Ken Pomeroy

[courtside]

...has new predictions out and has SLU at 16-12 (8-8 in league play)

SLU @ Pomeroy

[cheeseman]

...has new predictions out and has SLU at 16-12 (8-8 in league play)

SLU @ Pomeroy

Given what has happened I would think this is reasonable - sad to say. I so was hoping for a break out year.

[kshoe]

...has new predictions out and has SLU at 16-12 (8-8 in league play)

SLU @ Pomeroy

Its never entirely clear to me how these computer models can predict rankings and wins/losses before any games are played. I suppose its based primarily on last year's rankings and adjusted for the amount of scoring coming back. No idea how a computer factors in recruits but it likely does.

His predictions seem reasonable, but I'm pretty sure if he told his model that Kwamain was going to be back in January (a high likliehood event) we'd have a better conference record than 8-8.

[MusicCityBilliken]

Its never entirely clear to me how these computer models can predict rankings and wins/losses before any games are played. I suppose its based primarily on last year's rankings and adjusted for the amount of scoring coming back. No idea how a computer factors in recruits but it likely does.

His predictions seem reasonable, but I'm pretty sure if he told his model that Kwamain was going to be back in January (a high likliehood event) we'd have a better conference record than 8-8.

If you follow Pomeroy throughout the season it constantly corrects itself. That was most notable when we moved into conference play last year.

[SpencerFilibuster]

If you follow Pomeroy throughout the season it constantly corrects itself. That was most notable when we moved into conference play last year.

Yeah, I'm curious to know how Pomeroy puts together his algorithm for preseason rankings.

He's got Iowa at 102, Georgia at 69..etc.

[bsheldon]

Yeah, I'm curious to know how Pomeroy puts together his algorithm for preseason rankings.

He's got Iowa at 102, Georgia at 69..etc.

If you look at it, it (the computer) basically just chalks up losses for road games and wins for home games. The only exceptions are really weak or really strong teams--I think only 3 times total. This is very similar if not identical to other prediction sites based on rpi. They will all adjust as the season goes on and data is collected. The predictions really don't mean much until at least about 6-8 games are played so some OK comparisons and rankings can start to come together a bit. Last year most of these sites had us going under .500 until we mathematically couldn't and even then it still had us losing most of our games down the stretch in conference. I think they had us for something like 12-16 at the beginning of the year. Again, we greatly improved as the season went along and got stronger as we got into conference play. We had the one bad loss at home to MOST (home losses are much worse than road ones) when KM was hurt which kind of skewed our rankings a bit and took a while to recover from.

I like to talk about and discuss preseason rankings that are done by people. The computers are too rigid and don't really take into consideration factors like injuries. X on paper looks great once again, but they are missing 2 of their main horses right now so in reality maybe not that great, for now. The computers just look at wins and losses and where those wins and losses took place--that is it. Again, 10-15 games into the season they start to make a bit of sense, but I wouldn't bet with the rpi predictors even then.

[SpencerFilibuster]

If you look at it, it (the computer) basically just chalks up losses for road games and wins for home games. The only exceptions are really weak or really strong teams--I think only 3 times total. This is very similar if not identical to other prediction sites based on rpi. They will all adjust as the season goes on and data is collected. The predictions really don't mean much until at least about 6-8 games are played so some OK comparisons and rankings can start to come together a bit. Last year most of these sites had us going under .500 until we mathematically couldn't and even then it still had us losing most of our games down the stretch in conference. I think they had us for something like 12-16 at the beginning of the year. Again, we greatly improved as the season went along and got stronger as we got into conference play. We had the one bad loss at home to MOST (home losses are much worse than road ones) when KM was hurt which kind of skewed our rankings a bit and took a while to recover from.

I like to talk about and discuss preseason rankings that are done by people. The computers are too rigid and don't really take into consideration factors like injuries. X on paper looks great once again, but they are missing 2 of their main horses right now so in reality maybe not that great, for now. The computers just look at wins and losses and where those wins and losses took place--that is it. Again, 10-15 games into the season they start to make a bit of sense, but I wouldn't bet with the rpi predictors even then.

I agree with your points, I'm just optimistically hoping that there is more to take away from Pomeroy's site.

The other RPI predictor that was posted on this board, was exactly what you were describing...home wins, away losses. It seems slightly different with Pomeroy. Again, I don't take much away from his preseason rankings/estimations, but it is just interesting to me that he has some sort of Offensive/Defensive Efficiency ratings already calculated. What is this based on? Last year alone?

But if you look at Iowa's page, who is ranked 102 in his preseason, and ended last year in the 170 area, and they are probably worse this year than they were last year (skill wise). However, they picked up a new coach from Siena who plays a much faster pace than Lickliter (old coach). The predicted efficiency ratings are much better than what Iowa ended with last year. So his computer has to be doing something more than just away losses and home wins. (I'm originally from Iowa)

However, for SLU, it appears as if they just stuck SLU in there, based on where they were last year, and gave us cookie cutter predictions.

I don't understand why pomeroy would even publish stuff in the preseason if he didn't think the information was notable in some way.

[kshoe]

I agree with your points, I'm just optimistically hoping that there is more to take away from Pomeroy's site.

The other RPI predictor that was posted on this board, was exactly what you were describing...home wins, away losses. It seems slightly different with Pomeroy. Again, I don't take much away from his preseason rankings/estimations, but it is just interesting to me that he has some sort of Offensive/Defensive Efficiency ratings already calculated. What is this based on? Last year alone?

But if you look at Iowa's page, who is ranked 102 in his preseason, and ended last year in the 170 area, and they are probably worse this year than they were last year (skill wise). However, they picked up a new coach from Siena who plays a much faster pace than Lickliter (old coach). The predicted efficiency ratings are much better than what Iowa ended with last year. So his computer has to be doing something more than just away losses and home wins. (I'm originally from Iowa)

However, for SLU, it appears as if they just stuck SLU in there, based on where they were last year, and gave us cookie cutter predictions.

I don't understand why pomeroy would even publish stuff in the preseason if he didn't think the information was notable in some way.

He clearly is making some adjustments for what teams added or lost in the offseason. He's got Illinois #11 overall and UNC #12 and neither made the NCAA tourney last year. But they are bringing back most of their players.

What would be real interesting is if we could have Pomeroy re-run the analysis assuming WR and KM were on the team. Does 16-12 go to 22-6 or is it as at least one poster suggested, just a 1 or 2 game difference.

p.s. many of the "RPI Predictors" actually base the expected wins and losses off the analysis that Pomeroy or Sagarin do, so effectively they tell you what our RPI will be if we do indeed end up 16-12.