Trend vs. regression

silentbob · #1 01-03-2006, 07:03 PM

I struggle with this issue when betting any sport -- when there's a stat indicating that "team X is 8-2 in their last 10 games ATS," couldn't this stat be used to justify betting either way? I mean, it indicates X is more than likely to cover, but at the same time it's also likely that X's ability to cover isn't really 80% and will probably regress towards a more reasonable figure.

How do y'all decide what to do with a "trend" statistic? I'm a relative noob at sports betting and don't really know how to react sometimes.

eternalnewbie · #2 01-03-2006, 09:30 PM

Not a statistics expert, but I'm pretty sure that something saying that a team is 8-2 against the spread, statistically speaking, has absolutely no predictive ability one way or the other. Need an N larger than 10 I think.

In other words, I'm pretty sure that saying a team is 8-2 against the spread is about as useful as saying they have blue uniforms for purposes of predicting their future performance.

A lot of the posters here understand probabilities much better than me, though, so maybe they'll correct me if I'm wrong.

NoChance · #3 01-03-2006, 10:27 PM

[ QUOTE ]
In other words, I'm pretty sure that saying a team is 8-2 against the spread is about as useful as saying they have blue uniforms for purposes of predicting their future performance.

[/ QUOTE ]

It does show that they have either been undervalued or have played over expectation 8 times out of their last 10 games. That does mean something. However, it does not help you in the matchup yet to be played. You still need to determine whether they are STILL being undervalued or not. In some cases Vegas/Public catches up. In other cases they don't.

eternalnewbie · #4 01-03-2006, 10:47 PM

[ QUOTE ]
[ QUOTE ]
In other words, I'm pretty sure that saying a team is 8-2 against the spread is about as useful as saying they have blue uniforms for purposes of predicting their future performance.

[/ QUOTE ]

It does show that they have either been undervalued or have played over expectation 8 times out of their last 10 games. That does mean something. However, it does not help you in the matchup yet to be played. You still need to determine whether they are STILL being undervalued or not. In some cases Vegas/Public catches up. In other cases they don't.

[/ QUOTE ]

Actually, it doesn't show that they were undervalued at all. All it shows is that they happened to cover the spread. This says absolutely nothing whatsoever about whether they were overvalued. Just think about it - even if the book values them absolutely perfectly, and even if betting on them is an insanely bad bet, we know for a fact that you will nevertheless win a certain percentage of the time.

Saying that they were undervalued is like saying it's a good idea to play 27s in holdem UTG just because you happen to pull out the flush one time - or even 8 times in a row.