning procedure, not one set of numbers: in the learning part rolling windows learn on 3 segments and trade the next 1, four test segments are stitched together and the criteria are calculated on them. Every grid combination counts as an attempt. The check and the bot get the last window's parameters. The bot does not retune itself — the report says every how many days to repeat the Lab.
Bot generator (full version)
By default the bot is created only with a PASSES verdict. With another verdict you can generate it deliberately — then the file header and a message on start say plainly which criteria the strategy failed. The bot is one .mq5 file to compile in MetaEditor: risk as % of equity, stop and target in the order at the broker, a journal of every decision. It runs only on the symbol and timeframe of the test. In the Strategy Tester match test the bot repeated 243 of 243 of the Lab's exits. The bot trades on its own once you start it — your rules, your decision. Start on a demo account.
Lite and full version
| Lite (free) | Full (€149) | |
|---|---|---|
| Freezing, costs from ticks, attempt counter, report in EN/DE/PL | yes | yes |
| Random entries | 200 series | 1,000 series |
| Out-of-sample check (one look) | — | yes |
| Highest possible verdict | TRACE | PASSES |
| Walk-forward (up to 2 parameters, ≤ 50 combinations) | — | yes |
| Trade export (CSV) | — | yes |
| MT5 bot generator | — | yes |
Limits — read before testing
Not a strategy search engine
The Lab does not search thousands of variants to show you the best one. The best of many looks good mostly because you picked it after the fact. We count every attempt and charge it against the result.
One period, one instrument
Even PASSES applies to one symbol and one period. Check another period and another instrument before you believe it.
Price jumps within a minute
The Lab books a stop exit exactly at the stop level. On a sharp jump the broker fills worse — in the match test −0.013 R per trade on average, −0.54 R in the worst case.
Market indicators
They may refuse to run in the tester or inside another EA. That depends on the indicator's author.
MetaTrader 5 only
The MT5 Strategy Tester with real ticks. It does not run in MT4.
How we tested this tool
Every calculation has a Python reference, and the MQL5 result must match it to the last digit: criterion 8 on 266 vectors, walk-forward on 36 grids and 40 random window sets — zero differences. In the tester the Lab's result equals the reference computed on the same ticks in every control run (EURUSD and XAUUSD, M15 to H1, walk-forward included). A bot generated by the Lab reproduced 599 of 599 signals and 243 of 243 exits in the tester; every difference has a documented cause.
Questions
Does the Lab trade?
No. The Lab runs in the Strategy Tester, where everything is virtual; on a normal chart it only reads your account history. Only the bot trades — the one you generate and start yourself.
Does it send my data?
No. Account history, costs, reports and bots stay on your computer, in MetaTrader's files.
Why doesn't my strategy pass?
Because most don't — ours included. The report shows every criterion with its value, so you see whether trades are missing, the edge over random entries is missing, or the result hangs on a few best trades.
What does TRACE mean?
The effect points the right way but is too weak or too short to count as evidence. In Lite it is the highest verdict, because without the check nothing can confirm the result.
Why doesn't it run in optimization?
On purpose. Optimization picks the best of many variants, and that is exactly why results from history shrink later. If you want to tune parameters, use walk-forward: it counts every combination as an attempt.
How many accounts does the licence cover?
Up to 3 MetaTrader 5 accounts for 12 months; the key is tied to the account number.
An analysis tool. Not investment advice and no forecast of results. The Lite version is free, the full version is paid; Terms of sale and licence and Refund policy apply.