Today morning I ran my strategy (backtest) looking for new stocks to buy. The date range used (Most recent years -> 22) resulted in totally different results than the results shown in a backtest yesterday evening! Also, the backtest took tremendously long, much longer than the days ago! How come?
Rename
It sounds like you reconfigured your data providers in some way.
Other possible factors:
1. usage of 'unstable' indicators
2. peeking into the future error
1. usage of 'unstable' indicators
2. peeking into the future error
I did not change the data provider or the order of the providers. I just installed WL8, downloaded the extension for WL8 and migrated the data. Then made a backtest and the results were totally different. To check that I backtested my strategy in WL7 and suddenly the results were bad also in WL7.
QUOTE:
Also, the backtest took tremendously long, much longer than the days ago! How come?
1. Superfluous Event providers and their data collection.
2. First time data backfill.
QUOTE:
To check that I backtested my strategy in WL7 and suddenly the results were bad also in WL7.
See "other possible factors" above.
In a case we investigated last year we indeed found an unstable indicator in @Braunstein's strategy but my interim conclusion pointed to a change in the data. However there was no response or resolution:
https://www.wealth-lab.com/Discussion/Signals-list-changed-compared-to-yesterday-6934
https://www.wealth-lab.com/Discussion/Signals-list-changed-compared-to-yesterday-6934
Your Response
Post
Edit Post
Login is required