*Note: The visualizations featured in the article are intended to be viewed on a tablet/desktop. If you are reading this on your phone, come back to it when you get home and can look on a larger screen. Thanks.*
Way back before I had even started, and I guess also now finished, my tenure with the UW Dawg Pound I put together a database on players using the data from 247 Sports. I downloaded the information about every player in their rankings each season beginning with the 2014 recruiting class (Chris Petersen’s first) as well as who they had offers from in order to be able to answer questions like “when’s the last time UW beat out Texas for a prospect from outside the PNW?”
Then I went in and started adding information about each player’s college career to try to assess how well they were being developed. The initial version included NFL Draft status as well as All-American and All-Conference picks. I put that all together into one career score metric.
Something was missing though. What about the players who were backups as sophomores and juniors then started as a senior but didn’t quite make it to an all-conference level? Those careers shouldn’t be viewed as a complete zero. So I went back in and had to chose to manually add in games started and snap count data to differentiate players who never saw the field from ones who did but never rose above merely good college players.
Oh yeah, plus now I have to track the player movement of thousands of players every offseason to keep it up to date. Still wondering why I don’t have time to write full-time anymore? (I mean there’s also the whole being a dad thing but this is obviously more important).
The end result is now my new and improved College Football Talent Development dashboard which can be found on Tableau Public right here.
I’ll go through each of the features here to show what kind of cool questions it’s possible to answer with the tool and hopefully you’ll enjoy killing a bunch of time looking at it like I do. If you have the chance to open it up in a separate browser window and go back and forth, I promise it will look a lot better.
% of P5 Recruits
The % of P5 recruits tab may need to be renamed with the death of the Pac-12 (RIP) but I’m still including Oregon State and Washington State until further notice since a lot of the features are about looking back historically. However, the first tab shows you the crux of the stars matter argument.
From the 2014-18 classes: 26.1% of 5-star recruits made an AP All-American team, 51.6% of them made at least 2nd team all-conference, and 61.8% of them were drafted. Compare that to 5.9% of 4-star recruits becoming All-Americans, 18.1% making all-conference, and 24% getting drafted. The 5-star recruits hit those benchmarks 2.5-4x more often than the 4-stars. The difference between 3-star and 4-star recruits (more like 2x) is slightly less pronounced but still quite evident.
I’m a heavy proponent of re-doing the star system to make the divisions more meaningful. It’s more true in college basketball than football but the current boundaries between 2, 3, and 4-star prospects could be tweaked. Almost everyone these days is either a 3-star prospect or unranked. I understand a 17-year old’s feelings probably get hurt calling him a 1-star prospect. But how about using a 3 to 10-star system so no one is below a 3 but we more accurately portray how wide the range is between low and high 3-stars in the current system? A man can dream.
Let’s say you want to see how those numbers change at certain positions. Or maybe you want to see if it has changed over time. On the right there are filters for recruiting class as well as recruited position so you can customize if you want.
Then on the bottom there is the list of total career score by school. The default is set to 2014-18 since those are the classes that are (almost) entirely done with their careers. That also corresponds with most of the Chris Petersen era at Washington. You can see that Washington was 6th in the entire country in career score during that time period and only Alabama, Ohio State, Georgia, Clemson, and Oklahoma had players be more productive in that 5 year span.
Career Score Performance
So we know that Washington had players perform really well. We also know from having lived through it that the Huskies’ recruiting classes weren’t nearly as highly touted in the moment as those other schools I named. How do we assess how well they developed talent compared to their peers?
On to the Career Score Performance tab. This tab takes into account how highly rated a prospect was coming into college using the 247 Sports Composite rating. For each position and recruiting rating I assigned an expected career score total using blended historical data. I got granular enough to the hundredth of a point (e.g. 0.99, 0.98, 0.97, etc.) For those unfamiliar with the 247 system, the top-32 recruits are 5-stars which usually translates to between 0.985 and 1.0. Anyone below that but with above a 0.89 rating is considered a 4-star. And just about everyone below that who is recruited at the power conference level is a 3-star.
So for every prospect there’s what you would expect their career score to be based on their recruiting rating and position, then what their career score has ended up being (current prospects have it listed up to this point but top-level ones will get bumps when they eventually get drafted). I took the expected career score and subtracted it from the actual career score in order to end up with a performance compared to expected. Add that up for each individual player and you can see which programs were the best at developing talent over a certain period of time.
Once again, there are year and position filters so you can choose to look at the period you choose. For instance, set the years to 2014-2021 and the position to just WR. You’ll see that Oregon and Florida State are in a tight competition for who does the worst job developing wide receivers while the Huskies are in 2nd place in the new Big Ten behind only Ohio State.
The Buckeyes though are in just 3rd place despite their reputation because almost every receiver they recruit is expected to be really, really good. Wake Forest and Iowa State are just ahead of them. Iowa State has had 5 all-conference receivers with 2 drafted in that time despite most being unheralded while Wake Forest had nearly 3x as much of a career score total than you’d expect based on how lowly recruited those receivers were.
Current Roster
Want even more detail but in a table form instead? The career score reference tab will show you how many expected All-American, All-Conference, and NFL Draftees each program should have had based on your year and position filters and then the actual totals. Washington was expected to have 6 All-Americans, 20 all-conference players and 23 draft picks from their 2014-20 classes. Instead, Washington had 14, 34, and 32 respectively to well outpace those totals in every category.
Let’s say you want to see who DB U was during that 7-year timeframe for classes that Chris Petersen recruited. Well set the draft position to CB and S and the years from 2014 to 2020 and see that Washington DBs had the 2nd highest actual career score in that time behind only Alabama (and barely ahead of Ohio State). And sort by career score performance and Washington was #1 in development during that time ahead of Iowa, Illinois, and Utah.
Current Team Talent
Now we move to what I‘ve been doing lately. Namely, updating the current rosters for every power conference team. The first current roster view takes a look at two parts. How well a team has recruited (the sum of their expected career scores) and how much the players on that team have actually accomplished (the sum of their actual career scores so far). I set both axes to show the highest scoring team right now as nearly 100.
It’s fun to see that some programs that are closely tied show up as so similar. Texas and Oklahoma are the closest to one another with Texas having a slight edge in both talent and experience this year. Miami and Florida State are also nearly neck and neck in that upper right quadrant. LSU hired away Notre Dame’s coach but their logos nearly overlap with average experience and above average roster talent.
Husky fans may be disheartened to see that this view isn’t very favorable to Washington. The Huskies are in a similar neighborhood to the academic powers with Stanford, Vanderbilt, Northwestern, and Duke among the closest schools. If Washington ends up underperforming this season then this graph will be something to point to as to why we should have seen it coming. If Washington ends up being a top-25 to 35 program then it’ll be easy to shrug it off as a weird artifact of the irregular situation of a team replacing almost their entire roster.
If we do put stock into this graph then it shows that Ohio State and Oregon both are clear national title contenders while Georgia and Alabama have more raw talent than either but there is a lot more emphasis on the raw part of that phrase (particularly at Alabama which is even behind Washington in experience). It also says watch out for Oklahoma State as a real contender in the Big 12 and that FSU/Miami will likely duke it out for the CFP berth from the ACC.
Current Team Talent
If you’re still with me then you probably read at least some of the my Team Talent Preview series for the entire Big Ten conference. I figured out how to automate the formula and so have done the same calculation now for every power conference team (although I didn’t go through the other conferences with quite the same fine-toothed comb so I’m sure I missed a player or two here or there). There are two parts to this tab.
The first is similar to the last tab, a graph to try to assess where each program stands on their current roster. The X-axis is the team’s defense and the Y-axis is the team’s offense. The previous tab looked at every player I have listed on their roster but this one tries to weight for who the starters are going to be so it doesn’t care if your 4th string QB was a four-star recruit. The starter is weighted more heavily than the backup while the true depth pieces don’t count into the formula. You can see that there’s a strong correlation between a team’s offense and defense and most programs are along a pretty tight diagonal. Teams don’t usually recruit that much better on one side of the ball than the other.
Washington is skewed a little more towards defense this season (not surprising since they aren’t returning any offensive starters) and this view has them closer to the middle of the pack among power conference teams. Michigan and Ohio State though are clear outliers in terms of teams that project to have significantly more talented defenses than offenses this season. Meanwhile, Georgia is on the other side of that equation with the most offensive talent in the country followed by fellow SEC schools Alabama, Texas, and Texas A&M.
Looking for some sleepers? How about SMU in the ACC who finishes above Florida State or Miami (the Hurricanes are practically hidden behind Florida State as they are once again almost tied). Their numbers were accumulated in the AAC which probably over-inflates them and it may take a year to correct that but they’re intriguing and loaded at wide receiver. Florida was another shocker given how poorly they fared last year and their brutal schedule means most are expecting Billy Napier to get canned, with Jedd Fisch as a prime replacement candidate. We’ll see.
The second part of the tab allows you to see the rankings filtered by school and/or position. The rankings are out of 100 but QBs get 1.5x weight due to their outsized importance on the field so the first 91 names are all QBs until you get to Travis Hunter at Colorado. The rest of the top-non-QBs? LSU OT Will Campbell, Ohio State S Caleb Downs, Ohio State RB TreVeyon Henderson, Ohio State WR Emeka Egbuka, Ohio State ED Jack Sawyer, Michigan CB Will Johnson, and Arizona WR Tetairoa McMillan. So yeah, Ohio State is kind of loaded. And it would’ve been really nice if T-Mac had opted to follow the coaching staff to Seattle.
******
If you have suggestions about other views you’d like to be able to see or questions about the data, feel free to drop it in the comments below.
