Game odds, win probabilities, power rankings

This isn't predicting future win losses using previous seasons' results or something along those lines.

This is regressing a correlation for the historical association of a certain point spread toward an outcome. How often do you think a 4 TD favorite loses? How often do you think a 2 TD favorite wins? How about a pickem? I can accept that there's likely a statistically significant difference between percentage that the favorite wins in each of these. If that is the case and there is also a trend, then I should be able to translate the spread into an expectation percentage for victory.

I'm not using a regression to predict, I'm using the historical association of vegas's 'predictions' to wins.

 

And by I in the post above I just mean a random I, not me. Obviously I didn't do the work.

 

Each season has too much turnover on starting rosters to consider each one anything but discrete events. That's why I wouldn't consider blanket history as a good indication. Now, if you want to look at the last 5 or 10 years, I could buy that there is enough program consistency as to make some sort of prediction off of recent events, with the knowledge that the specific roster differences were a source of error.

 

I don't understand that point though. How do the point spreads not already account for that turnover. It's basically a question of how good Vegas can be.

The real argument against it should be that point spreads are designed to move money not predict how much Vegas thinks a team will win by. So based on matchup or fanbase the historical tie between spread and odds of winning will vary. That's a risk of using the regression for a single team for individual games in my opinion. Not the team changes / turnover stuff.

 

The royal we.

 

I see Vegas odds as a direct measure of bettor collective expectations. How the games turn out is irrelevant so long as the line evenly divides the wagers. But I'm listening, I never considered vegas' accuracy as a reflection of a team's ability. For example, a team beating the spread often just means a team is underestimated by the public, not necessarily vegas. But I don't really gamble

 

I generally agree with this. Of course this is likely opening line so it isn't just the public's read....it embeds all the tricks of Vegas. However, if there is a good regression coefficient over let's say the last 10 years then I'm listening.

Also as TV said, he views this more as a tool for setting expectations. And I think that's a good use.

 

http://blog.minitab.com/blog/custom...ity-that-your-favorite-football-team-will-win

 

.

 

12-0

 

A lot of bettors look at the percentage of the time a team goes over the spread, or under the spread. And they look at the opening line to figure that one out, not the way the public moves it. So if you crunch your numbers before the line comes out, you can say "If the line is +7, and team X is 8 and 1 against +7, then I'll put on team X. Otherwise team Y."

 

The initial line is still influenced by the perception of the public by Vegas. Doing it that way just gives the only consistent point in the lead up to the game, before injuries or storylines influence it.

 

I actually think the closing line tends to be the most accurate. The big money usually comes in late and works as a corrective at the end.

 

You could just wait a few hours after the closing line and find out the actual result. The problem is that you typically can't bet after that.

 

I'd hate to go on hearsay though. I'd probably want to witness the game first hand before putting my money down.

 

You don't understand what you're talking about.

 

It doesn't appear you understand what you're talking about either.

 

Closing line is the most accurate.

 

But also not useful for this exercise

 

Bet I do. I haven't looked at the article or how the data is being used, but historical outcomes mean nothing to this year. I assumed that was the methodology.