I will backtest and stress test your trading strategy in python, no overfitting


About this gig
Most backtests are too good to be true. I find out whether yours is real.
I turn your rules into exact code and test them like a skeptical quant. Answer: real edge, fragile, or luck.
What I check
- Signal on bar close, fill next open; no look-ahead
- Repainting audit from past data only
- Commission + slippage, and a 2x slippage stress
- IS/OOS: fit in-sample, judge on unseen data
- Parameter neighborhood vs a lucky spike
- Monte Carlo / random-entry baseline
- Multiple-testing luck control
- Futures rollover and bar building
Method only, no promises: Donchian 20/20, 0 look-ahead mismatches, rollover false signals; ADX ~157 variants; random-entry 3,000 paths, 0 profitable; 26,880 configs, 4 passed vs ~4.3 by luck, no edge.
I also code THS/TDX indicators (DRAWTEXT, BARSLAST, ADX/DI, Donchian).
You get a report (P&L, PF, win rate, drawdown, IS vs OOS), equity curve, trade CSV, and Python on Standard and Premium.
Message me rules, market, and timeframe before ordering.
This is a coding and research service, not financial advice. I don't sell signals or "profitable strategies", and I never guarantee returns. Past results do not predict future performance.
Get to know Jun Zhang
AI Avatar and HeyGen Product Videos
- FromChina
- Member sinceMar 2026
- Avg. response time14 hours
Languages
English, Chinese
Other Software Development Services I Offer
FAQ
Will you make my strategy profitable?
No one honestly can promise that. I code your rules precisely and test them rigorously. If the edge is weak or luck, I'll tell you.
Do you sell ready-made strategies or signals?
No. I only work on rules you provide, and I don't give buy/sell recommendations.
Which markets?
Futures (incl. Chinese commodity futures with rollover), crypto, stocks, FX, if OHLCV available.
Can you code TradingView Pine / MT5?
Core is Python backtesting plus Chinese formula languages (THS/TDX). Message me first for other platforms.
What if the result is negative?
Still a valid deliverable. Revisions fix errors or misunderstood rules, not make the numbers better.

