jarrett

Syracuse University SAL class of 2025, UVA MSDS class of 2026. Check out my GitHub for research projects:

[jarrett @jlmalytics]

@Eugenetiwat_ This is heat

[jarrett @jlmalytics]

@arjun_menon14 Oh my gosh Arjun Menon from the SU Club Tennis team with my just had an awesome hit tweet

[jarrett @jlmalytics]

ELECTRICCCCCCCCCCC LETS GOOOOOOOOOOOOOO

[Virginia Men's Tennis @UVAMensTennis]

CHAMPS.

[jarrett @jlmalytics]

Absolutely unreal

[Virginia Men's Tennis @UVAMensTennis]

AND TWO.

Santamarta wins 3-6, 7-6, 6-4

4-3 HOOS‼️

[jarrett @jlmalytics]

@Colin_Hanley1 Awesome, really cool stuff!

[jarrett @jlmalytics]

@jonahsoos24 Awesome tweet Jonah Soos

[jarrett @jlmalytics]

Both models predicted 36/37 at-large teams both had Auburn (RF over Miami OH, XGB over Missouri) and were 36/68 for seeds.
Models were VERY influenced by predictive metrics, which Auburn was great in. Seed performance is bad (especially for 11+) due to the limited no. of classes.

[jarrett @jlmalytics]

Updated bracket with Penn instead of Yale. Definitely think the seed line performances will be quite poor (I'd be happy with around 75%), but hope at least one model captures the at-large teams correctly. Regardless, with more data, the models will be even more accurate!

[jarrett @jlmalytics]

Updated bracket with Penn instead of Yale. Definitely think the seed line performances will be quite poor (I'd be happy with around 75%), but hope at least one model captures the at-large teams correctly. Regardless, with more data, the models will be even more accurate!

[jarrett @jlmalytics]

Final bracketology! It will come down to Auburn/Missouri/Texas/SMU/Miami OH for the last 4 in (assuming VCU wins today). Really curious to see how the models perform along seed lines, and if they perform poorly, there's always more data to train/test on next year :)

[jarrett @jlmalytics]

Final bracketology! It will come down to Auburn/Missouri/Texas/SMU/Miami OH for the last 4 in (assuming VCU wins today). Really curious to see how the models perform along seed lines, and if they perform poorly, there's always more data to train/test on next year :)

[jarrett @jlmalytics]

Bracketology 3/14! Random Forest (RF, left) and XGBoost (XGB, right). Along the seed lines, the RF model looks better, but in terms of the first four, XGB seems aligned with consensus. Would not be shocked if Missouri was left out of the tournament for Auburn or Texas...

[jarrett @jlmalytics]

Seems like Miami OH’s ceiling is Dayton. Missouri also seems like they are likely in Dayton. SMU w/ BJ Edwards seems to have a floor of Dayton. Expect NC State and Santa Clara securely in. I think Auburn’s metrics gets them in the first four…

[jarrett @jlmalytics]

Build another bracket, but this time with XGBoost rather than Random Forest. Seed lines seem pretty volatile. Last 4 in are SMU, Texas, Miami OH, and Auburn. Which means we have Missouri out of the tournament. Excited to compare both models with what the committee does

[jarrett @jlmalytics]

Build another bracket, but this time with XGBoost rather than Random Forest. Seed lines seem pretty volatile. Last 4 in are SMU, Texas, Miami OH, and Auburn. Which means we have Missouri out of the tournament. Excited to compare both models with what the committee does

[jarrett @jlmalytics]

Bracketology 3/13. I'm shocked that Oklahoma is now in. This assumes Utah St. wins the MWC title. Texas is the 1st team out, and Miami OH is the 5th. A team with predictive/resume/quad metrics like Auburn's has consistently been put in the tournament, and Miami OH is the inverse

[jarrett @jlmalytics]

Bracketology 3/13. I'm shocked that Oklahoma is now in. This assumes Utah St. wins the MWC title. Texas is the 1st team out, and Miami OH is the 5th. A team with predictive/resume/quad metrics like Auburn's has consistently been put in the tournament, and Miami OH is the inverse

[jarrett @jlmalytics]

@2ndChancePoints That would be really fun and fascinating. I suppose if Auburn gets to the SEC conference semis they might miss the first four altogether. Might not be possible if Utah St. doesn't win the MWC title or if someone steals a bid in ACC/B10/SEC tourneys

[jarrett @jlmalytics]

@2ndChancePoints KPI, WAB, and SOR are definitely good enough, but I expect the UMass loss (Q3/Q4) plus some poor predictive metrics to put them in last 4 in territory

[jarrett @jlmalytics]

@2ndChancePoints You’re right; sorry I thought by resume metrics you were saying things the committee takes into account - which predictive metrics are something they take into account. And no team with predictive metrics like that has gotten in before

[jarrett @jlmalytics]

Bracketology 3/12. As stated previously, the model loves Auburn's metrics, so they are projected in the field as one of the last 4 byes. Missouri is the last team in, and the model really dislikes their metrics, wouldn't be shocked if the committee was lower on Missouri

[jarrett @jlmalytics]

@kenzbrooksbets How are you calculating SEL%? Are you using some type of model or simulation process?