In partnership with CBSSports.com
The largest and most active MSU Spartans board on the web
The place to ask questions to SpartanTailgate's recruiting experts
"The Duff" is dedicated to Michigan State football recruiting discussion
"The Bres" is dedicated to Michigan State basketball recruiting discussion
This is your pulpit to preach to the masses about everything from politics to religion
The place to buy, trade or sell Michigan State tickets
You have no favorite boards.
The most viewed topics.
The most replied to topics.
The most up-voted topics.
The most down-voted topics.
The most up-voted posters.
The most down-voted posters.
The most followed posters.
Who said the KenPom rating's predictions have been "so good"? It wasn't me. The fact is that compared to other ratings with the stated goal of predicting scores, KP's has been pretty good, but not the best.
What do the KP ratings have to do with Wisconsin's performance vs the betting line?
Draymond Green, Magic Johnson, and Oscar Robertson are the only players to have two triple-doubles in the NCAA Tournament.
From what I gathered the KenPom model would predict a higher margin of victory for UW than the point spread?
So this is the game with the Highest Thrill Score being played tonight:
52 Oklahoma at 35 Iowa St. Iowa St. 74-69 (68%) 7:00 PM Ames, IA Hilton Coliseum
lets see what happens.
And if you want to see a game on tv tonight:
50 Notre Dame at 7 Syracuse Syracuse 74-62 (87%) 7:00 PM Syracuse, NY Carrier Dome
Signature for rent
I believe you overrate his significance. That's all.
Big MOnday sucks
Here's the site I was talking about.
I don't know who runs the site or how they decided which rating systems to include. They left off some pretty well-regarded ones (Dolphin, LMRC, KenPom, and others). Also not sure why they include Sagarin's ELO, as it is not a prediction-oriented system.
Of the ratings tracked by this site, Jon Dokter, StatFox and Sonny Moore are the three most accurate for picking winners and losers. Against the spread, Sevenovertimes.com, Sonny Moore, and Powerratings.com have been the best.
I remember being a bit surprised when I learned that the combined Sagarin rating is a slightly better predictor of winners and losers than either the Predictor or the ELO in isolation.
NCAA Basketball Prediction Results
Daily NCAA Basketball Predictions
Football ratings, NFL, NCAA Football Basketball Baseball Computer ratings predictions
This post was edited by red_cedar 17 months ago
I didn't say it would have turned out differently only that it very easily could have. The Palace that weekend was louder than the Breslin has ever been. Those were two home games for State and there is no doubt that being at home helps. Both games were tight and I could see Syracuse holding their lead if the game was in NYC or Iowa State squeaking out a tough game in Kansas City.
There is a reason Duke and UNC make so many final four runs in comparison to other teams just as talented. Nearly every NCAA tourney they play at home for 2/3 of the tourney.
If KP ratings were accurate they should indicate errors in the betting line, however they don't. KP is not an accurate indicator of a team's strength as based on results ATS.
BPI takes into account players that missed the game, though I don't believe that it factors rest. Though if I'm not mistaken it's been long proven that rest is completely irrelevant in college basketball (no back to back games like the NBA).
Did anyone, anywhere, claim that KenPom's computer prediction ratings should or could be more accurate than the Vegas line?
I believe you are mistaken, but do you have a link to back up your claim?
Syracuse won 63-47 - 16 point margin, KP had 12, Vegas: Syracuse -8.5
Iowa St. won 83-64 - 19 point margin KP had 5, Vegas: ISU -6
inconclusive. But Vegas wasn't close on either.
Any of the other posted power ranking and predictors have MSU somewhere between 11-20. It's really just a matter of what they value more.
This post was edited by Kerndaddy 17 months ago
The betting line is not a score predictor, it's an accurate representation of the betting. Vegas doesn't gamble, it just sets it up with the line so that bets equal out both ways, so whatever the outcome they just charge commission on the transfer of money from one party to another.
The closing line is effectively a crowdsourced score prediction.
Since Vegas doesn't set odds after the fact (like horse racing, I believe), and bettors are interested in making money (although I'm sure some homerism/entertainment plays into it) and potential bettors with lots of stats and potential financial backing would try to exploit profitable inaccuracies, I think it's pretty safe to say that the Vegas line is a good predictor of what will happen. Additionally, I've seen claims (NYT article by Stephen Dubner comes to mind) that Vegas does try to exploit bettor biases. However. That 10% juice doesn't provide a ton of room to move a line from the "true line," before they expose themselves to smart money coming in. I guess they can try to limit this by limiting individual bets, etc.
UM is a -7.5 at some casinos today, damn!
It would be if fans couldn't bet, and it was only professionals (or at least people doing it strictly for financial gain). The sharps bet the best game (the assumed disparity in spread and likely results), fans bet their own team regardless, or on a big game.
Our elected officials are a crowd sourced prediction on who should be the best leaders for our country too. But elections, like betting, are often about what/who we wish was the best. And like voters in the next election, betters often aren't willing to admit past mistakes.
And while Vegas does want money on both sides of it, most books aren't looking for a true 50/50 split (the lines would be much more volatile if they did). The Vegas books are essentially dollar cost averaging their picks, and if the line has moved a couple points, there is a sweet spot where they can win both sides. Vegas is giddy when betting fans take the bait, and bet the line down to where they actually think it should be.
Thanks for the link.
The assumption goes something like this:
Teams' relative strength as viewed by the public is going to roughly mirror the current polls (AP, Coaches, etc.) If a team like Wisconsin is consistently ranked lower in polls and other sources than they are by KenPom, then the public perception of Wisconsin would be lower than KenPom's ranking, the public would bet roughly in that manner, and if KenPom were correct then Wisconsin would usually beat the spread. But they don't, which means that either the betting public agrees more with KenPom than they do with other sources (doubtful) or that KenPom isn't an accurate predictor (ding ding ding).
1. Seems to me that you are fighting against a strawman. You didn't answer my question about who claimed that KenPom could or should be more accurate than the betting line? Nobody who knows anything about computer ratings would claim what you seem to think they are claiming.
2. According to the link I provided, the opening line is the most accurate predictor of the winner/loser, and the closing line is the most accurate predictor of the scoring margin (wisdom of the crowds).
3. Any systems that provide insight into the relative strengths of teams end up being factored into the betting lines.
Nice first sentence.....
Jeebus. Move on.
In this very thread there is at least one person who states they use KenPom as their predictor and that they win regularly with it. Scroll through if you want to see who.
I have no idea why you state your 2nd point .
I don't personally know anyone who sets betting lines. I'm suggesting that the ones that are leaders in the field do not give much or any weight to KenPom. I'd be happy to be proven wrong.
Let me just say I am impressed with you dorks. I often thought of venturing further into the statistical world of sports and am surprised that there are those here that have done so. I am going to keep this thread in mind for these references in case I ever get around to it.
Just out of curiousity, what would motivate you to venture into the statistical world when a majority of the very data is inconclusive? I hope if you dabble into this world, you keep it specific to a TEAM.
I figured the guy that said he makes money using KP as his guide is talking about office bracket pools. You won't beat Vegas in the long run with any of the computer formulas we've mentioned in this thread.
247Sports In partnership with CBS Sports