2023 Season Preview Release Announcement
TNG Season 2 incoming! Season Preview to be released May 31st...
It’s been awhile since we’ve touched base, but we have not been idle. While Spring practice has largely wrapped up across the state and programs have laid the foundation for their next classes to take the field in August, we here at TNG have been working to hone our game as well. Since the season ended we’ve been working on a number of things to try to improve our model and to improve what we bring to you, our nearly 300 subscribers. We’ve got some exciting things to share, so let’s jump right in.
The TNG 2023 Season Preview will be released on Wednesday, May 31st!
As we did last year, we’ll start your summer off right with our preview of the upcoming 6A Texas high school football season. Here’s what you can expect to see next week:
Preseason State Top 50
District Picks with projected order of finish, playoff picks including D1/D2 reps, Playoff Odds for every team (new!)
Playoff Bracket Projections with commentary and odds to advance
Top District Races to Watch
Most Difficult Schedules
Feature Articles (new!)
Is it crazy to release such projections so early? Yeah, probably! But, we’re just fans like y’all and since we have the data compiled, we simply can’t wait to share it with you (plus, last year’s effort fared pretty well). We don’t put these projections out lightly or without reservation. We know we’ll miss on some teams and we’re ok with that. It is not possible to model everything in Texas High School Football perfectly and, frankly, it wouldn’t be worth following if that were the case. We’re simply trying to provide the most objective view of the 6A landscape as is possible in order to capture the great performances, to quantify the magnitude of the upsets we love (and lament), and to properly credit teams on the merit of their achievements. There’s virtually no end to the tweaks we could make to our system and we’ll constantly look to improve. Next, we’ll talk about some steps we’ve taken this offseason to improve our processes and our offerings.
Self-scouting
In our last post to close out the 2022 season, we mentioned a number of things we wanted to take a look at to improve our model. One of the great things about an Elo-based model is that once the season starts, it does its thing and does it very well (What is Elo?, Is Our Model Any Good?, 2022 recap). Where we have the greatest opportunity to affect the accuracy of the model is in our preseason ratings and the process by which we arrive at those ratings. Like any model, if you feed it with good data, you have a better chance of seeing better results.
Last year, we made a strong effort to study each team in as much detail as we could while developing the model and launching the site at the same time. We scratched out our ‘22 preseason evaluations, quite literally, with pencil on paper and got it done, but looking back, there were holes in our process as well as in the legibility of our feverishly scribbled madness. Going into Year 2, with the foundation of the site well-formed, this offseason provided an opportunity to really focus on our research. We were able to create a standard template to collect and record the data that we used to do our individual team-by-team analysis. With this template, we were able to more easily identify specific team trends which we believe has led us to better inter-seasonal evaluations, and it gives us a great platform to potentially expand upon those evaluations in the future.
We also took a hard look back at which teams most overperformed last year’s preseason ratings and which most underperformed and tried to understand what we missed in making our initial assessments. In some cases it was as simple as a team having really strong underclassmen making their mark at the varsity level, and in other cases some teams had a tougher time than we'd expected in replacing a strong senior class.
Where we found the most opportunity for improvement, however, was in simply trusting the historical Elo ratings of the teams a little more. As much insight as we feel our adjusted scoring metrics give us (AOS, ADS, DOM), they aren't as dynamic or as well-calibrated as Elo is proving to be. Whereas last year's preseason evaluations used the adjusted scoring metrics to derive an Elo, this year's initial ratings are more of a blend of the two with heavier emphasis on the Elo side. While that may sound like a pretty big shift, it's really just fine-tuning. We did extensive testing in the offseason to see how much our preseason ratings could wreck the overall system, and the great thing is that due to the huge amount of data feeding the model each week and due to the genius of Mr. Elo's math, the model will find the proper level pretty quickly. Our attempts to improve our preseason adjustments will simply give us a chance to have a better model early in the season and will hopefully help it perform better in sorting out teams rated near each other. Our goal is to push the model from its 78.2% accuracy mark last year (link) to somewhere over 80%.
Another modeling variable we tried to better understand was the immediate effect of coaching changes. There will always be turnover in the coaching ranks and we know how coaching can affect a program's outlook, but can we actually model a coaching change impact in-between seasons? It might seem simple enough on the surface, but the data is confounding. Looking deeper into the data this offseason and seeing examples of coaches with middle-of-the-road records at one program then taking their next programs to historic heights (and vice versa) as well as having a great appreciation for the coaching profession and all that goes into making a successful program, we are convinced that attempting to proactively adjust for coaching changes in-between seasons is a perhaps futile endeavor. Some coaches just simply fit certain programs and communities, but even then it takes a blend of many factors, most notably time and patience, to translate to success on the field. Beyond that, there's defining what 'success' looks like at a given program. There are many outstanding coaches out there coaching their tails off and not seeing it reflected in the win column for many varied reasons. Again, the explanations aren’t simple as one might think. In any case, coaching changes take time to reveal their impact and with how well, and how quickly, the Elo system is able to adjust to a team's performance, we won't be making any offseason adjustments to team ratings based on coaching changes.
Looking ahead to August…
Last year, in our first season, we tried a whole bunch of stuff to see what worked, what didn’t, and what we could reliably and realistically produce. We jumped right in with a ton of different features and kept adding throughout the season until we nearly ran ourselves ragged. Our plan this year is to try to streamline our offerings so that we bring you high-quality analysis and commentary while also not overloading ourselves or your inboxes.
The main change we’re looking to make this fall is to consolidate many of our weekly feature releases into one main, more traditional, newsletter which will include snippets and links to features like the Weekly Top 25, Model Performance, and more. We’re thinking that the main newsletter will include both the review of the prior week as well as the preview of the upcoming week. We’ll provide more details as we flesh it out and as we get closer to the season.
We’re also coming back with a new season of the Pick ‘Em contest (click to see last year’s contest introduction), featuring a larger weekly set of games to pick from and hopefully much larger participation from our wonderful community.
Last but certainly not least, we’ll be coming back with the Podcast this fall and we’re still planning to release it on YouTube as well as on Spotify and Apple. We had a much better than expected response to the Pod last year and we’re hoping to keep that going strong. This year’s episodes will feature a two-man booth with Jeremy and Josh leading the way and Justin hopefully making some cameos. Be on the lookout for the first episode of the new season, most likely in late July.
Excited to read what you have prepared for this upcoming season. Always enjoy your content and analysis each week. Keep up with the exceptional work you provide. Let's get it on!!!!!
Let's gooo!!! Can't wat for the pick'em contest this year!! Keep up the great work fellas, excited for the new changes and focus!