When I have a new signal, I typically submit it over as many weeks as possible to see how it looks across a long time period in the Numerai Signals diagnostics. I’ve noticed that the “Invalid Tickers” seems to be pretty high and I cannot figure out why that might be.
Here’s an example:
My one thought is that the Numerai ticker universe changes over time, and some tickers may not have existed in the past. Any ideas on how to eliminate “Invalid Tickers” and does this impact my diagnostic scores at all? Thanks.
Oh interesting! I’m currently using bloomberg_ticker for the column header. I’ll re-name that column and submit to see if that improves it. Thank you! (And I’ll update this thread if it does.)
Update: It didn’t seem to fix anything. I’m going to investigate the tickers that I’m using in my submission dataframe and compare it to the tickers in the Numerai universe file and check for discrepancies.
Update: The tickers used in my submission dataframe seem to match the tickers in the Numerai universe file (linked in the comment above). I still can’t figure out why I’m receiving so many “Invalid Tickers” in the diagnostics.
