I swear there was a post on back testing potential systems
Something about how much is too much backtesting
or
how much is enough etc......

Which leads to the following question
How much back testing is needed before running a system?
5 nba seasons using all teams?
2 mlb season using national league only, american league only, or both?

Is there a comfortable number of games regardless of what sport?
100 games of data?
500 games?
1000 games?
10000 games?

looking into an interesting under over system and in the early process of back testing
any help would be great