Backtests are not winning guarantees. Rankings are historical after assumed futures fees and slippage; ongoing strategy search repeatedly reuses the data and can overfit. Signals are based only on closed bars; entries occur at the following bar. Real orders remain disabled.
Strategies tested
—
Combinations generated so far
Active combinations
—
Weak variants retired over time
Validation + holdout positive
—
Historical only; not independently proven
Data coverage
—
BTC · ETH · SOL futures · 1m
Running indicator calculations and backtests. Data will appear after the first completed research run.
| # / Strategy & Confirmations | Exit plan | Train trades | Validation trades | Validation win% | Validation net/trade | Validation total ($) | Holdout win% | Holdout net/trade | Assessment |
|---|---|---|---|---|---|---|---|---|---|
| Waiting for historical tests… | |||||||||
bps = 0.01% of notional. Approximate simulated P/L for $1,000 notional: 1 bps = $0.10. Holdout is reported, not used to rank, but repeatedly inspecting it weakens independence.
Entry indicator / price actionVariantsBest net / tradeBest win
Loading entry rankings…
Confirmation indicatorVariantsBest net / tradeBest win
Loading confirmation rankings…
01 / No repainting
Candle-close confirmationCompute signals using known completed candles. Simulated entries occur on the next 1-minute candle's open, never retroactively.
02 / Conservative fills
Stop first if both levels hitIntrabar sequence is unknown from OHLCV alone. Ambiguous TP/SL events assume the adverse stop triggers first. Apply assumed 0.05% taker fee and 0.01% slippage per side.
03 / Chronological validation
60% / 20% / 20%Train, validation and later-date holdout windows with 40-minute gaps. New indicator combinations are added on recurring test cycles; repeatedly viewing a holdout still introduces bias.