Reading your results

A backtest looks like a prediction. We tell you whether to believe it.

With every test result we give you an indication of how usable it is in a trading strategy: the headline MinervaScore, and the statistical pass/fail seal, along with the full P&L report behind them. Here is what each one means.

scroll
0
MinervaScore · luck or real? a 0–100 robustness grade
Minerva
robustness
sealSEALED
MarkdownHTML
MinervaScore91✦ Seal · Pass
DSR
PASS
PBO
PASS
SPA
PASS
MinTRL
PASS
Window Stability
0.71
Min TRL prob
0.93
CAGR
+12.9%
Cumulative return
+34.5%
Sharpe
1.84
Sortino
2.61
Calmar
1.79
Max drawdown
−7.2%
Profit factor
2.13
Win rate
58.4%
Expectancy
322
Final equity
268,946
Avg winner
1,040
Avg loser
−686
Trades
214
Winners
125
Losers
89
Equity curve
+ drawdown, monthly returns, distribution, signal quality, trades, disclaimers
01 · The MinervaScore

One number for robustness evidence.

A 0–100, verdict-consistent ranking of how far a validated result sits from five robustness thresholds. The paper demonstrates strong ranking performance in synthetic ground truth.

It’s seal-gated: 80 or above passed every mathematical gate (higher = wider combined margin) · below 80 failed at least one Seal gate. A separate evidence-floor warning describes sample breadth without changing that verdict.

02 · The robustness seal

Five gates decide the Seal.

Where the score grades, the Seal decides. Every gate must clear its threshold. The evidence floor is a separate warning about sample breadth and does not change a passed or failed Seal.

The report names a failing gate and its margin, or explains why the evidence is too thin to judge. Tap a gate for its math.

03 · The full report

Then the full accounting.

Below the score and seal sits the complete P&L report — performance, risk, trades, costs, charts, and the validation evidence.

On an optimization, the headline Sharpe is computed from daily returns on the untouched final holdout — a reserved slice of history the search never touched — with a Newey–West adjustment for serial correlation.

Every field is tappable for what it means and how to read it.

Understand it, rank it, trust it.

Run a sealed optimization →