YinYang Harness

Inventory-Accurate Backtest

This run replays saved market samples through the same YinYang quote, fill-through-limit, inventory hedge, pair-completion, and settlement logic used in paper trading.

Scope: -- Markets: -- Samples: -- Strategies: --

Idle

Portfolio Equity
--
Portfolio PnL
--
Settled Markets
--
Portfolio Win Rate
--

Overlay Mode Comparison

Direct A/B/C comparison from the same dataset: spread-only baseline, spread + directional bias, and spread + directional bias + regime filters.

Mode Portfolio PnL Portfolio Win% Top Strategy Top Score

Market Pattern Research Lab

Scientific exploration on Binance candle history for directional edges. This complements YinYang spread simulation with regime and indicator research.

Research idle.

1m Candles
--
5m Candles
--
5m Baseline Up
--
Top Edge Lift
--

Top Edge Candidates

Edge Side Hit Rate Lift Support

Research Guidance

No guidance loaded.

Strategy Leaderboard

Sorted by YinYang score, then total PnL. Click a strategy row to inspect its equity curve and market-level inventory outcomes.

Strategy Score PnL Settled Win% Max DD

Selected Strategy

Select a strategy from the table.

Recent Settlements

Market Result Equity Pair Shares Excess

Signal Attribution

Per-mode decomposition of what drove performance uplift: market momentum score vs research score (hour + ATR), plus regime blocking rate.

Mode Active Signal % Regime Block % Avg Strength Avg Market Score Avg Research Score Hour Score ATR Score

Out-Of-Sample Split

Train vs forward split to reduce overfitting risk. Train uses earlier markets; forward uses the most recent held-out markets.

Split: --

Train Window

Mode PnL Win% Top Strategy

Forward Window

Mode PnL Win% Top Strategy