← sources

Algos that are consistently successful when back testing but dont pan out in real trading?

source: reddit (reddit:algotrading)published 2026-11-15 17:45ingested 2026-11-16 02:39

retained text

Does/Has this happened to you?  If so, why do you think that is?  A friend of mine created one that when working with an A.I. produced incredible results, but then the market changed, and so did the strategies.  Trying to understand if this is a common phenomenon, and why.  Maybe various algos are required for different regimes, and we are supposed to know what regime we will get on any given day before the day begins.

Ive been working on one that consistently produces a 70%+ win rate, but it does invalidate signals which isnt obvious when backtesting.  Its the first time Ive experienced this due to using 5minute qualifiers on a 1minute setup, with a 2minute base code 😅

Thank you for any feedback.  I guess for the owners of the codes that end up working, they quietly disappear for obvious reasons.

document fetched 2026-11-16 00:38 · canonical https://reddit.com/r/algotrading/comments/1xx1qk2/algos_that_are_consistently_successful_when_back