QuantProof Does Your Backtest Survive - Catch overfitted backtests before real money does
by•
My Sharpe was 2.4. Win rate: 68%. Looked perfect until I traded it live. My entire edge was 3 lucky trades in 2020.
So I built QuantProof. Upload your backtest CSV and find out if your edge is real or just luck.
38 checks in 2 seconds: overfitting detection, profit concentration, slippage sensitivity, alpha decay, regime testing, and 5 historical crash simulations.
Connect Alpaca for broker-verified results.
Free to validate. ₹999 for the full PDF + AI improvement plan
Replies
Hey PH 👋, Hardik here. I built Quantproof after a backtest that looked great blew up in real life, and I lost capital, and my dreams of a prop firm shattered. I have since improved my strategies, running them by my system and improving the feedback it provided.
A few things I'd love brutal feedback on:
Is the 0–100 score intuitive or confusing?
What checks are we missing?
Would you pay ₹999 (~$12) for the PDF, or is that mispriced?
Happy to answer anything about the methodology. We run deflated Sharpe, CPCV path stability, alpha decay, bootstrap resampling, and 5 crash sims. Ask me anything. We are working on international payments; they should be sorted in 48 hours
One week in and we have made upgrades. Thanks for evreryone whoupvoted.
Since some people have asked what has been happening since launch—here is the honest update.
What we found in validations this week:
The most common finding across strategies from India, the US, and the UK: Sharpe ratios above 2.0 that still fail the Deflated Sharpe test because of parameter overfitting. The strategy looks real. The edge is noise. QuantProof catches it before a $400 prop-firm challenge fee does.
Why does QuantProof actually run when the launch post undersold it?
44 institutional checks plus 5 historical crash simulations, including 2008 GFC, 2020 COVID, 2022 Bear Market, 2010 Flash Crash, and 1998 LTCM.
The check stack includes things most retail validators do not touch:
1. Deflated Sharpe Ratio with Newey-West HAC correction for serial correlation
Ljung-Box autocorrelation test at lags 1, 5, and 10
Purged walk-forward cross-validation with embargo periods
4CVaR / Expected Shortfall at 95% and 99%
5The information ratio vs. SPY benchmark detects beta-capture disguised as alpha
6Beta Exposure check flags strategies that are just leveraged index exposure
7. Fractional Kelly criterion and generalised ruin probability at institutional position sizing
8 Almgren-Chriss market impact capacity model
9HMM regime detection across Bull, Bear, and Consolidation periods
10. Regime auto-detection via SPY 20-day SMA-sounded CSVs still get regime context
11AR(1) autocorrelation penalty in crash exposure modelling
Single-regime backtest warning when more than 85% of trades are in one market condition
Who can connect:
1 CSV upload for any strategy with a PnL column
2 Alpaca paper and live trading direct API verification with position-state tracker
3. Zerodha Kite TradeBook and Orders CSV both supported
4IBKR FlexQuery integration for institutional verification
TradingView webhook session for live alert-based strategies
6MT5 trade history export
7 Options CSV with convexity warnings
The API is live at v2.7.0 for anyone who wants to build on top of the validation engine.
The price is ₹299 for the full PDF report. Deliberately accessible. Institutional methodology should not cost $25,000 a year.
If you have a backtest and want to know whether it will survive a prop firm challenge statistically, drop a comment or DM me, and I will run it for free today. I am still checking every validation personally at this stage.
Transparency is a big for us, so we added a methodology page, How QuantProof Works — Validation Methodology