NO_GO
Latest hourly candle is within expected window.
The strategy has not cleared the research gates
Treat the failed gates as the result: the current evidence does not support progression.
3 of 11 gates passed- FAIL Net return after costs -1.29% · needs > 0
- FAIL Beats the best baseline -1.29% vs best baseline 25.66% · needs strategy > best baseline return
- FAIL Beats baselines risk-adjusted Sharpe -3.52 vs best baseline 4.25 · needs Sharpe > best baseline
Supporting metrics Gross result, costs, equity, win rate, profit factor, and sample size
- Gross P/L
- +$10.84
- Transaction costs
- $23.74
- Equity
- $987.10
- Win rate
- 11.8%
- Profit factor
- 0.53
- Round trips
- 34
Cost-aware equity
Chronological walk-forward holdout · 1393 persisted points
- Start
- $1,000.0024 Jul 2026 · 14:00 UTC
- Current
- $987.1020 Sep 2026 · 13:00 UTC
- Low
- $986.1816 Sep 2026 · 13:00 UTC
- Max drawdown
- 2.18%
Research gates
Actual result versus the condition required to progress
FAIL Net return after costs -1.29%
Required: > 0
The strategy must remain profitable after fees and slippage.
FAIL Beats the best baseline -1.29% vs best baseline 25.66%
Required: strategy > best baseline return
The strategy must outperform the strongest passive return benchmark.
FAIL Beats baselines risk-adjusted Sharpe -3.52 vs best baseline 4.25
Required: Sharpe > best baseline
Return must beat the baseline after accounting for volatility.
FAIL Sharpe interval excludes zero 95% CI [-9.33, 1.12]
Required: bootstrap lower bound > 0
The confidence interval must stay above zero, not merely the point estimate.
FAIL Better than no edge p = 0.1337 over 500 rotations (alpha 0.05)
Required: permutation p < alpha
The observed alignment must be unlikely under the no-edge rotation test.
FAIL Round-trip count 34 (need 100)
Required: meets the configured minimum
A larger sample reduces the chance that a few trades dominate the result.
FAIL Profit factor 0.53
Required: >= 1.20
Gross wins must exceed gross losses by the required margin.
PENDING Forward paper run 1674 settled over 5.9d, -0.66%
Required: profitable over the required window
Only decisions recorded before the next candle count as forward evidence.
PASS Out-of-sample observations 1392 over 9 walk-forward folds
Required: >= 500
Enough unseen hourly observations are needed before the result is credible.
PASS Maximum drawdown 2.18% (limit 10.0%)
Required: within the configured limit
Peak-to-trough loss must remain inside the configured risk budget.
PASS Risk limit respected never tripped
Required: drawdown breaker never tripped
A tripped drawdown breaker means the strategy breached its safety limit.
A passed historical gate does not authorise live trading. Forward evidence and separate legal, security, and execution reviews remain mandatory.
Forward paper run
Decisions recorded before the next candle arrives—the page’s strongest protection against hindsight
Settled decisions
1674 / 300
Elapsed window
5.9 / 30d
Forward net P/L
-$6.58
-0.66% return
Forward drawdown
0.89%
14 entries · 1 open
How forward evidence is protected
Each candle timestamp can be written once. A decision is settled only when the following candle arrives; late gaps are voided rather than filled at a price that was never reachable.
BTC market context
BTC-USD · 1h candles · persisted read-only series
- Start
- $65,494.52
- Current
- $80,496.05
- Low
- $58,243.98
- High
- $81,817.26
Evidence quality
9 walk-forward folds · 1392 unseen bars
The model is refit on an expanding window and scores only bars it has not seen, separated by a 24-bar embargo.
- Permutation p-value
- 0.1337500 rotations
- Annualised Sharpe
- -3.5295% CI [-9.33, 1.12]
- Null mean return
- -1.97%95th percentile -0.73%
- Out-of-sample bars
- 13929 independent folds
Baselines and null-test method
Buy And Hold: +25.66% return · Sharpe 4.25. 100% exposure, no execution cost, not reachable in practice
Always Long: +2.50% return · Sharpe 4.25. 10% of capital held throughout, one round trip
The null rotates the return series while preserving the signal sequence, trade timing, persistence, and volatility clustering.
Backtest round trips
Newest first · latest 20 · holdout replay, not forward evidence
| Entry | Signal | Confidence | Entry price | Notional | Gross P/L | Cost | Net P/L | Equity after |
|---|---|---|---|---|---|---|---|---|
| LONG | 82% | $80,014.12 | $98.62 | +$1.61 | $0.69 | +$0.92 | $987.10 | |
| LONG | 56% | $76,193.42 | $98.72 | -$0.38 | $0.69 | -$1.07 | $986.18 | |
| LONG | 56% | $75,930.77 | $98.80 | -$0.05 | $0.69 | -$0.74 | $987.25 | |
| LONG | 61% | $75,831.87 | $98.89 | -$0.17 | $0.69 | -$0.87 | $987.99 | |
| LONG | 60% | $75,645.58 | $98.98 | -$0.26 | $0.69 | -$0.95 | $988.85 | |
| LONG | 61% | $76,912.76 | $99.25 | -$1.98 | $0.69 | -$2.68 | $989.81 | |
| LONG | 57% | $76,487.74 | $99.34 | -$0.18 | $0.70 | -$0.88 | $992.48 | |
| LONG | 57% | $79,208.67 | $99.43 | -$0.28 | $0.70 | -$0.98 | $993.36 | |
| LONG | 56% | $78,167.61 | $99.45 | +$0.50 | $0.70 | -$0.20 | $994.34 | |
| LONG | 59% | $77,570.56 | $99.53 | -$0.02 | $0.70 | -$0.72 | $994.53 | |
| LONG | 65% | $79,157.09 | $99.83 | -$2.33 | $0.70 | -$3.03 | $995.25 | |
| LONG | 58% | $78,535.16 | $99.86 | +$0.40 | $0.70 | -$0.30 | $998.28 | |
| LONG | 74% | $80,503.99 | $100.01 | -$0.86 | $0.70 | -$1.56 | $998.58 | |
| LONG | 62% | $78,590.31 | $100.04 | +$0.39 | $0.70 | -$0.31 | $1,000.14 | |
| LONG | 55% | $78,669.88 | $100.16 | -$0.49 | $0.70 | -$1.19 | $1,000.45 | |
| LONG | 57% | $78,465.98 | $100.23 | +$0.05 | $0.70 | -$0.66 | $1,001.64 | |
| LONG | 57% | $77,990.42 | $100.29 | +$0.11 | $0.70 | -$0.59 | $1,002.30 | |
| LONG | 56% | $77,740.15 | $100.38 | -$0.16 | $0.70 | -$0.87 | $1,002.89 | |
| LONG | 58% | $77,910.28 | $100.48 | -$0.38 | $0.70 | -$1.08 | $1,003.75 | |
| LONG | 56% | $80,190.32 | $100.53 | +$0.24 | $0.70 | -$0.46 | $1,004.83 |
Economic assumptions
The conservative inputs applied before any result reaches this page
- symbol
- BTC-USD
- interval
- 1h
- fee bps per side
- 25
- slippage bps per side
- 10
- minimum notional
- 10.0
- starting capital
- 1000.0
- position fraction
- 0.1
- enter probability
- 0.55
- max drawdown
- 0.1
- stale after minutes
- 90
- label threshold bps
- 70
- forward run days
- 30
- forward run min decisions
- 300
- walk forward min train bars
- 720
- walk forward step bars
- 168
- embargo bars
- 24
- permutations
- 500
- significance alpha
- 0.05
- min trade count
- 100
- random seed
- 42
- data source
- coingecko
- data quality note
- Synthetic/CoinGecko data is for research plumbing, not execution-quality conclusions.
CoinGecko and synthetic data are research plumbing only, not execution-quality candles.
Freshness and safety
Persisted timestamps and the non-negotiable paper-only boundary
- Last candle
- Last signal
- Last paper trade
- Current state age
- 4m ago
Paper mode only. This dashboard reads persisted simulation state and cannot place live exchange orders.