Building a backtesting engine that never stops searching. 24/7. Next-gen Backtesting Engine - AUTO MODE update: → Claude scans the web: Reddit, forums, trading communities & research → Grok scans X for new trading concepts and setups → Both feed fresh strategy ideas directly into the platform → The... engine automatically converts them into backtests → Profitable ideas get identified → Bad ideas get rejected → Then the search continues. Some strategies work. Most don't. And that's exactly the point. Instead of manually hunting for the next edge, I'm building a system that can search → test → reject → discover → repeat… 24/7. Still building. Still improving. But this is getting seriously interesting. Would you let an AI research engine search for your next trading edge while you sleep?show more

🥹🔥

The next generation of backtesting 💪

You’ll have my fully support about this!! 💪

Thanks Ben! Means a lot. Expect more in the future!

🫡

I'm actually building this too. Different stack and my approach is different than yours but yes... working on it.

Took me a couple of days - and still improving it. Let me know how it goes

how do you make it search 24/7

Spent all last year building mine. Would suggest trying other cheaper models or the costs will be a bottleneck. Had success with Google’s models

Ohh dont mention that :) Cost of the business.

👏

Great work man, congratulations

Thanks buddy. Still working on the engine - its a beast already.

goat

We’ve been building a very similar loop inside TheSystem Next — but the interesting part for us starts after the idea is found. Research → strategy generation → backtest → optimization → walk-forward / out-of-sample validation → robustness filters → reject or promote. The goal isn’t to find as many “profitable backtests” as possible. It’s to automatically kill fragile ideas fast, keep statistically defensible ones, and continuously repeat the process across markets. 24/7 discovery is powerful. 24/7 falsification is probably even more important. @thesystemnext

Beautiful setup! What platform do you do the backtests on?

Thanks. OHLCV file from databento

can yu share more detials

SUre- will post a more videos

Any further updates

Of witch feature buddy?

I am looking for a profitable strategy which i can execute, do you have practice option selling or buying strategy which is profitable

You mean an Enigma on a silver plate? I am reading this right?

Internal autosearch > AI autosearch 1000x AI autosearch > manual 1000x Internal cannot be public, AI can

Both are in place. Plus auto improve

Genuinely interesting build — the search→test→reject→repeat loop is the right shape. A few questions from someone building CFD strategies for ProRealTime: On data: What are you backtesting against? Which instruments, and where do you source the historical data — your broker's feed, a provider (Dukascopy, Polygon, etc.), or something else? And are you testing on cash/CFD prices or futures? It matters a lot whether the "profitable" ideas were found on the same price series you'd actually trade. On timeframes: How does the engine decide which timeframe to test a given idea on? Is it fixed, swept across several, or does the idea itself imply one? A concept can look dead on 15M and alive on 1H purely by chance, so I'm curious whether TF is part of the search space — and if so, how you stop that from just multiplying the false positives.

where do you get historical data for such backtesting?

databento

While I love the general unblocking potential of AI, I still much prefer deterministic systems. Might be the old school dev in me though (can't believe I'm saying that at my age)... absolutely love it though that we're reaching a point where we can build at the speed of thought!

Thanks Tiago. Means a lot when coming from an old school dev.

Really nice, keep posting about it

Sure- still improving it. Thanks

@RHerman you are doing great👍👏

Thanks buddy. Its just the beginning...

💪🏼💪🏼

Spent a few months building my trade ideas sandbox. It has 20 years of price history and a rigorous testing methodology. I’ve fed it countless ideas. Some do great in the sample years and collapse from regime change. A few survive and get certified for paper trading. When…

Thats what its all about - those just few but special ones

can u just hard code and teach the LLM your own edge with context , rather than dicovering a news one