KU Win/Loss Projection based on Seniority/Talent Scale

KU PROJECTED TO GO 34-6

Hopefully you are familiar with my famous and highly accurate (TIC) seniority/talent graph. I simply plot the winning % vs. the seniority scale. The seniority scale is the weighted average of the class (freshman - 1, sophomore=2, etc) by minutes played. This has proven to be an incredibly interesting chart over the years. For instance, our 4 most senior seasons are 2005, 2008, 2012 and 2022. Notice a trend. 3 of those 4 seasons KU went to the National Championship game and 2 of them were National Championships. The other was one of the 2 biggest flameouts in the NCAA tourney for Kansas. Seniority is a much bigger factor than many realize. To wit, how many of the super over the top recruiting classes have won a National Championship in the last 15 years. Duke had the top three recruits several years back and still didn't win. Yet Kansas took it's most senior class by far (2012) with what I judge to be slightly below average talent and went to the National Championship game. As I said, of our 4 most senior classes, 3 have been to a NC game. That is another reason I am really liking the recent recruiting classes by KU. We are getting good talent but it is mostly players who will be around for 3 or 4 seasons. A senior Frank Mason is a far better than a freshman Frank Mason and also better than most or all incoming freshmen. Devloping talent is every bit as critical to program success as recruiting the top 10 or 15 players and this seniority scale chart demonstrates that.

Having said all that, I have projected the minutes for every player (WAG at this point). I also judge this team to be a solid above average talent team. Not in the ilk of 2008, 2010 and those teams. But better than last season (talentwise). The seniority scale, based on my projected minutes is 2.42. If that pans out that will make this team a thoroughly middle of the road seniority scale team. Not young or old but a good mixture of all levels. And if they are on a par with the 2016 and 2017 teams talentwise, then the chart projects them to win 85% of their games which is 34-6 for a 40 game schedule.

This definitely is an interesting graph, Corpus. Thanks, as always, for putting in the work to create it.

I see you shared where the senior scale comes from (win % is straight data). Can you just remind me/us how how you determined the "talent" metric.

Thanks,
Port

Port, that is a great question but one that is a little tricky to answer. The quick answer is that the iso-talent lines just emerge from the data. Let me see if I can explain. The data points themselves are easy to understand. It is the calculated seniority scale on the x axis and the winning percentage on the y axis. All hard data. So the Logic is that a more talented team will win more games. If you assume that experience has zero impact on winninG then the iso-talent lines would be horizontal. But experience demonstrably impacts winning. The question is how much. So I went back for all KU teams and plotted win% vs experience (seniority scale) to try to see if I could get a quantifiable impact of experience. The next thing I did is I took every KU player and calculated their player rating by year to see if I could get a quantifiable value of improvement from year to year. I did this back to 1989. The combination of these the evaluations led me to use around an 8% impact per year of experience. In other words, a team of all freshmen might have a win% of 70%. That same team a year later would win 8% more games or a winning percentage of 75.6%. As juniors they would win around 81.5%. So now I haVe essentially defined the slope of the iso-talent curves. Now I just draw lines with 8% per year slopes through the beat team (2010) and that is the top talent line>. Draw that same slope line through the worst team (2019) and that is the worst talent line. Then I just equally space lines for below average, average and above average. That actually sets the entire chart. But just to give it a reasonability check, I went through each team and gave it a score of talent (mostly subjective but used things like HS ratings and draft status). When I did that, pretty much all of the teams seemed to be close to where I would put them especially considering the accuracy of the win% data. I judge this graph to be pretty useful. I predicted we would go 33-7 last season. But this team turned out to be a little more talented than I initially judge.

Hope this makes sense.

This is actually very helpful, Corpus. Thanks for taking the time to explain.