For individual traders

Your backtest passed.
Will your strategy survive live?

Upload your backtest. Get a 7-layer validation in under 5 minutes. Find out if your edge is real before you risk capital.

The problem

Why backtests lie

316published factors tested by Harvey, Liu, and Zhu. Most didn't replicate.

You spent weeks optimizing. The backtest showed 40% annual returns. You deployed it. Month one: down 8%.

That's not bad luck. It's overfitting. The strategy captured noise in the training data and mistook it for signal. A 2.5 Sharpe in backtesting becomes 0.4 once you account for spreads, slippage, market impact, and regime shifts.

Walk-forward analysis catches about 30% of overfit strategies. The other 70% look fine until they bleed money in live trading. The gap between paper performance and live performance is where real capital disappears.

Process

How it works

Upload, validate, decide. The full process takes under six minutes.

1
30 seconds

Upload your strategy

Pine Script, Python, MQL5, or CSV backtest results. Drag and drop. The system auto-detects your format and column structure. No configuration, no API keys.

2
Under 5 minutes

Seven layers validate

Each layer runs independently. Overfitting detection finishes first. Then causal inference, stress scenarios, regime analysis, execution costs, and integrity checks. You watch results arrive as each layer completes.

3
Instant

Get a clear verdict

Validated, Conditional, Weak, or Rejected. No ambiguity. The composite score (0 to 100) and per-layer breakdown tell you exactly where your strategy is strong and where it's vulnerable. Deploy with data, not hope.

Detection layers

What we catch

Four of seven layers, focused on the failure modes that cost individual traders the most.

L1

Statistical Core: Is the edge real?

Cross-validates your strategy across multiple time periods and parameter variations. If performance only exists in one specific slice of history, this layer flags it.

CPCV (Combinatorially Purged Cross-Validation) tests every possible combination of training and testing periods while removing contaminated data points. It catches over 80% of overfitting that standard walk-forward analysis misses.

80%+overfitting detection rate
CPCV FOLD MATRIXTRAINPURGED TESTF1F2F3F4F5F6F7F8
L3

Scenario Lab: What hasn't happened yet?

Your backtest only covers historical conditions. This layer generates thousands of synthetic scenarios your strategy has never seen: fat-tail events, liquidity shocks, correlation breakdowns.

The chart shows the core problem. A strategy that looks great in backtesting (the ascending curve) can follow two very different paths after deployment. Validated strategies continue to compound. Overfit strategies give back everything. This layer reveals which path yours is more likely to follow.

1000+synthetic scenarios tested
DEPLOYBACKTESTVALIDATEDOVERFIT+40%0%-30%
L4

Regime Intelligence: Will it survive the next shift?

Detects which market regime your strategy was optimized for and tests whether it degrades when conditions change. A strategy built for 2017's low-volatility calm can fall apart when the VIX spikes.

The framework identifies three distinct regime types using Hidden Markov Models on market data. Your strategy is tested against each regime independently. If it only works in one, you'll know before the market shifts.

3regime types: bull, bear, transition
BULLBEARTRANSITIONVIX
L5

Execution Realism: What are the real costs?

Models spreads, market impact, slippage, and fees for your specific strategy. The backtest assumes zero friction. This layer shows what +15.2% annual returns look like after the market takes its cut. Often: +3.1%. Sometimes: negative.

4cost components modeled

These are 4 of 7 layers. The full platform includes causal inference, live monitoring, and cryptographic integrity. See all seven layers

Crisis validation

Proven against real crises

2008Global Financial Crisis45.7-point separation
2018Volmageddon27.1-point separation
2020COVID-190.706 forward correlation

We ran 61 strategies through the pipeline against three decades of market crises. The framework correctly separated genuine strategies from overfit ones, and its rankings predicted which strategies would actually perform in the years after each crisis. Read the full case studies

Comparison

Compare your options

Sigmentic
Manual ReviewNo Validation
Time to resultUnder 5 minutes40+ hours per strategyN/A
Overfitting detectionCPCV + PBO (catches 80%+)Walk-forward only (~30%)None
Regime testingHMM regime detection + stressRarely doneNone
Execution costsFull cost modelEstimated or ignoredAssumed zero
OutputPortable Validation PassportInternal notesNothing

Get started

Start validating. Free.

3 validations per month. No credit card required.

Under 5 min to verdictZero data retentionPine Script, Python, MQL5, CSV