I'm re-implementing this algorithm in research: https://www.quantopian.com/posts/long-only-non-day-trading-algorithm-for-live#5ebe5112ea971b7d207ce8f0.
What is super interesting is how positive the backtest was until a year ago. It's really fascinating seeing how some models completely break during extreme uncertainty, even though in the original version the beta was low.
What I wanted to ask: in the lectures it was said that any p-value under 0.05 is good. If it's over 0.05 just ignore everything else and keep on digging for another factor. However what does it mean when the p-value is like below where 1day is bad but otherwise it's OK:
p-value(IC) 0.115 0.003 0.007
Are there exceptions from the rule that any p-value under 0.05 should be ignored?