Version 1.0 of the TG Index is live! I imagine that this is something that will experience multiple iterations, but I wanted to get this out in time for March Madness 2025.
Background
I initially had a theory in 2018 that it was not about selecting the overall winner of March Madness, it was about analyzing individual matchups. If a team has one bad matchup, it could doom their entire tournament, and I did not want to rely on a team to go all the way in that case, even if they had more favorable outcomes overall. So I created some views in a notebook of team playstyles (distribution and efficiency of scoring 2s, 3s, FTs). This heavily influenced how I selected teams in close games if one team had a more favorable matchup (eg. a team that is an elite 3pt shooting team coming up against a team that has poor perimeter defense. I never said this was revolutionary!). This resulted in my best March Madness bracket ever at the time.
Nonetheless, I failed once again to pick the correct March Madness champion. I realized that despite this method of selecting favorable matchups, it was still vital to identify the few teams in the field who had what it takes to go all the way. In 2019 I devised the first evolution of TGX. At the time, it wasn’t great, but the purpose was sound – create an index rating to identify the most important stats from the field based on historical champions’ stats. I have been working on this on and off ever since, and slowly incorporating other views to assist in my bracket-making.
How is TGX calculated?
The current iteration is utilizing eFG%, 3PT%, 2PT%, TO%, Non-Steal TO%, OReb% for offense, Opponent eFG%, 3PT%, 2PT%, Block%, and Average Height. These 11 stats are the most notable in terms of consistent performance and requirement for championship teams to be above average. The index weights these based on level of importance/correlation.
Please note that this is NOT an efficiency metric! I am planning on creating some efficiency metrics of my own, as well as resume metrics. These may potentially be baked in too, if they appear to be statistically significant historically. For now, this is based on raw stats that most meaningfully align with former champions.
Using TGX1 (which I am now dubbing this initial version), the lowest score a champion has finished with since 2007 is a 50.0, while the champion average is about 70.0. The highest was Kentucky’s 2012 team with a whopping 102.0.
Please enjoy as TGX and teddygeary.com grow and evolve! I am happy to have you along for the ride.
teddygeary 
Leave a comment