2,147 closed trades · 50.3% win rate · paper trading
Performance
Real win rate, profit factor, Sharpe, and drawdown per scalp signal type, computed from the actual resolved-signal log.
Why this page exists
Every signal type currently under test is published here, whether or not it actually works — nothing removed, nothing rounded up. Win rate alone doesn't determine profitability; profit factor and average win against average loss are shown alongside for exactly that reason.
- Signals tracking now
- 17
- Closed trades
- 2,147
- Hit rate
- 50.3%
- Average result
- −$8.67
open signals with a live stop and target
every paper trade resolved and recorded, win or lose — all time, from the trade ledger, not the signal ledger
over 2,147 closed trades
mean profit per closed trade, in dollars
Two populations below: journal trades and tracked signals. Never blend them.
Two different real populations below, never blend them: journal = trades this bot's auto-trader actually executed through a broker (small, real money); signals/backtest = every signal this bot posted and tracked to a real resolved outcome, whether or not it was ever traded (much larger population, same real price data — not a simulation or a re-run against history with different parameters, see PROJECT_DOCUMENTATION.md's backtest_engine.py notes). A stock/crypto win rate from one population can differ from the other without either being wrong — they're not the same sample.
- closed trades (journal)
- 2,147
- win rate (journal)
- 50.3%n=2147
- stock win rate (journal)
- 22.7%n=132
- crypto win rate (journal)
- 52.2%n=2015
- stock net P/L (journal)
- −$22,866−$173 per position
- crypto net P/L (journal)
- +$4,257+$2 per position
- entry slippage (journal, stock)
- −$5,502 0.18% avg on n=130
- share of the stock loss
- 26%execution, not thesis
- scalp signals (backtest, 90d)
- 1289
- swing signals (backtest, 90d)
- 1080
- stock max loss streak (backtest, 90d)
- 31
- crypto max loss streak (backtest, 90d)
- 28
By market regime, at entry
Every closed position, grouped by the market regime it was opened in. Regime is the broad market's state at entry, read from SPY's trend structure and applied to crypto positions too.
| Regime at entry | Asset | Positions | Net P/L | Per position | Win rate |
|---|---|---|---|---|---|
| Trending | crypto | 1,511 | +$5,684 | +$4 | 51.9% |
| Range Bound | crypto | 492 | −$266 | −$1 | 53.3% |
| Trending | stock | 99 | −$12,185 | −$123 | 28.3% |
| Range Bound | stock | 33 | −$10,681 | −$324 | 6.1% |
| Unknown | crypto | 12 | −$1,160 | −$97 | 41.7% |
How the regime is read
The regime is read from SPY's own trend structure and is applied to every position opened in that window, including crypto — which is a framing, not a claim about the coin. Rows are ordered by population, because a bucket of eight and a bucket of seven hundred are not equally informative. Regime snapshots are taken periodically, so each position takes the most recent reading at or before its entry; anything opened before the first snapshot is counted as unknown rather than assigned to the nearest one.
Win Rate by Signal Type
Every type that has resolved at least one trade, swing and scalp together. The dashed line marks 30%, a visual anchor only.
How to read these bars
Every real signal type, swing and scalp together — green marks types clearing a 30% win rate, not a claim of profitability by itself, just a visual anchor since profit factor/avg win vs loss (tables below) is what actually determines that. Each n counts the resolved signals that carry a recorded percentage result; one resolved without it is left out rather than counted as 0%, so these rows can total fewer than the resolved count shown elsewhere on the site.
| Experimental (shadow) | Resolved | Win Rate |
|---|---|---|
| PULLBACK | 2 /14 open | 0.0% (n=2, under 30) |
| CONTINUATION | 0 /2 open | no resolved trades yet |
| C_MOM_TREND | 0 | no resolved trades yet |
| MOM_TREND_ONLY | 1 | 0.0% (n=1, under 30) |
| C_VWAP_RS | 2 | 0.0% (n=2, under 30) |
| C_CVD_OK | 8 /20 open | 25.0% (n=8, under 30) |
| MOM swing | 0 | no resolved trades yet |
| ORB swing | 0 | no resolved trades yet |
| VWAP swing | 0 | no resolved trades yet |
| CANDLE swing | 0 | no resolved trades yet |
| GAP swing | 0 | no resolved trades yet |
| HOD_BREAK swing | 0 | no resolved trades yet |
| PM_HIGH swing | 0 | no resolved trades yet |
| EMA_PULL swing | 0 | no resolved trades yet |
| FIB_BOUNCE swing | 0 | no resolved trades yet |
| BULL_FLAG swing | 0 | no resolved trades yet |
Experimental setups running as a live shadow: never posted, never traded, and not counted in the chart above or any total on this page. Pullback and Continuation: logged and resolved on the same daily-bar rules they were backtested under. Out of sample, neither beat a random entry in the same uptrend. Crypto combination rules (C_MOM_TREND, MOM_TREND_ONLY, C_VWAP_RS, C_CVD_OK): every crypto scalp signal is recorded with whether each rule would have kept it, then followed under the live exit ladder on Binance 5-minute closes, costs included. A signal with no order-flow read in the six hours before it is recorded as no data for C_CVD_OK, never as a pass. In the study none of them was strong enough to trade. Swing variants (the _SWING rows): every stock scalp is also recorded as a swing on the same trigger, stop 1.5 x daily ATR, target 2R, held up to 10 sessions, and resolved on daily bars. In the replay of July to November every one of them lost, and lost more than its scalp. Fewer than 30 resolved trades is too few to judge.
Scalp — By Signal Type (90d)
Profit factor total gained on winning trades ÷ total lost on losing ones; above 1.00, winners outweighed losers. · Sharpe average result per trade ÷ how much results varied, per trade, not annualised. · Max DD (drawdown) the largest peak-to-trough fall when the trades are compounded in order.
| Type | Resolved | Win Rate | Profit Factor | Sharpe | Max DD | Avg Win | Avg Loss |
|---|---|---|---|---|---|---|---|
| VWAP | 158 /160 logged | 45.6% | 0.920.68 net | -0.03 | 15.7% | +1.03% | -0.94% |
| GAP | 84 | 44.0% | 1.241.08 net | 0.10 | 20.5% | +2.67% | -1.69% |
| MOM | 359 /360 logged | 41.8% | 1.020.78 net | 0.01 | 27.3% | +1.33% | -0.94% |
| ORB | 313 | 40.3% | 0.770.59 net | -0.10 | 47.3% | +1.21% | -1.06% |
| PM_HIGH | 91 | 37.4% | 0.660.55 net | -0.18 | 28.9% | +1.69% | -1.56% |
| HOD_BREAK | 100 | 32.0% | 0.790.66 net | -0.10 | 27.0% | +2.24% | -1.33% |
| EMA_PULL | 47 | 31.9% | 0.730.62 net | -0.14 | 17.0% | +2.21% | -1.41% |
| CANDLE | 125 | 22.4% | 0.500.43 net | -0.32 | 49.8% | +2.40% | -1.37% |
| BULL_FLAG | 10 | — | — | — | — | — | — |
| FIB_BOUNCE | 2 | — | — | — | — | — | — |
Sample-size warning applies below 30 resolved trades — treat those numbers as directional, not conclusive.
How “net” and “resolved” are counted
“net” beside a profit factor is the same figure after an assumed 15.0bps round-trip cost (spread plus price impact), which is the bar this system holds itself to: `validate_strategy` accepts or rejects a rule on the net number, not the gross one. It is an ASSUMPTION rather than a measurement — distinct from the entry slippage reported above, which is measured on real stock fills — and it is shown so a reader sees the same figure the engine judges by. “Resolved” only counts entries with a real recorded exit price — the “/N logged” figure shows how many were logged in total, including any still open or never filled.
Swing — By Signal Type (90d)
| Type | Resolved | Win Rate | Profit Factor | Sharpe | Max DD | Avg Win | Avg Loss |
|---|---|---|---|---|---|---|---|
| SETUP | 79 | 34.2% | 0.250.21 net | -0.47 | 64.6% | +1.21% | -2.55% |
| MEAN_REVERSION | 1001 | 26.6% | 0.720.64 net | -0.09 | 98.5% | +3.45% | -1.73% |
Sample-size warning applies below 30 resolved trades — treat those numbers as directional, not conclusive.
How “net” is counted, and why some rows show “—”
“net” beside a profit factor is the same figure after an assumed 15.0bps round-trip cost (spread plus price impact), which is the bar this system holds itself to: `validate_strategy` accepts or rejects a rule on the net number, not the gross one. It is an ASSUMPTION rather than a measurement — distinct from the entry slippage reported above, which is measured on real stock fills — and it is shown so a reader sees the same figure the engine judges by. Rows under 5 show their raw count instead of a colored pill since a single trade can otherwise read as a false 0%/100%. Real swing volume is still thin compared to scalp, so profit factor/Sharpe/drawdown will often show “—” until more trades resolve.
Crypto Backtest
- resolved (90d)
- 5679
- win rate (90d)
- 44.5%
- profit factor (90d)
- 2.25
| Type | Resolved | Win Rate |
|---|---|---|
| ALT_RS | 1169 | 52.0% |
| BREAKOUT | 1166 | 45.9% |
| MOMENTUM | 1420 | 43.8% |
| VWAP | 1924 | 39.6% |
Sample-size warning applies below 30 resolved trades — treat those numbers as directional, not conclusive. Rows under 5 show their raw count instead of a colored pill since a single trade can otherwise read as a false 0%/100%.
Known Issues
The min_rr gate checked R:R against a signal's planned entry, so a bracket order filling away from the signal price could realize a materially worse ratio than the gate cleared — confirmed live: GSAT opened at 0.49:1 despite clearing a 1.5 minimum. R:R is now re-tested at submission against a live quote, and a signal whose edge has drifted below the minimum is skipped rather than entered. The ratio each position actually got is also recorded from its real fill, so any residual slippage is measurable instead of only findable by hand.
13 of 43 closed positions (all BTC) exited on a 2-hour hold-time timeout rather than a stop or target. Root cause: crypto signals were labelled “scalp” by default, but their stops and targets are sized from daily ATR — a target sits about two daily ranges away, which two hours cannot reach. Only positions that hit their stop quickly ever resolved, and BTC, the least volatile name in the universe, mostly didn't — so the timer collected BTC. The hold limit is now derived from each position's own geometry, so a multi-day target is given a multi-day horizon instead of being closed at 120 minutes, with a longer backstop still preventing a slot being held indefinitely.
Scale-outs (a partial profit-take followed by the final close of the same position) were counted as two separate closed trades, inflating both trade count and win rate. Now grouped by position — a partial fill shows as a badge on its position's single row, not a second row.
Signal-Level Factor Research and Feature Redundancy on Research Lab were silently checking field names that don't exist in the real data (a stale key list left over from an earlier factor-consolidation change) — every real correlation/bucket read was against the wrong keys. Restored to the real shape and verified against live production data.
Publishing the defects next to the numbers is the point — a performance page that never changes its own methodology isn't being audited.