Pomeroy upgrade

[tornado]

Kenpom upgrades, ready for 2010-2011, and BU ranked 86

http://kenpom.com/rate.php

He projects BU to go 18-11..and thinks Duke will beat us...

http://kenpom.com/team.php?team=Bradley

[shaunguth]

Kenpom upgrades, ready for 2010-2011, and BU ranked 86

http://kenpom.com/rate.php

He projects BU to go 18-11..and thinks Duke will beat us...
http://kenpom.com/team.php?team=Bradley

I obviously don't understand his system.....

His game-by-game projected results have us going 22-7 (13-5), but his projected record has us at 18-11 [10-8].

[tornado]

he needs to hire our stat guys

[squirrelgotdead]

I obviously don't understand his system.....

His game-by-game projected results have us going 22-7 (13-5), but his projected record has us at 18-11 [10-8].

His numbers are based on efficiency calculations deduced from actual performance factors.

The simplest explanation is predictions and projections are two separate matters.

There are two separate calculations at play. His predictions measure the overall likelihood of outcome in a single game.

His projections have the Pythagorean theorem applied which acts as a sort of equalizer and supposes every team plays the same or similar schedule. Doing so projects out the season and allows for unexpected results to be figured into the equation, hence the variance.

[thefish7]

I obviously don't understand his system.....

His game-by-game projected results have us going 22-7 (13-5), but his projected record has us at 18-11 [10-8].

He explains it somewhere... Basically the stats applied to individual games may lead to different results than the season taken on the whole. That's what I remember it saying, with clearer detail somewhere on that site.

[squirrelgotdead]

He explains it somewhere... Basically the stats applied to individual games may lead to different results than the season taken on the whole. That's what I remember it saying, with clearer detail somewhere on that site.

If you click the giant it takes you to the explanations.

It's a lot of math, but efficiency is generally regarded as the most effective measure for basketball. His Offensive Ratings for individual players are basically the same number that drives NBA market forces.

[shaunguth]

His numbers are based on efficiency calculations deduced from actual performance factors.

The simplest explanation is predictions and projections are two separate matters.

There are two separate calculations at play. His predictions measure the overall likelihood of outcome in a single game.

His projections have the Pythagorean theorem applied which acts as a sort of equalizer and supposes every team plays the same or similar schedule. Doing so projects out the season and allows for unexpected results to be figured into the equation, hence the variance.

So he projects that we'll win 22 games, but predicts that we'll win 18 games.

Makes perfect sense now.

[tornado]

So he projects that we'll win 22 games, but predicts that we'll win 18 games.

Makes perfect sense now.

numerous parallel universes, remember??

[squirrelgotdead]

So he projects that we'll win 22 games, but predicts that we'll win 18 games.

Makes perfect sense now.

Not quite. On paper, we should win 22 games. That is the predicted outcome.

However, all things being equal, 18 wins is projected and what the most probable outcome would be.

The terminology seems backwards, yes.

[Tulsa Brave]

Here's the best way I can explain the differences...

BU plays 29 games. Let's assume for a minute that in each game we have a 52% chance of winning. Thus taking each game individually, they each would be a "win" since the win percentage is above 50%. Thus our record would be 29-0 taking each game individually.

However, when all 29 games are put together, since our win percentage in each individual game is only 52%, that means our final overall record would be 15-14 taking all 29 games together.

Now in the real-life situation, our percentages vary for each game. Taking each game individually, you get a 22-7 record for BU. Taking all 29 games together with the percentages for each one, you come up with 18-11 as the most likely final record for BU.

Hopefully this makes sense.

[Beninator]

Last Saturday, a friend of mine asked me to predict our record for the year. I responded that it looked to me that we would be a .500 team or maybe a couple of games over .500. We were in agreement with that projection and it appears Ken Pom is fairly close as well. Also, Ken Pom only projects Duke to score 80 points against us. Given what I have seen so far, if Duke only scores 80 points, I would be happy!

[real fan]

Not quite. On paper, we should win 22 games. That is the predicted outcome.

However, all things being equal, 18 wins is projected and what the most probable outcome would be.

The terminology seems backwards, yes.

This team should win 22 games.

[TheAsianSensation]

There should be a Caution: Math Ahead warning in this thread.

[squirrelgotdead]

This team should win 22 games.

Agreed. But will it?

Here's another thing. In terms of the difference, if you look at the predicted outcomes, essentially 4 conference road games have the same score (1 has the same result, but each team one fewer point) that show us as one point victors.

You can probably assume two of those will in actuality be losses. Two other unexpected, but likely results such as a home loss and at WCU is where the likely projected number is derived from. Just using those as an example.

[shaunguth]

Projected MVC standings....

12-6 Missouri State
12-6 Wichita State
11-7 Northern Iowa
10-8 Bradley
10-8 Creighton
8-10 illinois state
7-11 Evansville
6-12 Drake
6-12 Indiana State
6-12 Southern Illinois