FanPost

Evaluating UW's Recruiting Using Expected Outcomes

It is that time of year when there are stories about how ‘stars matter' in recruiting and such. This is my take on the subject for this year. Despite the use of a bunch of math throughout this fanpost, this in no way should be viewed as anything other than a ‘quick-and-dirty' analysis.

This year among Husky fans there is more than a little grumbling about how low UW's recruiting class is rated. We know that both Lake and Petersen have said that they don't care about those recruiting ratings from the recruiting services. But we also know that UW does offer recruits that are rated highly on those recruiting sites-along with ones that are not rated as highly. So I wanted to look at the question of how closely UW's rating of recruits matches the recruiting services. The question is: Should we trust Lake and his staff when it comes to recruiting? (The answer is 'yes', and this post will explain why.)

We know that UW does have some prioritization even among recruits that it offers. We also know that they aren't going to offer a recruit unless they think that they can play at a Power-5 level. And there is no way that I (or likely any other Husky fan) is going to get their hands on UW's actual recruit ratings. So, how do we evaluate or compare their ratings with the recruiting services?

The way I decided to do it was to look at comparing outcomes; in other words, answer the question: How do the results of the players that UW signed compare to the expected outcome for the recruits in the rankings with similar ratings?

It is well established that the more ‘stars' a recruit has in high school, the more likely they are to get drafted into the NFL. There is plenty of data which breaks down the odds by the number of ‘stars' a recruit has. My thought was to compare UW's ability to put players in the NFL with that data, comparing the percentage of UW recruits at various ‘star' levels. The hypothesis that I have is that if a larger percentage of UW's recruits make it to the NFL compared to the established levels for all players, then UW's (Petersen, Lake, and the rest of the coaches') ratings for their recruits must be higher (better) than the recruiting services.

NFL Draft Projection of Recruits

The first step I took was to look at a 'baseline', that is the amount of recruits that would be expected to be drafted based on their recruiting ratings. There are a lot of places that have data on the percentage of recruits at different ‘star' levels that are drafted. The values that they have vary, likely because of the data that they use. The one that I decided to use was from a paper submitted to the Journal of Statistics and Applications called "Statistical Modeling of Success in College and NFL for a Star-Rated Football Recruit". Besides having some academic rigor (not normally found in the normal on-line article on football recruiting), the paper developed equations which model success. This allowed me to get some granularity that I couldn't find in other articles (as you'll see later). As it mentions in the paper, they have based their recruiting data on Rivals, which is not normally the source I would use for recruiting data. But while there may be differences with the 247Sports Composite, I assumed that the aggregate data would still be useful.

In the paper they developed 2 different equations for the probability of a recruit making the NFL as a function of their ‘star ranking'. (I won't bore you with the equations. You can check the article if you are interested.) I decided to use the one that they thought was a more realistic representation. Here is the data in tabular form showing the percentage of recruits drafted by the NFL depending on the number of ‘stars' they had as recruits.

Stars

NFL Draft %

5

55.5%

4

29.9%

3

5.4%

2

0.3%

While you may quibble with the exact numbers, I think that most fans would agree with the trend shown in that table. A very small percentage of ‘low 2-star' recruits would be expected to be drafted while most ‘5-star' recruits would be (and most of them in the first round or two).

What can make that data a bit confusing to some fans is that the number of recruits with 2 stars is much higher than the ones with 5 stars, so the absolute number of recruits with 2 stars that get drafted may seem high (especially when they sometimes stand out because they weren't highly recruited); still, the percentage is quite low. As an example, if there are 600 recruits with a 2-star rating, 18 could be expected to be drafted while there could be only 16 or 17 of the 5-star recruits drafted; so, more total 2-star's drafted than 5-stars, but still a low percentage of them.

UW's NFL Draft Results

That established the ‘baseline' for all recruits. The next step was to look specifically at UW's recruits.

I looked at all of the recruits that have been recruited since Petersen arrived (class of 2014) and have completed their time at UW, which goes through the class of 2017. (I'm not including those who are still on the team or have said that they are making themselves eligible for the 2021 draft-guys like Molden and Tryon.) I have included recruits that were not originally given scholarships but earned them since giving them scholarships was likely justified like Myles Bryant. I ended up with 60 recruits on the list. I then looked at their rating on Rivals. A few were never rated on Rivals (Will Dissly, Ezekiel Turner, Tony Rodriguez, Jared Pulu, and Ryan Bowman); so I gave them the low score of 5.2 (low 2-star). Finally, I recorded how many of each were drafted into the NFL. The following table shows the number of recruits with the different 'star' ratings and how many of each were drafted.

Stars

Scholarships

Drafted

5

0

0

4

10

3

3

34

7

2

16

3

According to my data, 13 of the 60 UW recruits that had scholarships while at UW were drafted. I will note also that at least 8 others were on NFL teams as of the 2020 NFL season, including Salvon Ahmed, Myles Bryant, and Tristan Vizcaino, who all saw NFL action in 2020. But since the equation only looked at recruits that were drafted, not just made teams, I did not include those as being ‘drafted'.

The bulk of the recruits in this analysis were in the ‘mid-to-low 3-star' rating. A big part of that is due to the fact that most of the higher-rated recruits came in later during the Petersen years and have not finished up at UW.

Comparison

Now we can compare the actual percentage drafted with the percentage predicted based on that equation for recruits in general. Here's a look at the comparison.

Stars

Scholarships

Drafted

Draft %

Predicted Draft %

5

0

0

0%

55.5%

4

10

3

30%

29.9%

3

34

7

21%

5.4%

2

16

3

19%

0.3%

UW did not had any 5-star recruits during that period, but that will be changing over the next few years. UW was right on the mark with the percentage of 4-star recruits drafted so far. But, the ability of UW recruits with lower ratings to be drafted is much higher than for the 'baseline' for all recruits. This was true with both 3-star recruits, but even more so with 2-star recruits. By the way, the three 2-star recruits that have been drafted were: Drew Sample, Will Dissly, and Nick Harris. (Harris was a 2-star on Rivals, but a 3-star elsewhere.)

Comparison by Rivals Rating

‘Stars' are a very coarse system for rating players. Some players rated a ‘3-star' may be very close to being a ‘4-star' while others may be just barely better than a ‘2-star'. So the next question I looked at was whether some of the ratings of UW's recruits were just off by a little, and whether that could explain some of the better performance. In other words, if all of those 3-star recruits were just below being a 4-star, that may explain why UW's performance with '3-star' recruits is so much better.

As I mentioned above, I liked having an equation because it allowed me to achieve some additional granularity. The data that was used in that paper to develop the equation did not use the level of granularity that I'm trying to get from it, but I decided to use it like that anyway. As a result, the values that I got out of it for this next comparison may not be as accurate (skewed higher). But, there is a mitigating factor. The data in that paper looked across all of college football. In the paper they admit that the rates for schools in the Power-5 conferences is probably greater than that of all of college football. (A 3-star at a Power-5 school would have a higher likelihood of being drafted than a 3-star at a FCS school.) So, while the following numbers may not be completely accurate, I believe that they are at least reflective of the expectations for a Power-5 team, like UW.

Rivals uses a numeric rating system that, for most recruits, goes from 5.2 to 6.1 with a 5.2 being a ‘low 2-star' and a 6.1' being a ‘5-star'. The additional numbers in the ratings allow them to differentiate, for example, between a ‘high 4-star' and a ‘low 4-star'. The table below shows the ratings from Rivals along with the expected percentage of players at each rating to be drafted into the NFL. To get the additional values, I used values between the number of stars (not just the integers) and plugged them into the equation. For example, for a ‘high 4-star' I used a value of ‘4.6' in the equation. (This was the advantage of having an equation-it was easier to get this interpolation.)

Rivals Rating

Stars

NFL Draft %

6.1

5

55.5%

6

4.6

47.6%

5.9

4.3

39.3%

5.8

4

29.9%

5.7

3.6

17.6%

5.6

3.3

10.3%

5.5

3

5.4%

5.4

2.6

1.9%

5.3

2.3

0.8%

5.2

2

0.3%

You can see that the percentages are higher (much higher in some cases) for some recruits; for example a rating of 5.4 now is 1.9% where it was just 0.3% before because it was grouped in with the rest of the recruits that were 2-stars. But, as I wrote above, since I will be using these numbers as a comparison for a Power-5 program (UW), I thought the higher numbers were justified.

Now let's look at the comparison to UW by looking at the Rivals ratings.

Rating

Stars

Scholarships

Drafted

Actual Draft %

Predicted Draft %

6.1

5

0

0

-

55.5%

6

4.6

2

2

100%

47.6%

5.9

4.3

3

0

0%

39.3%

5.8

4

5

1

20%

29.9%

5.7

3.6

6

1

17%

17.6%

5.6

3.3

13

4

31%

10.3%

5.5

3

15

2

13%

5.4%

5.4

2.6

9

1

11%

1.9%

5.3

2.3

3

1

33%

0.8%

5.2

2

4

1

25%

0.3%

One thing that probably stands out is that none of three recruits with the 5.9 rating were drafted. Why weren't any of them drafted? Here's the list:

  • Jake Browning
  • Salvon Ahmed
  • Hunter Bryant

All three of them were on NFL teams in 2020: Jake on a practice squad, Hunter played in one game after battling some injury, and Salvon even started for the Dolphins for a time while Gaskin was out. If either Salvon or Hunter had been drafted (very surprising that neither was), the percentage drafted for UW would have been very close to the predicted amount.

The two ‘high 4-star' recruits were Budda Baker and Byron Murphy. So, every one of the recruits that UW has signed with a rating of 5.9 or higher is in the NFL (or at least was for the 2020 season).

Of the ‘low 4-star' recruits (5.8 rating), only Kaleb McGary was drafted, but Benning Potoa'e is in the NFL. Two of the other 3 had to give up football due to injuries: Austin Joyner and DJ Beavers. (KJ Carta-Samuels was the other recruit with a 5.8 rating.)

Of the ‘high 3-star' recruits (5.7 rating), the only one that has been drafted is Myles Gaskin (and that was in the 7th round). One that would have been drafted if he hadn't been injured is Trey Adams. There are, however, several ‘high 3-stars' which should be drafted eventually including Levi Onwuzurike, Jaxson Kirkland, and Cade Otton. So while the percentage of NFL drafts for a 5.7 rating may be just below the average now, it could surpass the average soon.

Looking at the table, it should be clear that UW has placed the ‘lower-rated' recruits (‘mid 3-star' and lower) into the NFL at a FAR GREATER rate than college football teams in general. UW was not just taking advantage of recruits that were just slightly under-rated. The expectation from the Rivals ratings for these recruits is vastly different from what actually happened at UW.

Projecting the Future

One could argue that the sample size so far is relatively small. Maybe UW's coaches just got lucky. So I looked at some of the other recruits from the 2016 and 2017 classes who should be leaving UW in the next year or two to see if the NFL draft rates could hold up.

There are 28 additional recruits in this list-those from the recruiting classes of 2014 through 2017. Of those, there are several that could be confidently projected as being drafted:

  • Levi Onwuzurike (3-star)
  • Elijah Molden (4-star)
  • Henry Bainivalu (4-star)
  • Keith Taylor (4-star)
  • Jaxson Kirkland (3-star)
  • Joe Tryon (3-star)
  • Cade Otton (3-star)

There are a few others that could be as well, like Terrell Bynum, Luke Wattenberg, and Josiah Bronson. (Bronson was not originally recruited to UW, but he did get a scholarship after transferring from Temple, so I'm including him.) However, if just those 7 get drafted (and none of the other recruits), here is the comparison:

Stars

Scholarships

Drafted

Projected Draft %

Predicted Draft %

5

0

0

0%

55.5%

4

21

6

29%

29.9%

3

48

11

23%

5.4%

2

19

3

16%

0.3%

Again, the 4-star recruits from UW match the predicted number very closely, while the numbers for the 3-star and 2-star recruits continue to be much higher. And these numbers could go up even higher if even more are drafted.

What does this mean?

In the four years' worth of recruiting that I analyzed, UW showed a good ability to get recruits into the NFL. (Maybe someone can compare UW's ability with other Power-5 programs.) Their ability to get lower-rated recruits into the NFL has been exceptional. If you are a 2-star recruit that signs with UW, your likelihood of being drafted is about the same as a ‘high 3-star' recruit in other programs (16% vs 17.6%), and a 3-star recruit has a similar chance compared to a 4-star recruit in other programs (21%, vs 29.9%). Based on this data, nobody who signs with UW should be considered rated lower than a ‘mid 3-star' (5.6 Rivals rating) and I would argue that we should consider every UW signee to be the equivalent of at least a ‘high 3-star'.

I did a little experiment where I increased the Rivals ratings on all of UW recruits (from 2014 to 2017). For those with 4-stars, I increased their Rivals rating by 0.1 (so, for example, instead of a 6, they became a 6.1, meaning a 5-star). For 3-star recruits I increased their rating by 0.2 (so 5.6 became 5.8, a 4-star). And for 2-star recruits, I increased by 0.3 (5.4 became 5.7, a 'high 3-star'). I made those changes and added in the recruits that I had projected to be drafted. Here are the results:

Adjusted Stars

Scholarships

Projected Draft

Projected Draft %

Predicted Draft %

5

2

2

100%

55.5%

4

48

13

27%

29.9%

3

38

5

13%

5.4%

You can see that these numbers match the predictions much better across the board-higher for 5-star and 3-star recruits, and slightly lower for 4-star recruits.

So for those complaining about the lack of 4-star commits in the 2021 class, look on Rivals and add 0.1 to the ratings of the 4-stars, 0.2 to the 3-stars, and 0.3 to the 2-stars to get a better assessment of this class. What you find is that UW really has the equivalent of about ten 4-star recruits out of the 15 commits instead of 5 (plus one 5-star).

I know that a lot of Husky fans look at the 247Sports Composite ratings when looking at recruiting. Even though the ratings for any single recruit may not be the same, I don't see any reason why the same methodology could not be applied to the ratings there. So I came up with this equation to make the adjustment between original rating and the adjusted rating based on UW's recruiting ability:

Original Rating + ((5 - Original Stars) x 0.0333) = Adjusted Rating

If you do that you find that every one of UW's commits for the 2021 class should be considered a 4-star, with the exception of Sam Huard, who is already a 5-star. (And yes, that includes Siaosi Finau whose adjusted composite rating becomes 0.8947.)

Concluding Remarks

Player development is obviously another factor in getting some of these recruits into the NFL. How much of the success can be attributed to that versus the initial recruiting assessment is hard to know. However, projecting development in a recruit is something that the coaching staff tries to do-not just look at where a recruit is while they are being recruited, but where they will be when they are a senior at UW. To that extent, I think that it is still fair to say that UW's recruiting evaluation is significantly better than the recruiting ratings.

It isn't just success at getting recruits drafted. Another way to view success from a recruit is with All-American and All-Conference selections. UW has also shown the ability to get their recruits on All-American and All-Conference teams as well. Most of the UW players that were drafted also made it on All-Conference first or second teams at some point in their career at UW. Some, like Jake Browning, made All-Conference without being drafted; others did both. I couldn't find a good model for comparing ‘stars' with ‘All-Conference' selections, but I don't see anything in UW's ability to get players on ‘All-Conference' teams that would indicate anything different than we see with getting players drafted: namely that UW does a very good job of getting players on ‘All-Conference' teams, even ones that may not have been highly-rated as recruits. Guys like BBK, Drew Sample, Greg Gaines, Dante Pettis, Will Dissly, Myles Gaskin, Sidney Jones, Nick Harris, and Taylor Rapp were not highly rated out of high school, but all made ‘All-Conference ‘ teams at least once in their career at UW.

It is hard to say how long this success at getting recruits to the NFL will last since there is starting to be a lot of coaching turnover. We can hope that with Jimmy Lake coming up from under Petersen (and some of the staff still in place) that success will continue to at least some degree.