NCAAB Kenpom Pyth Discrepancies 2014-2015

[spankmythighs]

So I have a current totals thread using Kenpom predictions and Bet365. This thread will be dedicated to LARGE discrepanicies between Kenpom Pyth calculations and Bet365.

Here's the equation:

(Away Team Pyth) - (Home Team Pyth + 3) / 2

Whatever the number is, check it against the posted spread. If there is a difference of 7 or more points between equation answer and spread then you will bet it.

Here is an example:

Dec 2, 2014

Georgia vs. Tenn Chatt
Current line Georgia -7.5

Georgia Pyth = 74.5
Tenn Chatt Pyth = 33 + 3 (Home advantage) so 36

(74.5-36)/2=19.25 which is >7 from current line so we bet Georgia -7.5

[spankmythighs]

DEC 2, 2014

Georgia -7.5
St Johns -17.5
Charleston -10.5
Texas -23.5

[spankmythighs]

I may dabble with giving more points too for value. Example: Texas -27 @ +135

[HeeluvaGuy]

Good stuff. Thanks for sharing. Has this been back tested at all?

[spankmythighs]

Not really. Tough to go back as the Pyth is updated daily.

[spankmythighs]

Was 2-0 last night though with Baylor and S. Caro. Is that enough backtesting lol

[THam12]

This may be a really dumb question, but what is pyth?

[spankmythighs]

Its basically each teams winning percentage broken down using statistics.

[spankmythighs]

Expected winning percentage. And it is updated after each game.

[HeeluvaGuy]

Not really. Tough to go back as the Pyth is updated daily.

Yeah. Figured that'd be tough but thought I'd ask. I'll be following along with interest. GL

[spankmythighs]

Yeah. Figured that'd be tough but thought I'd ask. I'll be following along with interest. GL

Thanks bud.

[spankmythighs]

What I will do is bet $50 units and any losses from a night will be distributed evenly throughout the next night.

[spankmythighs]

DEC 2, 2014

Georgia -7.5 WIN
St Johns -17.5 LOSS
Charleston -10.5 LOSS
Texas -23.5 LOSS

Tough start. 1-3 for -2.3 units or -$230. The $230 will be distributed over tomorrows plays.

[drfunkmaster]

all that mumbo jumbo. and you went 1-3. cant wait to see your plays tomorrow.

[HeeluvaGuy]

So it looks like this method is 2-0 picking road favorites?

I think Dayton fits that criteria today.

[RestlessSpirit]

Spank, I really like your equation approach using Kenpom. I'm a bit of a statistical nerd myself, so I'm always looking for new ways to calculate predictions and haven't been able to find a good approach using Kenpom ratings.

Have you ever played around with AOPR equations using Sagarinis rating? I found it online and if you haven't I can break it down for you.

As with any equation though it takes a good 6-8 games in worth of data to get a good edge. I'm just starting to run it myself.

Your total point approach is pretty dang good as well.

[Fred The Hammer]

KenPom talk is always appreciated by me! I study it everyday as well. I think the KP mismatch tonite would be Valpo/E. Kentucky.

[RJSingle]

Going by this formula and all the data looks like the plays of day are Cal -14 (23.81) Dayton -10 1/2 ( 23.42). Butler, UCLA, Old Dominion, and Southern Illinois also are 7-8 point discrepancies. GT, Wichita St, and San Francisco were 6 pt differences.

[spankmythighs]

Spank, I really like your equation approach using Kenpom. I'm a bit of a statistical nerd myself, so I'm always looking for new ways to calculate predictions and haven't been able to find a good approach using Kenpom ratings.

Have you ever played around with AOPR equations using Sagarinis rating? I found it online and if you haven't I can break it down for you.

As with any equation though it takes a good 6-8 games in worth of data to get a good edge. I'm just starting to run it myself.

Your total point approach is pretty dang good as well.

Not familiar with that one. Fill me in. But I do see a trend with away teams covering more often when using this equation. I will post my plays and unit sizes closer to 4pm MT

[chalk46]

KenPom talk is always appreciated by me! I study it everyday as well. I think the KP mismatch tonite would be Valpo/E. Kentucky.

In favor of Valpo?

[spankmythighs]

Plays for Dec 3

BUTLER -6.5
DAYTON -9
E. CARO +3
OD -3
CAL -12.5
UCLA -15

Last night lost 2.3 units so adding $130 to my bets as $50 = 1 unit. So adding $22 to each for $72 per pick.

[spankmythighs]

Plays for Dec 3

BUTLER -6.5 - AWAY
DAYTON -9 - AWAY
E. CARO +3 - AWAY
OD -3 - AWAY
CAL -12.5 - HOME
UCLA -15 - HOME

Last night lost 2.3 units so adding $130 to my bets as $50 = 1 unit. So adding $22 to each for $72 per pick.

Just updating where they are playing. So we can keep track.

[RestlessSpirit]

I ran three different statistical formulas, including the one you broke down above. All three confirmed our thoughts on Butler and Dayton so those where my two plays. they also confirmed the OD but it was not a substantial differential like the previous to. Let's see how they all roll out.

[HeeluvaGuy]

I ran three different statistical formulas, including the one you broke down above. All three confirmed our thoughts on Butler and Dayton so those where my two plays. they also confirmed the OD but it was not a substantial differential like the previous to. Let's see how they all roll out.

I don't have any models per se, but I ran a second analysis and rated the plays:

Butler (w)
Cal
Dayton (l)
ODU (w)
ECU (l) - this one was way off on my second
(I didn't include UCLA)

[spankmythighs]

No plays today. Year total -2.7u or
-$135

[RestlessSpirit]

Spank, I know you have a subscription to KenPom fan-match. Just out of curiosity what alls in their and is the breakdowns worth the money. I know it's pnly $20 but just wondering.

[RestlessSpirit]

Sorry for the delay on breaking down the AOPR analysis I previously spoke of, but I am just now having time to type this out. I'd first like to state that I take absolutely no ownership of this and actually just randomly ran across it while doing a web search looking for an analysis of its type. I am sure a lot of you either use it or something similar when trying to calculate projected scoring outcomes of games. I began using it full force a few days back, as it needs 5+ games of data to be effective and should get better as more games are played. It has been rather accurate thus far.

So this is how I understand the analysis and use it myself. Once I select a game I want to look at I do the following.

1) Get each teams strength of schedule thus far. I use sagarini's.
2) Get each teams ppg offense and defensive numbers
3) Divide the higher SOS by the lower SOS. This gives you a way to get the two teams on an equal playing field and give you an SOS differential.
4) Divide each teams offensive scoring by the D1 average, which I use 71 ppg.
5) Multiple each teams (offensive results divided by D1 average) against the SOS differential number
6) Now add the SOS differential to the NEW PPG calculation of the team with the higher SOS and subtract the SOS differential from the NEW PPG calculation of the team with a lower SOS to give you a new PPG for each team taking SOS into the analysis
7) Lastly you multiple each teams NEW PPG by their opponents Defensive PPG to get a total projected score of the game. Then add in 3 points to the home team for home court advantage.

So for an example I'll use the Butler vs Indiana State game from 12/3.

Butler -
Sos - 73.08
Off PPG - 72.7
Def PPG - 54.5

Indiana State-
Sos- 69.32
Off PPG - 66.3
Def PPG - 69.7

3) 73.08 / 69.32 = 1.054
OSS differential= .054

4) Butler - 72.7 / 71 = 1.024
IN. St. - 66.3 / 71 = .934

5) Multiple by OSS Differential
Butler - 1.024 x .054 = .055
IN. St. - .934 x .054 = .050

6) Add differential to (PPG / 71) to higher SOS and subtract from lower SOS
Butler - 1.024 + .055 = 1.079
IN. St. - .934 - .050 = .884

7) Multiple new PPG to opponents Def PPG to get projected score.
Butler - 1.079 x 69.7 = 75
IN. ST. - .884 x 54.5 = 48
- add +3 for home court advantage

Projected score
Butler 75
IN. St. 51

From their, like with the KenPom I look for game that have major differences between projected score and spread. I have played this analysis and gone 6-1 with it. Dayton was the one game that was a lose and it was substantially in their favor.

Please let me know of any fine tuning or fallacies in this approach as it's always a learning process.

[spankmythighs]

Spank, I know you have a subscription to KenPom fan-match. Just out of curiosity what alls in their and is the breakdowns worth the money. I know it's pnly $20 but just wondering.

Its pretty decent info with great stats and a prediction of the games score.

[spankmythighs]

Sorry for the delay on breaking down the AOPR analysis I previously spoke of, but I am just now having time to type this out. I'd first like to state that I take absolutely no ownership of this and actually just randomly ran across it while doing a web search looking for an analysis of its type. I am sure a lot of you either use it or something similar when trying to calculate projected scoring outcomes of games. I began using it full force a few days back, as it needs 5+ games of data to be effective and should get better as more games are played. It has been rather accurate thus far.

So this is how I understand the analysis and use it myself. Once I select a game I want to look at I do the following.

1) Get each teams strength of schedule thus far. I use sagarini's.
2) Get each teams ppg offense and defensive numbers
3) Divide the higher SOS by the lower SOS. This gives you a way to get the two teams on an equal playing field and give you an SOS differential.
4) Divide each teams offensive scoring by the D1 average, which I use 71 ppg.
5) Multiple each teams offensive results against the SOS differential number
6) Now add the SOS differential to PPG calculation of the team with the higher SOS and subtract the SOS differential from the PPG calculation of the team with a lower SOS to give you a new PPG for each team taking SOS into the analysis
7) Lastly you multiple each teams NEW PPG by their opponents Defensive PPG to get a total projected score of the game. Then add in 3 points to the home team for home court advantage.

So for an example I'll use the Butler vs Indiana State game from 12/3.

Butler -
Sos - 73.08
Off PPG - 72.7
Def PPG - 54.5

Indiana State-
Sos- 69.32
Off PPG - 66.3
Def PPG - 69.7

3) 73.08 / 69.32 = 1.054
OSS differential= .054

4) Butler - 73.08 / 71 = 1.024
IN. St. - 69.32 / 71 = .976

5) Multiple by OSS Differential
Butler - 1.024 x .054 = .055
IN. St. - .976 x .054 = .053

6) Add differential to higher SOS and subtract from lower SOS
Butler - 1.024 + .055 = 1.079
IN. St. - .976 - .053 = .923

7) Multiple new PPG to opponents Def PPG to get projected score.
Butler - 1.079 x 69.7 = 75
IN. ST. - .976 x 54.5 = 53
- add +3 for home court advantage

Projected score
Butler 75
IN. St. 56

From their, like with the KenPom I look for game that have major differences between projected score and spread. I have played this analysis and gone 6-1 with it. Dayton was the one game that was a lose and it was substantially in their favor.

Please let me know of any fine tuning or fallacies in this approach as it's always a learning process.

Ahh yes Ive seen something similar to this.

[spankmythighs]

Thanks for sharing RestLess!

[jtoler]

Good luck going forward Spanky.

[spankmythighs]

RestLess do you go through every game or how do you decide which games?

[spankmythighs]

Just a heads up on my Kenpom Totals strategy, I am testing it against inplay on Bet365 when there are no Totals that stand out before the game starts. It is great for early payout!

[mth61]

I recently bought a kenpom sub so i will be playing with a few equations and follow yours as well GL.

[spankmythighs]

I recently bought a kenpom sub so i will be playing with a few equations and follow yours as well GL.

Sounds good bud! Lets expose the value plays and get some winners!