Model Update - Past Month Only

[Aike]

I ran an update this morning to see how teams are tracking over the past month only.

Fairly interesting to me in that several teams have been playing at a really high level. UCONN has been a juggernaut.

Always good to keep in mind that a 7 or 8 game sample is going to tell you a lot less than the overall 31 game sample, but it is good directional information, imo.

Here’s how it shakes out:

1. UCONN 4.73
2. Houston 3.85
3. Auburn 3.75
4. Tennessee 3.14
5. Creighton 3.05
6. Gonzaga 2.77
7. Arizona 2.67
8. Kentucky 2.61
9. Purdue 2.57
10. Duke 2.53
11. Texas 2.13
12. St Mary’s 2.04
13. UNC 1.94
14. Marquette 1.87
15. St John’s 1.81
16. Baylor 1.79
17. Colorado 1.76
18. USC 1.61
19. Texas Tech 1.54
20. Florida 1.49
21. Cincinnati 1.47
22. Villanova 1.454
23. TCU 1.452
24. Pittsburgh 1.451
25. Ohio St 1.38
26. FAU 1.36
27. Nebraska 1.35
28. San Diego St 1.34
29. Iowa St 1.33
30. Clemson 1.30
31. Alabama 1.27
32. Iowa 1.26
33. Nevada 1.25
34. Michigan St 1.21
35. Kansas 1.17
36. BYU 1.16
37. Boise St 1.14
38. Illinois 1.12
39. Syracuse 1.11
40. Mississippi St 1.09

St John’s, USC, and Ohio St stick out as dangerous teams for upcoming conference tournaments.

Season is taking its toll on Iowa St, BYU, Kansas, and Alabama.

 

[WeWant9]

Auburn is an interesting team to me. I know they’re coming in solid in every metric, but there’s something off to me on them. Depending on their draw, I could easily see them going home early.

 

[ukdesi]

Thanks for posting this. Would be interesting to see how this model using recent data will perform in the tournament vs a model using season long data.

 

[Aike]

Auburn is an interesting team to me. I know they’re coming in solid in every metric, but there’s something off to me on them. Depending on their draw, I could easily see them going home early.

They play incredibly hard, but their guards are suspect. I don’t think anyone trusts them.

 

[LineSkiCat14]

Now do just November.

 

[Aike]

Thanks for posting this. Would be interesting to see how this model using recent data will perform in the tournament vs a model using season long data.

I’ve tested it extensively, and the full season model typically outperforms the recent model.

I think this is good info though to see which teams are really hot or cold. Can definitely help with upset picks.

 

[Aike]

Now do just November.

I did. In November.

 

[LineSkiCat14]

Yeah its probably not great to take the small sample size.. but for UK, this model might be more accurate than the whole-season model. We are CLEARLY not the same team. It also might be the same for Kansas, who is trending way down and is barely holding onto just 1 win in the tournament now.

 

[Aike]

Yeah its probably not great to take the small sample size.. but for UK, this model might be more accurate than the whole-season model. We are CLEARLY not the same team. It also might be the same for Kansas, who is trending way down and is barely holding onto just 1 win in the tournament now.

I think this validates what a lot of us are seeing. A surprise to me was how many teams have kicked it up a notch over the past month.

 

[JimBR]

Auburn is an interesting team to me. I know they’re coming in solid in every metric, but there’s something off to me on them. Depending on their draw, I could easily see them going home early.

Absolutely agree. They are great at home and mediocre or average away from Auburn.

 

[ukdesi]

I’ve tested it extensively, and the full season model typically outperforms the recent model.

I think this is good info though to see which teams are really hot or cold. Can definitely help with upset picks.

Ahh ok. Do you think there would be value in combining the model in some sort of way? I don't mean combining into 1 model -- I mean using the season long as the baseline, but then incorporating recent trends for possible upset/variation from the straight baseline version.

 

[Aike]

Ahh ok. Do you think there would be value in combining the model in some sort of way? I don't mean combining into 1 model -- I mean using the season long as the baseline, but then incorporating recent trends for possible upset/variation from the straight baseline version.

I’ve done that too. It’s fine. Doesn’t make a lot of difference either way. Tournament is random enough that one year it might help and the next it might not.

Honestly, what I’m publishing as my model right now is two different models blended together. So I tinker with it all the time. If you have two or three models that are mathematically really close, then I think blending them has some merit.

 

[ukdesi]

Gonzaga stands out as one team higher on the recent data list vs the season long list.

 

[Aike]

Gonzaga stands out as one team higher on the recent data list vs the season long list.

Good point. They have clearly figured some things out.

 

[nickhorvathsuxazz]

I ran an update this morning to see how teams are tracking over the past month only.

Fairly interesting to me in that several teams have been playing at a really high level. UCONN has been a juggernaut.

Always good to keep in mind that a 7 or 8 game sample is going to tell you a lot less than the overall 31 game sample, but it is good directional information, imo.

Here’s how it shakes out:

1. UCONN 4.73
2. Houston 3.85
3. Auburn 3.75
4. Tennessee 3.14
5. Creighton 3.05
6. Gonzaga 2.77
7. Arizona 2.67
8. Kentucky 2.61
9. Purdue 2.57
10. Duke 2.53
11. Texas 2.13
12. St Mary’s 2.04
13. UNC 1.94
14. Marquette 1.87
15. St John’s 1.81
16. Baylor 1.79
17. Colorado 1.76
18. USC 1.61
19. Texas Tech 1.54
20. Florida 1.49
21. Cincinnati 1.47
22. Villanova 1.454
23. TCU 1.452
24. Pittsburgh 1.451
25. Ohio St 1.38
26. FAU 1.36
27. Nebraska 1.35
28. San Diego St 1.34
29. Iowa St 1.33
30. Clemson 1.30
31. Alabama 1.27
32. Iowa 1.26
33. Nevada 1.25
34. Michigan St 1.21
35. Kansas 1.17
36. BYU 1.16
37. Boise St 1.14
38. Illinois 1.12
39. Syracuse 1.11
40. Mississippi St 1.09

St John’s, USC, and Ohio St stick out as dangerous teams for upcoming conference tournaments.

Season is taking its toll on Iowa St, BYU, Kansas, and Alabama.

Gotta love the fact that over the past month we've beaten 2 of the top 4 performing teams of the last month

EDIT: both on the road!

 

[Aike]

Gotta love the fact that over the past month we've beaten 2 of the top 4 performing teams of the last month

Kind of crazy. Even crazier that Auburn is that high with 2 losses. Lost to us and UT but smoked everyone else, including USC by 40.