Evidence before confidence
Historical backtest.
Test the strategy on real historical candles—not invented market results.
1. Load historical candles
Kraken returns up to 720 recent minute entries, including the uncommitted row that is excluded. For longer periods, import an official historical archive.
Header: time, open, high, low, close. Unix seconds or ISO timestamps with UTC offset. Minute boundaries, ascending order. Minimum 300 rows. No gap-filling.
2. Freeze training configuration
Tune only on the first chronological segment in Strategy settings. Freeze before inspecting the unseen test results.
3. Evaluate unseen candles
No historical data loadedStart with evidence
Load a genuine historical provider snapshot or import your archive. There are no fabricated results or pre-seeded performance numbers.
Each decision sees only candles closed before entry; swings require right-side confirmation. Entry is modeled at the next candle open and exit at its close. Real signals arrive after that open. Actual fills, latency, spread, slippage and Quotex payouts are not verified. Reusing a holdout after tuning makes it no longer unseen.
Crypto · historical performance
Saved tests for selected assetResolved trades
0
Win rate
—
Loss rate
—
Draw rate
—
Expected return / stake
—
Maximum drawdown
—
0 evaluations; NO TRADE excluded from trade rates. Draw refunds stake before modeled cost. Unit-stake return = win payout − losses − costs. No historical accuracy claim is made. These are provider-candle observations, not broker execution results.
Holdout decision ledger
No historical results yet
Freeze your configuration and run the unseen segment. Zero trades is a valid result—not a reason to fabricate performance.