West Virginia Numbers Recap
Written by: noreply@blogger.com (Rob Lowe)
Hey folks... we've got some interesting stuff coming this week from John Pudner again. However, before we get to that, I just wanted to recap the numbers associated with the West Virginia game.
(http://3.bp.blogspot.com/_Xpuy4viivhI/SWwCyXY7QFI/AAAAAAAAAjM/cUmUDaPQGOg/s400/West+Virginia.png)
(http://3.bp.blogspot.com/_Xpuy4viivhI/SWwCyXY7QFI/AAAAAAAAAjM/cUmUDaPQGOg/s1600-h/West+Virginia.png)
What should immediately stand out is the defensive efficiency. Marquette's offensive efficiency was about average, but the defense was great. Two of the last three games (this one and Cincinnati) have ended up with stellar defensive efficiencies. Considering our defensive ranking of #49 (http://kenpom.com/sr.php?team=Marquette&y=2009) lags behind our offensive ranking (#8), it's nice to see those results.
Marquette ended up winning on all four factors. Although OR% was essentially a wash, winning and holding WVU below their average helped the overall win margin. TO Rate was a huge key for Marquette, considering that WVU typically forces opponents into a turnover rate of 27% (http://kenpom.com/sr.php?team=West%20Virginia). The Free Throw Rate was a huge disparity, but more because WVU's rate was so low.
As for the recommendations that were posted on the preview (http://www.crackedsidewalks.com/2009/01/marquette-takes-on-west-virginia.html), the results and impact can be found here on MUScoop.com (http://www.muscoop.com/index.php?topic=11553.msg102489#msg102489). No need to recap twice.
Okay. Flat out. I. Was. Wrong. (http://www.crackedsidewalks.com/2008/12/new-years-resolution-expect-less-from.html)
(http://3.bp.blogspot.com/_Xpuy4viivhI/SWwT21UlOjI/AAAAAAAAAjU/K5lqK4jlQwU/s400/BE+Histogram+1.12.09.png)
(http://3.bp.blogspot.com/_Xpuy4viivhI/SWwT21UlOjI/AAAAAAAAAjU/K5lqK4jlQwU/s1600-h/BE+Histogram+1.12.09.png)Since my initial doom and gloom about what to expect this year, Marquette has surged out of the gate. The offense has been playing well, and the defense has been very strong in two of the four contests. Not only is 9-9 no longer the most likely scenario, but given the way the team is playing now, there is a 90% likelihood of finishing with 10 wins or more.
Now, I still see 12-6 as the most likely scenario (instead of Pomeroy's 13-5 prediction (http://kenpom.com/sked.php?y=2009&team=Marquette&t=p)). However, there is a 57% chance of getting 12 wins or more, a 33% chance of getting to 13 wins, and even a 15% chance of finishing with 14 wins. Things would have to go very well, especially on the road (PC, DPU, USF) and down the stretch. If you don't want to get too greedy with the current predictions, there's still about a 45% chance of finishing with 11 wins or less.
Regardless, I'd sign up for 11-7. Saturday's game will be a tough one. Life is hard on the road, but should Marquette win, it'd be reasonable to expect very good things this season.
http://www.crackedsidewalks.com/2009/01/west-virginia-numbers-recap.html
Looking at Pomeroy's predictions, he lists the final conference record as 13-5 but the game-by-game predictions he lists has us going 12-6. Does that mean that the close games (49%, 50% or 51%) just go in as wins?
I was looking at this earlier today and was unsure why the total wins/losses doesnt match up with the game-by-game predictions.
Pomeroy's game by game prediction isn't tied to his overall record prediction. His game by game is a comparison of how the two teams rank, give or take home court advantage. His overall record is based on how team's with our stats are expected to fair statistically regardless of who they play.
Ultimately they should be fairly close, as it is in this case but not always the same.
edit: What he said
That said, I have come to mistrust the Pomeroy win percentage predictions. He predicted us as big favorites last year at West Virginia, at UConn, both UL games, and at Syracuse. He predicted WVU this past weekend as a favorite over us as well.
That's why they play the game...
I know what you're saying.
However, Marquette was predicted to win those five games by 15 points, and the win% prediction was about 75% for each game. Then, not only did Marquette not win, but they wound up losing big in each game. According to Pomeroy's predictions, MU should have won four of those games pretty easily... or maybe won three of those games and been competitive in the other two... or maybe lost four and been competitive in all of them. But MU got smoked in each game.
Either it's a complete aberration that MU would lose big in each game... or something is off with Pomeroy's statistical prediction.
<hint: something is off with Pomeroy's statistical prediction>
I believe Pomeroy's total tally comes from adding up all of the percentages for the remaining games and rounding them to a whole # of wins (e.g., four games at 49% are each losses individually, but account for about 2 wins in the total).
WVU must've done something very well within the bounds of his statistical model - they're still ranked #4 overall with the #4 defensive efficiency (maybe it's not giving enough credit to the opponent's free throw rate, as they send players to the line a lot?)