jjvors
Cavs fan since 1970. Browns Indians/Guardians 1964
- Joined
- Jan 18, 2022
- Messages
- 6,103
- Reaction score
- 6,633
- Points
- 113
A statistically accurate statement! It's rare to see on a sports forum. I use 32 as a cut-off for minimally statistically accurate, but I prefer 40-50 as a minimum sample size. You want your sample to be as large as possible.There's no strict cutoff between "statistically significant" sample, and "not statistically significant sample" -- actually that doesn't really even mean anything unless you say what it is statistically significant compared to....
In his last 13 games he is 15/31 = 0.4838 , with a range (1-sigma) of 0.38 -- 0.585
In his first 32 games he is 13/58 = 0.224, with a range (1-sigma) of 0.167 -- 0.292
The trouble with small samples is that they're not random. If you play ten games against the 10 worst defenses you'll get different results than 10 games against the best defenses. I'd want a sample against all the teams in the league, say, half a season.