Strategy Performance
Backtest Results 2021–2026

One agent · one account · one strategy · your signal source
Three reference strategies with transparent backtests · Binance USDⓈ-M Futures · 5% margin · x5 leverage
RSI Crossover — 16 coins · BearBull — 7 coins · BearBull Alt — 18 coins
A 50/50 mix (BB Alt / RSI) is two agents on two accounts — not one book inside one process.

5x Leverage 5% Margin 0.75% Risk/Trade

Self-hosted Your keys Past performance ? future results

Reference Strategy

RSI Crossover

2021+9.21%

2022+130.33%

2023+100.35%

2024+144.97%

2025+83.93%

2026 (Jan–Sep)+37.31%

Total PnL (6y)+506.10%

Avg Win Rate~53.9%

Worst Max DD−19.1%

Capital Exposure (max)80%

Risk Exposure0.75% – 12%

Total Trades (6y)8 621

Market Orders Only Fee: Taker 0.05%

Reference Strategy

BearBull

2021+48.89%

2022+53.91%

2023+32.06%

2024+69.72%

2025+19.28%

2026 (Jan–Sep)+14.24%

Total PnL (6y)+238.10%

Avg Win Rate~59.1%

Worst Max DD−10.54%

Capital Exposure (max)35%

Risk Exposure0.75% – 5.25%

Total Trades (6y)4 435

Market Orders Only Fee: Taker 0.05%

Reference Strategy

BearBull Alt

2021+54.11%

2022+107.77%

2023+51.99%

2024+269.56%

2025+26.06%

2026 (Jan–Sep)+34.62%

Total PnL (6y)+544.11%

Avg Win Rate~59.3%

Worst Max DD−17.65%

Capital Exposure (max)90%

Risk Exposure0.75% – 13.5%

Total Trades (6y)9 785

Market Orders Only Fee: Taker 0.05%

Methodology

Same 0.75% risk per trade across all reference strategies (position risk from margin × leverage × structural stop)
Risk exposure scales with concurrent positions — 0.75% per trade, up to coin count (RSI 0.75–12%, BearBull 0.75–5.25%, BearBull Alt 0.75–13.5%)
Non-compounded annual returns — balance resets each year; six-year total = sum of yearly PnL
Capital exposure differs by strategy — margin per position (RSI 80%, BearBull 35%, BearBull Alt 90% at 5%)
Sharpe / Sortino — daily returns (strategies: annual balance reset; benchmarks: compounded B&H); risk-free rate = 0; period 2021–Aug 2026

S&P 500 and Bitcoin buy-and-hold use compounded total return when compared in benchmark sections.

Benchmark Comparison

Three reference strategies vs Bitcoin and S&P 500 buy-and-hold · 2021–2026

Strategy	Total PnL	Avg/Y	Win Rate	Max DD	Sharpe	Sortino
BearBull Alt	+544.11%	+90.7%	59.3%	−17.65%	1.93	3.53
RSI Crossover	+506.10%	+84.4%	53.9%	−19.1%	1.77	3.40
BearBull	+238.10%	+39.7%	59.1%	−10.54%	1.92	3.56
Bitcoin B&H	+116.13%	~19.4%	—	−64.3%	0.43	0.64
S&P 500 B&H	+115.72%	~19.3%	—	~−18%	0.84	1.22

Sharpe and Sortino from daily returns; strategies use non-compounded annual segments, benchmarks use compounded buy-and-hold (BTC-USD, SPY). Color scale: >1.5 strong · >1 good · <1 weak.

Oct 10–11, 2025: during the historic flash crash, all three reference strategies were correctly positioned (BTC SHORT), took profit within the first hour, and did not open new trades at the liquidation peak — resilient behavior with no manual intervention.

Risk Management Model

Common to all reference strategies — 0.75% risk per trade; risk exposure scales with open positions; capital exposure (max) differs by coin count and margin per position

Margin Model

Isolated

Per-coin isolation, no cross-exposure

ROA per Trade

Margin allocated per open position

Leverage

×5

Fixed, applied uniformly on all pairs

Risk per Trade

0.75%

Of total capital at base multiplier = 1.0

Adaptive RiskController

Multi-window PF

Tracks profit factor across fast (30), mid (100) and slow (300) trade windows simultaneously. Weighted composite shifts toward the fast window during win streaks.

Asymmetric EMA

Risk falls fast (?=0.3) when PF deteriorates, recovers slowly (?=0.1) when PF improves. Win streak of 3+ moderately accelerates recovery; 5+ makes recovery symmetric with the fall.

Drawdown Gate

Penalty starts at 10% drawdown and scales linearly to 50% risk reduction at 30% drawdown. Independent of the PF multiplier — applied on top.

Time-in-DD Penalty

If equity stays in drawdown zone for more than 3 days, a second penalty reduces risk to 0.7? over the next 3 days. Resets immediately on new equity high.

Risk multiplier range

0.35? 1.0? base 1.2?