Two-Bot Portfolio: Relative Strength (Aggressive) + Bollinger Breakout
Two-bot portfolio, moderate risk (~-30% drawdowns possible). Half the money runs Relative Strength Long/Short (Aggressive), half runs Volatility Breakout (Bollinger); the split is reset on the first day of every month, with the cost of resizing positions charged.
Two-bot portfolio · Moderate risk: ~-30% drawdowns possible Backtest/paper only for now
Real risk. At its worst point in our test this portfolio was down -27.9% from its peak, and a future drawdown could be deeper. Only use money you can afford to lose.
Ruin odds: In 2,000 simulated years built by reshuffling 10-day chunks of this backtest: 0.0% hit a drawdown of 50% or worse, 0.0% lost 90% or more of the account, 7.2% ended the year with a loss, and the middle outcome was +69%. These come from the same past data, so real losses can be bigger: fat-tail events the backtest never saw, exchange outages and fills worse than modelled are not in them.
The two bots
| Bot | Share | Avg month alone | Max drawdown alone | Deflated Sharpe alone |
|---|---|---|---|---|
| Relative Strength Long/Short (Aggressive) | 50% | +5.43% | -49.4% | 0.99 |
| Volatility Breakout (Bollinger) | 50% | +3.18% | -32.7% | 1.00 |
Daily returns of the two bots are correlated 0.32 (1 = move together, 0 = unrelated), which is why the portfolio's drawdown is smaller than the riskier bot's.
Tested result
Passed out-of-sample Profitable on data it was not tuned on and passed every luck/overfitting check. Still not a guarantee.
Test: Two bots, 50% Relative Strength Long/Short (Aggressive) / 50% Volatility Breakout (Bollinger), rebalanced monthly; Binance USD-M Futures, 2019-09-08 to 2026-11-19 (common period of both bots), out-of-sample from 2024-09-21. Run 2026-11-19 20:09 EST.
| Average month (whole test) | +4.56% |
|---|---|
| Average month (out-of-sample) | +3.60% |
| Max drawdown | -27.9% |
| Average month (in-sample / design period) | +4.97% |
| Walk-forward periods profitable | 5 of 5 |
| Luck checks | beats 100% of random-entry versions of both bots combined; deflated Sharpe 1.00 after 35 variants tried across both bots and all portfolios (needs 0.95+) |
| Rebalancing costs | 86 monthly resets, about 0.002% of the account each (0.03% a year). Charged as money moved × each bot's open position size × (taker fee + half-spread). Small because each reset only moves the drift between the two bots. |
| Trades (both bots) | 6199 |
| Profit factor (all / out-of-sample) | 1.21 / 1.20 |
By market regime
| Regime | Days | Avg month |
|---|---|---|
| bull | 1434 | +6.00% |
| bear | 950 | +3.60% |
| chop | 246 | +4.80% |
How the portfolio works
Each bot trades in its own sub-account. The money starts split by the shares above. On the first day of every month the split is reset: money moves from the bot that did better to the one that did worse, and both bots resize their open positions, which costs fees and spread. The backtest charges that cost every month. Both bots run on Binance USD-M Futures, where moving money between sub-accounts is free.
Template results are real backtests from this engine on the listed coins and dates, after fees, slippage and funding. None is a promise. A template is labeled "Passed out-of-sample" only if it passed every check.
Open this portfolio in the workshop
Portfolios are backtest/paper only for now: live trading isn't available for them yet. You can also open either bot on its own from its template page.
Results not guaranteed. Backtests are hypothetical; trading risks loss of capital.