EA EdgeAudit ← Back to home
Sample outputs

Three real backtests.

Below are three example outputs straight from the bot — one PASS, one PROMISING, one REJECT — so you can see exactly what the tool produces before installing it.

Example 01 Pass

Momentum signal on QQQ — survives both checks.

EA
EdgeAudit · just now
/backtest custom company: QQQ strategy: "12-month momentum, top quartile rebalance monthly" years: 5
Backtest · QQQ
12-month momentum signal, monthly rebalance into the top-quartile percentile
Performance
Annualised return: +12.3%
Sharpe ratio: 0.94
Max drawdown: -28.4%
vs buy-and-hold: +0.6 pp/yr
Statistics
Round-trips: 62
Win rate: 58.1%
95% CI: [+4.1%, +20.5%]
Bonferroni 95% CI: [+1.8%, +22.8%]
Verdict: PASS · Bonferroni LB > 0, robust to multiple-comparison adjustment

Why this passes. The momentum signal produced 62 round-trips over the test period, enough statistical power to drive a meaningful confidence interval. Both the standard 95% CI and the Bonferroni-corrected 95% CI exclude zero — meaning the lower bound of plausible returns is still positive even after adjusting for the multiple parameter variants tested. The strategy also marginally outperforms buy-and-hold rather than just tracking it. This is the strongest verdict EdgeAudit issues; everything above this is just degrees of confidence.

Example 02 Promising

EMA(12/26) crossover on MSFT — survives the lighter check, not the harder one.

EA
EdgeAudit · just now
/backtest ema company: MSFT fast: 12 slow: 26 years: 5
Backtest · Microsoft (MSFT)
EMA(12) above EMA(26) → long; cross back → flat. Long-only, the classic MACD-style momentum signal.
Performance
Annualised return: +8.4%
Sharpe ratio: 0.62
Max drawdown: -19.7%
vs buy-and-hold: -6.1 pp/yr
Statistics
Round-trips: 34
Win rate: 55.9%
95% CI: [+0.6%, +16.3%]
Bonferroni 95% CI: [-2.4%, +19.2%]
Verdict: PROMISING · 95% LB > 0; Bonferroni LB ≤ 0

Why this is only "promising". The strategy looks profitable in isolation — the 95% confidence interval just excludes zero (lower bound +0.6%), so by the conventional retail standard you'd call it a winner. But once we widen the threshold to account for the multiple parameter variants typically explored before arriving at the chosen 12/26 setting, the Bonferroni-corrected lower bound drops below zero (-2.4%). That means we can't statistically rule out that the apparent edge is multiple-comparison noise. It also underperforms buy-and-hold by 6.1 percentage points per year. "Promising" is the verdict you should expect for most plausible-but-unconvincing strategies — interesting enough to research further, not safe enough to deploy.

Example 03 Reject

RSI(14) mean-reversion on AAPL — neither check survives.

EA
EdgeAudit · just now
/backtest rsi company: AAPL period: 14 oversold: 30 overbought: 70 years: 5
Backtest · AAPL
Classic RSI mean-reversion. Long-only, fully out of market when no signal.
Performance
Annualised return: +3.2%
Sharpe ratio: 0.21
Max drawdown: -22.1%
vs buy-and-hold: -7.4 pp/yr
Statistics
Round-trips: 87
Win rate: 51.7%
95% CI: [-8.5%, +14.9%]
Bonferroni 95% CI: [-11.2%, +17.6%]
Verdict: REJECT · Both 95% and Bonferroni intervals cross zero

Why this is rejected. The headline annualised return is positive but small, and the confidence interval is wide. Even the more permissive 95% threshold can't rule out that the strategy makes money by luck rather than skill — the lower bound is firmly negative. The Bonferroni-adjusted threshold is wider still. The strategy also underperforms buy-and-hold by over 7 percentage points per year. Without statistical discipline, you might look at "+3.2% annualised over 87 trades" and call this a working strategy. EdgeAudit calls it noise.

Try your own strategy.

The guided builder walks you through picking a company and a strategy template — no syntax to memorise. Free tier, no card needed.