wanna be pissed?

[sieken]

Again, I don't want to put myself in a position where I sound like I think FPI is "good". There were more predictive models this year. A FPI critique I haven't seen anyone make yet is that they don't show their math relative to others. For example, it would be nice to know how win probability gets weighted relative to offensive or defensive efficiency ratings. It's always been too much of a black box for my tastes.

But to your question...just looking at Texas' schedule tells the tale:

-Lost to OU by 7 (ranked 5th).
-Lost to OK St by 8 (ranked 7th).
-Lost to Baylor by 7 (ranked 15th).

You can also play with the Resume and Efficiency rankings too to dive a bit deeper. At the end of the day Texas is ranked 0.4 points higher than UK which is pretty marginal imo. Looks to me like UK gets dinged that it's win probability suggests it was really a 8.5 to 9 win team based on the fundamentals.

I think Arkansas is just as bad if not worse than UK and Iowa off of there. If the program digs as deep as you are suggesting, Iowa and Ark should have been caught by the program. I don't think there is a bias against any one team, I think it is totally useless. I can't imagine using it to help decide a winner to bet on.

 

[satcheluk]

What I read on the FPI is that their ~75% accuracy is comparable to Vegas’ ability to predict outcomes. I would think that for the FPI to be a useful tool to rely on to predict game outcomes it would need to exceed the prediction accuracy of Vegas. Otherwise you would just take moneyline favorites and win 75% of the time. I almost always look at the FPI/BPI predictor when I bet and my anecdotal experience is that there isn’t a significant deviation from Vegas’ lines. If you use it to bet on sports and you have success, I’m really happy for you. To me, I’m not that big on a model with 75% accuracy on picking the winner of the game.

Fair points. I don't use it as my only source of data, but I look at it for every game I bet on, mainly looking for discrepancies between multiple data points hoping to find an edge I can arbitrage.

 

[YaketySax]

I think Arkansas is just as bad if not worse than UK and Iowa off of there. If the program digs as deep as you are suggesting, Iowa and Ark should have been caught by the program. I don't think there is a bias against any one team, I think it is totally useless. I can't imagine using it to help decide a winner to bet on.

Ya I wouldn’t use any one model if betting -115 odds on spreads or totals. Best I’ve seen was SP+ this year hitting 65% ATS on FCS the first six weeks of the season but that’s just not sustainable.

I spent two seasons aggregating six models that I liked and played around weighting them differently. Hit 51% and 53% but got tired of spreadsheets and betting 0.2 units for 10-15 games. Gambling started to remind me of work instead of leisure it’s intended to be.

Pretty common approach so lots of Google docs and other instructions for doing so out there.

 

[Cat06]

It's an algorithm.

ESPN is responsible for the Algorithm. It didn't create its self.

 

[satcheluk]

ESPN is responsible for the Algorithm. It didn't create its self.

Indeed or maybe they subcontract it out, but the point is ESPN is not showing bias towards UK, rather their algorithm values/disvalues traits that don't seem to work out in our favor recently. If we improve in those areas, then we will benefit. ESPN is not out to get us which is how many commenting seem to feel.

 

[YaketySax]

SP+ creator and proprietor:

 

[ganner918]

You seem to be under the mistaken impression I care enough to convince anyone. lol. Like I said earlier, FPI wasn't good when it ranked UK higher than most in 2016 and it's not good now when it ranks UK lower than others. A better reading comprehension would realize I'm explaining why FPI came up with the numbers it did. Not that it's "good".

...as for the other assertions without evidence...if you don't believe there's luck in sports I don't know what to tell ya lol. A game where sometimes the wind catches the ball wrong in the air after it's kicked, or when a gang of 300+ pound sweaty dudes are dogpiling on the ground for a slick, bouncing oblong ball, or a knee injury going terribly wrong at a key position, etc DEFINITELY NEVER HAPPENS. No sir. No such thing as catching all the breaks or none of the breaks in sports.

You know all of this. You just don't like the messenger, so you think it's wrong. lol.

I've been in your shoes trying to explain how the NET works in basketball. They won't accept an explanation of "it's a computer model, and it has its flaws, but it has a logic behind what it does." Try to explain why it has teams where it does and they'll ignore that you said you don't like the model and don't trust it. People just want to be mad.

 

[JC CATS]

That’s Iowa STATE.

Iowa, Iowa State, it's all the same to them, talking points