Have you ever wondered which coaches maximize their talent, which ones do more with less, or which coaches just can’t seem to win despite stockpiling a shit-ton of talent? Are you tired of arguing with OSU fans about whether Harbaugh or Day is the better coach, as if the answer wasn’t already obvious? Do you just like looking at graphs? If you answered ‘Yes’ to any of these questions, then this diary’s for you.

This is a continuation of a previous diary where I compared ‘team talent’ with performance in 2023 to see which teams got more with less or less with more. There was some great feedback and questions in the comments and, since I’m the type of guy that generally aims to please, I thought I’d take a crack at addressing a few of them.

By far, the most common question was what these data would look like over a longer period of time. WestQuad also started a thread after Saban’s retirement wondering what the data could say about his performance over time and whether he was, indeed, the GOAT. 

So my plan in this diary is to combine these two questions by (1) extending the data back to 2015 (the first year available for the 247 Team Talent Composite) and (2) using it to evaluate the success of P5 coaches.

Let’s get started, shall we?

THE MAIN VARIABLES

• 247’s Team Talent Composite is a team-wide composite measure of every player's recruiting rating, accounting for transfers (in and out). I readily admit that this is not a perfect measure of ‘team talent’: recruiting rankings tend to be flawed, development matters, freshman stars are added even though they might not see the field, etc. But it’s the best measure we have, so I’ll be using it.
• FEI’s Win Differential measures a team’s win differential relative to a hypothetical elite team (defined as a team with an FEI score that is 2 standard deviations above the FEI average). Positive values mean that the team had more wins than would be expected from an ‘elite’ team, negative values mean it had fewer wins. 

Putting this all together, here’s what the data look like for all P5 teams from 2015 to 2023. 2020 is excluded for obvious reasons.

• The red shaded area represents ‘elite results’: teams that had at least as many wins as an ‘elite’ team would be expected to have against the same competition. 
• ‘Elite recruiting is represented by the green shaded area (get it?): teams with a talent composite score that is 2 standard deviations above the P5 average for this time period. 
• The dashed lines along the x- and y-axis are the mean talent composite scores and FEI win differentials over the sample period.

The plotted regression line shows the expected/predicted win differential at each corresponding level of team talent (thanks, RJWolvie for recommending this). I calculate actual vs. predicted performance for each year by subtracting the expected win differential (from the regression) from the actual win differential. 

• Positive values indicate that the team had a higher win differential than would be expected based on its given talent. 
• Negative values indicate that the team had a lower win differential than would be expected based on its given talent. 

The graphs for individual years are at the end of this diary. Here’s the graph with the average actual vs. predicted win differential for all P5 coaches from 2015-2023:

Oof. That’s ugly. Let me try that again. To conserve space, I’ll also drop coaches that only have one year of data for the 2015-2023 time period.

That’s a bit better. 

Saban comes in at #11, Harbaugh at #14, and Day at #20 (behind PJ Fleck). I personally wasn’t expecting to see coaches like Mike Gundy, Kirk Ferentz, or Paul Chryst in the top 10, but I guess it makes sense. Lane Kiffin at #9 was also a surprise.

It’s also kind of interesting to compare coaches as they moved from one school to another. Riley at Oklahoma (the good) vs. Riley at USC (the bad) seems about right. Mario Cristobal’s tenure at Oregon puts him within the top-30, but his time at Miami so far has earned him a spot in the bottom-5 of all coaches in the sample. Mike Leach, on the other hand, was pretty successful at both WSU (#4) and Mississippi State (#29). 

In case anyone is interested, here are the same data organized by conference.

Apologies for the messy presentation and the overlapping labels.

Anything you find interesting?

IMPORTANT CAVEATS

There is at least one thing to keep in mind about these data that limit their usefulness. The 247 Team Talent Composite only goes back to 2015, which unfortunately omits the early years of some coaches’ tenures. For example, Jimbo Fisher’s most successful seasons at FSU (2013 & 2014) are not included which probably decreases his average score (his time at FSU is currently in ‘The Bad’ category). Saban had a number of elite seasons at ‘Bama before 2015 (same with Meyer at OSU) that were excluded and may have weakened his overall score. Alternatively, Leach’s 3 worst seasons at WSU (3-9, 6-7, 3-9) are not included in the data. Only a small fraction of Bob Stoops’ 18-year tenure (1999-2016) or Kirk Ferentz’s 25-year tenure (1999-) are included in the sample. Etcetera, etcetera, etcetera.

YEARLY GRAPHS

In case you’re interested, here are the results for each year, 2015-2023.

 

 

 

 

 

 

 

 

Thanks for reading. Go Blue!