Hey @everyone, I wanted to give you a quick sneak peek at what I’ve been working on lately.
Over the past few weeks, I’ve been building, improving, and debugging my own locally running Python-based strategy script optimizer. This means we can finally move past TradingView’s optimizer limitations.
On TradingView, optimization is slow and limited to one backtest range one asset and timeframe and strategy combination at a time. With this new optimizer, I can run iterations 10x, 100x, or even 1000x faster and on multiple asset-timeframe-strategy combinations in parallel. (I’ve already seen 1-hour timeframe optimizations running at 900–1500 iterations per minute!)
The optimizer also uses much more advanced optimization methods, including:
Walk-forward optimization with multiple train/test windows
Holdout checks on unseen data
Neighboring parameter robustness checks
Various additional robustness checks to help avoid overfitting
Advanced ranking and sorting based on factors not included in a traditional TradingView backtest report, such as Calmar, top 5/10/25% performance, and many more
Beggining of this week, I finalized the first version of the optimizer and over the past few days, I’ve been running tests and optimizations across various assets and timeframes, including:
USDJPY, GBPJPY, EURUSD, XAUUSD, XAGUSD, US100, US500, BTCUSDT, ETHUSDT, and SOLUSDT
on 5m, 15m, 30m, and 1-hour timeframes.
After around 5 million iterations, I’ve started going through the results and have already found a few great ones (check out the attached screenshots)
(optimization date range with many train and test walk forward windows: 2021.11.01-2025.11.01.)
The optimizer has currently been running on three of the newly updated V2 strategies, using only a limited set of filters for now — around 8 out of 30.
I’ll be uploading these optimizations as presets to the V2 scripts in the coming days, so stay tuned!
