How we test strategies
Exact rules, real costs, data we did not tune on.
Many trading videos make a strategy look easy. We take some of them and check: if you followed the rules exactly, on years of real price data, after paying realistic trading costs, would it have made money? Each tested video gets one of five labels.
1. Turn the video into exact rules
We take the strategy extracted from the video and write it down as mechanical rules: when to enter, where the stop goes, when to exit. Where the video leaves room for judgement, we pick the simplest reading and say so. If no fair set of rules can be written, we label the video "Too vague to test" instead of guessing.
2. Test on historical minute data
The rules run on historical minute data from Dukascopy, 2016 to 2026, across stock indices, major FX pairs, gold and crypto. Longer timeframes are built from the same minute data.
3. Charge realistic retail costs
Every trade pays the spread and slippage a retail CFD trader would see. Positions held overnight also pay funding. Many strategies that look good on a chart stop working once these costs are included.
4. Fix the rules first, then check the future
Rules are fixed before we look at any results. We judge them on 2016–2022 (in-sample) and then on 2023–2026 (out-of-sample), a period we do not tune on. A strategy only holds up if it makes money after costs in both.
What the labels mean
The mechanical rules made money after realistic trading costs, both in the period we designed the test on and in the later period we held back.
The rules could be tested, but after spread, slippage and funding they lost money or the result did not survive the held-back period.
The video made specific claims (win rate, returns, profit factor). Testing the rules as described did not come close to those numbers.
The rules could be tested, but over the whole test period they produced fewer than 30 trades. That is not enough to judge the strategy either way.
The video does not give rules precise enough to turn into a test without us inventing large parts of the strategy.
The rules are clear, but they need data we do not have (for example order flow, options chains, news or individual stocks).
Loading counts…
Limitations
- Our encoding of the rules may differ from how the creator actually trades them. Discretion, experience and context are hard to capture in rules.
- Tick and range charts are approximated with time-based bars.
- We only test markets we have data for: indices, FX, gold and crypto. Individual stocks, options and order-flow strategies are out of scope.
- Past results do not predict future results. A strategy that held up in our test can still lose money.
- Nothing here is financial advice.
Think we encoded a strategy wrongly? Tell us with the feedback button. Browse strategies →