Intraday signal ensemble · All signals
The same method as the selected version on every signal family, half size per slot.
2022-09-27 – 2026-10-05
Backtest · Platform backtestBacktest period returns
Backtest cutoff 2026-10-05 · Fixed backtest results
| Window | Cumulative return | CAGR | Max drawdown | Closed trades | Trade win rate | Profit factor | Monthly win rate | Average holding time | Dates / days |
|---|---|---|---|---|---|---|---|---|---|
| Full period | +1256.11% | +91.12% | 31.23% | 1,096 | 52.01%n = 1,096 | 1.68 | 60.42%48 complete months · Flat 2 months | 1.86 days | 2022-09-27 → 2026-10-051,469 days |
| Start through 2022 | +1.44% | +5.60%Short-period extrapolation | 1.68% | 9 | 55.56%n = 9 | 4.76 | 33.33%3 complete months · Flat 2 months | 1.42 days | 2022-09-27 → 2022-12-3196 days |
| 2023–2025 | +759.92% | +104.74% | 31.23% | 895 | 52.40%n = 895 | 1.72 | 63.89%36 complete months | 1.88 days | 2023-01-01 → 2025-12-311,096 days |
| 2026 through cutoff | +55.46% | +78.78%Short-period extrapolation | 30.90% | 192 | 50.00%n = 192 | 1.62 | 55.56%9 complete months | 1.78 days | 2026-01-01 → 2026-10-05277 days |
Counts use trades closed in this window; win rate uses profit after fees and funding. Average holding time includes closed trades only.
Trade and monthly statistics definitions
One trade = one of the strategy's positions from entry to exit; every step of a staged entry and every slot open at the same time on one instrument is a trade of its own. Only closed trades count. Win rate = the share of trades with a net profit (after fees, slippage and funding) above 0. Profit factor = the same trades' total gains / total losses; above 1 means more was won than lost. Long trend strategies often win big on a few trades and lose small on many: a low win rate with a high profit factor. Without any losing trade there is no profit factor. Monthly win rate = the share of calendar months with a positive return on the equity curve, complete months only (the partial month where the backtest starts and the one where it ends do not count); it does not depend on the trades, and months with an open position count too; a month spent wholly out of the market (a return of exactly 0) is counted, not as a win.
Returns since activation · simulated funds
Current run starts: 2026-11-12
2026-11-12 → 2026-11-21 · UTC · Under 30 days: CAGR is not displayed
This run uses simulated funds; fills and liquidity can differ from real trading.
Previous-close allocation
2026-11-21 · UTC
No open exposure
How the strategy participates
BTC / ETH perpetuals; only the top model scores trade, at most 3 slots per instrument.
Compare returns from different starting points.
The 2023-onward curve is a slice of the same backtest, carrying over its exposure and risk state. Each curve is rebased to 100 at its own starting point.
Full-period backtest Platform backtest
2022-09-27 – 2026-10-05Same-period buy-and-hold benchmark · CAGR 45.1% / MDD 53.9%
2026-10-05 · Equity 1356.11
The chart uses display samples. Returns and drawdowns come from the same-period source summary, not a recalculation from sparse points.
Costs, assumptions & engine differences
Framework backtest: the same process and risk controls as live trading, taker fees, slippage estimated from the liquidity at the time, perpetual funding as paid, liquidations simulated on 1-hour mark prices. The curve starts at 100; whole period.
Backtest from 2023 Platform backtest
2023-01-01 – 2026-10-05Same-period buy-and-hold benchmark · CAGR 55.0% / MDD 53.9%
2026-10-05 · Equity 1336.81
The chart uses display samples. Returns and drawdowns come from the same-period source summary, not a recalculation from sparse points.
Costs, assumptions & engine differences
Framework backtest: the same process and risk controls as live trading, taker fees, slippage estimated from the liquidity at the time, perpetual funding as paid, liquidations simulated on 1-hour mark prices. The curve starts at 100; from 2023.
Read these limits before choosing the return.
- A backtest is a simulation of the past, not a promise of future returns.
- The backtest runs the same process and risk controls as live trading, with fees, slippage and funding; real fills can still differ.
- Returns since activation come from the exchange's paper account (fake funds); its fills can differ from the real market and it has run for a short time.
- This page shows test results and the previous close's position shares only: no live positions, no entry or exit signals, no investment advice.