Research

We test the trading ideas we teach on real Binance futures data, with the same code our tools run and a random-entry control for every result. Most ideas don't survive. We publish what we find either way.

7 Nov 2026 · 11 min read

We found an edge in 4-hour chart patterns. Then we tested it properly.

Our first test of 4H chart patterns looked like a real edge: +0.24R a trade. Re-run on 30 coins and 1,603 trades, it shrinks to +0.02R better than random — and to nothing on coins the rule was not built on.

7 Nov 2026 · 10 min read

Why a 44% win rate at 1.2R still loses — and why changing exits won't fix it

A day of paper trading: 44% winners, a 1.2R target — and a loss. We replayed the strategy over 1,621 trades with five exit rules. Every one lost about the same. The real problem was 0.09R of fees on a +0.03R edge.

How we test

The real rule. Backtests run the same code as the app and its bots, not a simplified copy.
No look-ahead. Every decision sees only the candles and market data that existed at that moment.
Real costs. Binance's standard fees and slippage on every fill; when a candle touches both stop and target, the stop counts first.
A random control. Every strategy is compared with random entries using the same stops, targets and fees.
Results in R — profit or loss in units of the amount risked — with 95% ranges, so you can see how certain each number is.