League Wide Draft Success 2017-2019 - A Follow Up

[GunnerBill]

2 hours ago, JGMcD2 said:

@KzooMike

 

In comparison to Buffalo, this is New England (2017-2019 as I just grabbed it quick). The worst drafting team over the last 4 years (#32). They've made the playoffs as often as Buffalo has, extremely competitive team. They actually have a better record over the last 4 years than we do, they're actually the 4th most winning team over that time span. Belichick has done an absolutely horrible job of finding any sort of value consistently. He pretty much gets 1 contributor each year, but a lot of his early picks are just horrible values. He's done better with guys like Wise (4th), Bentley (5th), Winovich (3rd) than he has with anything in earlier rounds. 

 

 

I'll throw in GB (#23) seeing that they've been in the conference championship the last two seasons. That's a lot success... out of the team's listed starters on Sunday, they had 8 drafted between 2017-2019. 

 

 

Houston (#13). Generally do a pretty good job finding some value, but really propped up by Watson. Rightfully so, he's the most important draft pick they have made and most important player on the field for them each week. 

 

Any system that only rates Kevin King as a -1.6 is flawed. He is worth about that many points to the opposition each week. Equally Jaire Alexander is only +1.0. 

 

That just doesn't come close to passing the eye test. One of them is a top 10 corner and one of them is among the worst starting corners in the NFL. 

[JGMcD2]

7 hours ago, GunnerBill said:

 

Any system that only rates Kevin King as a -1.6 is flawed. He is worth about that many points to the opposition each week. Equally Jaire Alexander is only +1.0. 

 

That just doesn't come close to passing the eye test. One of them is a top 10 corner and one of them is among the worst starting corners in the NFL. 

I really enjoy when people don’t read and make assumptions as to what they’re looking at. THIS ISN’T RATING HOW GOOD A PLAYER IS. 
 

It’s rating them in comparison to the value in the round they were drafted. It’s a tool to grade GMs extracting value. 
 

Look at it this way... the 2018 first round was very talented. Jaire Alexander was a great pick, nobody is debating that, he netted them positive value. At this point he’s just not THAT much more valuable than the average player from the 1st round in 2018... because there was that much talent. 
 

Kevin King was a bad pick, he netted the Packers negative value in the round. The talent in the 2nd round in 2017 was a mixed bag, all it’s showing you is that Packers drafting King there didn’t really do all that much worse than any other team in finding value in the 2017 2nd round. 
 

Yeah, it’s flawed. Trying to quantify this isn’t easy and will never be perfect. It’s an attempt at something I don’t think I’ve ever seen someone try and do publicly. If you actually read what it was doing before commenting, it would be helpful. 
 

As I’ve been saying to everyone who says it’s flawed, help me out and provide some feedback as to how to make it better. 

[GunnerBill]

3 minutes ago, JGMcD2 said:

I really enjoy when people don’t read and make assumptions as to what they’re looking at. THIS ISN’T RATING HOW GOOD A PLAYER IS. 
 

It’s rating them in comparison to the value in the round they were drafted. It’s a tool to grade GMs extracting value. 
 

Look at it this way... the 2018 first round was very talented. Jaire Alexander was a great pick, nobody is debating that, he netted them positive value. At this point he’s just not THAT much more valuable than the average player from the 1st round in 2018... because there was that much talent. 
 

Kevin King was a bad pick, he netted the Packers negative value in the round. The talent in the 2nd round in 2017 was a mixed bag, all it’s showing you is that Packers drafting King there didn’t really do all that much worse than any other team in finding value in the 2017 2nd round. 
 

Yeah, it’s flawed. Trying to quantify this isn’t easy and will never be perfect. It’s an attempt at something I don’t think I’ve ever seen someone try and do publicly. If you actually read what it was doing before commenting, it would be helpful. 
 

As I’ve been saying to everyone who says it’s flawed, help me out and provide some feedback as to how to make it better. 

 

I know what it is trying to do. It still doesn't pass the sniff test. If a "great pick" can only net you +1.0 then the methodology surely has to be flawed. 

 

We had this discussion before..... to me it comes down to three criteria: playoff games won by drafted players, pro bowl selections (excluding alternates) by drafted players and all pro votes (not just selections but votes) by drafted players. That is what I am interested in and then I think of those three criteria I would weight the all pro votes heaviest in my assessment. 

[JGMcD2]

27 minutes ago, GunnerBill said:

 

I know what it is trying to do. It still doesn't pass the sniff test. If a "great pick" can only net you +1.0 then the methodology surely has to be flawed. 

 

We had this discussion before..... to me it comes down to three criteria: playoff games won by drafted players, pro bowl selections (excluding alternates) by drafted players and all pro votes (not just selections but votes) by drafted players. That is what I am interested in and then I think of those three criteria I would weight the all pro votes heaviest in my assessment. 

Well let’s look at Jaire Alexander over his first 3 years in the league by PFF grade 

 

2018: 72.4

2019: 72.3

2020: 90.5

 

Here’s his AV over those 3 years

 

2018: 5

2019: 6

2020: 10

 

While he may be a shutdown corner now, he wasn’t playing like one all 3 years in the league. He’s been better than the average player in the first round of his draft class... you have to remember this is also just a snapshot in time as well. Things will change. If he continues to be a shutdown corner, they will continue to gain more value. At this point, he hasn’t provided that much more value than the average player drafted next to him in the first round. 

I am really glad that you listed your three criteria! Approximate Value incorporates both All-Pro selections and Pro Bowl selections when grading each player. It also weights All-Pro selections greater than it does Pro-Bowl selections. And while it doesn’t have the capability to include playoff games won (personally that’s an unstable number to throw in) it does reward players for playing/starting on better teams.  Sounds like AV is pretty darn close to your criteria! 

 

Edited January 27, 2021 by JGMcD2

[GunnerBill]

17 minutes ago, JGMcD2 said:

Well let’s look at Jaire Alexander over his first 3 years in the league by PFF grade 

 

2018: 72.4

2019: 72.3

2020: 90.5

 

Here’s his AV over those 3 years

 

2018: 5

2019: 6

2020: 10

 

While he may be a shutdown corner now, he wasn’t playing like one all 3 years in the league. He’s been better than the average player in the first round of his draft class... you have to remember this is also just a snapshot in time as well. Things will change. If he continues to be a shutdown corner, they will continue to gain more value. At this point, he hasn’t provided that much more value than the average player drafted next to him in the first round. 

I am really glad that you listed your three criteria! Approximate Value incorporates both All-Pro selections and Pro Bowl selections when grading each player. It also weights All-Pro selections greater than it does Pro-Bowl selections. Sounds like AV is pretty darn close to your criteria! 

 

 

None of that explains a 2.6 difference between one of the best starting corners in the league and one of the worst. 

 

Good it sounds like it is some way there but it obviously includes other metrics as well. Saying Kevin King isn't that bad of a pick because a lot of other teams made bad picks in the 2017 second round is weak to me. I can count 14 players who went after him in round 2 who have been without question better pros. That is nearly half the round. 

[JGMcD2]

24 minutes ago, GunnerBill said:

 

None of that explains a 2.6 difference between one of the best starting corners in the league and one of the worst. 

 

Good it sounds like it is some way there but it obviously includes other metrics as well. Saying Kevin King isn't that bad of a pick because a lot of other teams made bad picks in the 2017 second round is weak to me. I can count 14 players who went after him in round 2 who have been without question better pros. That is nearly half the round. 

It includes everything you want. It’s publicly available, you can read about it. Probably would’ve helped to understand what was in it before you blasted me and then ended up telling me your criteria is pretty much the exact same thing. 
 

The difference is that they were in two separate drafts. It’s all about GMs extracting value based on what is available to them in the player pool that year. 
 

It’s not saying Kevin King isn’t a bad pick, he’s a bad pick. He was rewarded a bit for having a decent season in 2019, albeit it’s probably a career year. It’s just that other GMs as a whole didn’t do a very good job of extracting value in that round, so Green Bay isn’t going to be punished severely when other teams missed as well. So yeah, you have half the round that are better pros than King, but half the round are just as bad, if not a little worse in terms of their return on value.  All it’s doing is rewarding teams for getting value or punishing them for not. In this case they’re punished. 

 

It was a easier to miss in 2017 in RD2, so the Packers are punished, but not severely, based on how King has performed compared to his peers in that round thus far. 
 

It was harder to miss in 2018 in RD1, so the Packers are rewarded, but not significantly, based on how Alexander has performed compared to his peers in that round thus far. 
 

If they continue to go on the paths they’re going, Green Bay is only going to continue to be rewarded by Jaire Alexander and penalized by Kevin King. Again, it’s a snapshot. This is living and breathing as the players continue to play. You’re focusing too much on the 2020 version of Jaire Alexander and 2020 version of Kevin King. 
 

I’ve said it a million times, this isn’t perfect. Not even close... it’s my first crack at something that nobody else has attempted. 

Edited January 27, 2021 by JGMcD2

[GunnerBill]

4 minutes ago, JGMcD2 said:

It includes everything you want. It’s publicly available, you can read about it. Probably would’ve helped to understand what was in it before you blasted me and then ended up telling me your criteria is pretty much the exact same thing. 
 

The difference is that they were in two separate drafts. It’s all about GMs extracting value based on what is available to them in the player pool that year. 
 

It’s not saying Kevin King isn’t a bad pick, he’s a bad pick. He was rewarded a bit for having a decent season in 2019, albeit it’s probably a career year. It’s just that other GMs as a whole didn’t do a very good job of extracting value in that round, so Green Bay isn’t going to be punished severely when other teams missed as well. So yeah, you have half the round that are better pros than King, but half the round are just as bad, if not a little worse in terms of their return on value.  All it’s doing is rewarding teams for getting value or punishing them for not. In this case they’re punished. 

 

It was a easier to miss in 2017 in RD2, so the Packers are punished, but not severely, based on how King has performed compared to his peers in that round thus far. 
 

It was harder to miss in 2018 in RD1, so the Packers are rewarded, but not significantly, based on how Alexander has performed compared to his peers in that round thus far. 
 

If they continue to go on the paths they’re going, Green Bay is only going to continue to be rewarded by Jaire Alexander and penalized by Kevin King. Again, it’s a snapshot. This is living and breathing as the players continue to play. You’re focusing too much on the 2020 version of Jaire Alexander and 2020 version of Kevin King. 
 

I’ve said it a million times, this isn’t perfect. Not even close... it’s my first crack at something that nobody else has attempted. 

 

I really am not. I don't buy that King was good in 2019. He had a few picks but that was because people were throwing at him a lot. He still gave up a hell of a lot of plays. Nor do I buy that the 2nd round of 2017 was a poor round. I dispute that as well. It wasn't. It just doesn't pass the sniff test. I get what you are trying to do, I don't think it is worthless and it is an indicator, but it has too many funky results like this one for me to consider it anything more than an indicator. I would not use it to draw any general conclusions about drafting success. There is no logic by which you are ever going to get me to accept that Jaire Alexander and Kevin King are within 2.5 points of each other in terms of success as pros. They are miles apart. 

[JGMcD2]

21 minutes ago, GunnerBill said:

There is no logic by which you are ever going to get me to accept that Jaire Alexander and Kevin King are within 2.5 points of each other in terms of success as pros. They are miles apart. 

Well I am not trying to get you to accept that Jaire Alexander and Kevin King are within 2.5 points of each other in terms of success as pros - they're not. That's where there is a disconnect with what I am doing and what you are saying. It's not in terms of success, it's in terms of value added based on their draft position. The way you're verbalizing your interpretation of what this is, is wrong.

 

I am telling you that Green Bay, by drafting Jaire Alexander in the first round got extra value as compared to their peers in RD1 of 2018. I am telling you that Green Bay, by drafting Kevin King in the second round got less value as compared to their peers in RD2 of 2017. 

 

It's not an apples to apples comparison between Alexander and King because they were in two separate drafts in two separate rounds. The whole premise here is that each draft is its own event and teams can only succeed or fail in drafting in comparison to what their peers do in a given draft.

 

Just for fun, if GB were to have taken a player who's performed like Kevin King in RD1 of 2018, say Donte Jackson (CB, CAR) and then in RD2 of 2017 they were to have taken a player who's performed like Jaire Alexander, say Marcus Williams (S, NO). They would be penalized for their RD1 pick in 2018 (bad value) and rewarded for their RD2 pick in 2017 (good value).

 

It's grading GMs success based on value added compared to their peers, it's not grading player success.

Edited January 27, 2021 by JGMcD2

[JGMcD2]

18 hours ago, Billl said:

I guess I didn’t realize that this was your own creation.  You’re never going to achieve consensus trying to quantify the unquantifiable, but you need it to pass the common sense test.  If I were to put something similar together, I would start by combining methods that are already established.  For example, draft value charts aren’t exact, but they’re pretty good in terms of estimating trade equivalencies.  My step 1 would be to total up the draft value points used.

 

My second step would be to find a measure of determining a player’s contribution irrespective of draft position.  There are plenty of versions of these as well, so it’s a matter of picking your favorite.  For the sake of argument, I’ll go with Pro-football-reference.com and their Approximate Value stat.  It’s far from perfect (I don’t think Fred Warner had a better season than Patrick Mahomes, for example), but it’s a decent jumping off point.  (That said, it has Willie Gay at 4 and Sneed at 3, so even this method is going to be terrible.)

An possibly better, though more tedious, way would be to reference a redraft and assign players values based on the draft chart value of their redraft position.  Here’s one that I found, but I have no idea on its quality.  https://bleacherreport.com/articles/2926937-redrafting-the-2020-nfl-draft

It shows Sneed at 25 and CEH at 32, for reference.

 

From there, I would take the sum total of AAV (or redraft score) of the draft class and divide it by the total draft chart points.

 

A quick example using Sneed and CEH would show that 627 points of draft value were used on them in the actual draft.  The redraft positions are worth 1310 points, so they would have a “surplus value” of 109% (they returned 2.09 points of value for every point spent on them).  
 

Justin Jefferson returned 1700 points in the redraft versus the 780 points spent on him, so he had a surplus value of 920 points.  (As  I think about it, I don’t really like the percentage method much, as it really overvalues a seventh rounder who may have had a draft value of 3 points who would have gone in the sixth round with a value of 12 points for a 400% return whereas a Justin Herbert was drafted at a cost of 1600 and a redraft taking him first overall would score him at 3000 points for less than a 200% return.)

 

One note is that UDFAs should absolutely be included in any method and given a draft capital score of 0?

 

 

@Billl

 

I replicated your method as you laid it out. It's not perfect either, but it's a different method conducted as objectively as possible based on your recommendation. 

 

I took every draft from 2017 to 2020 and matched up each player picked with the assigned draft points for that slot. This is what I called OLD_VAL. 

 

I used the draft points from Meers' Harvard Draft Value Chart. My rationale for that can be found in the then italicized section below. 

 

The Draft Value Chart (DVC) gained prominence in the 90’s after the Cowboys overhauled their roster through the draft, winning three Super Bowls that decade with a modest payroll. The actual usefulness of the chart for pre-draft trades has been debated for years afterwards, and in 2011 Harvard economics student Kevin Meers sought to improve the chart, explaining how the point values from the original basically were created arbitrarily and not based on any analysis or statistics.

 

I used AV to objectively "re-draft" each draft class from 2017 to 2020. Each player in the re-draft was then matched up with corresponding points Meers' Harvard Draft Value Chart based on their new draft spot. This is what I called NEW_VAL.

 

After finding both OLD_VAL and NEW_VAL I divided the NEW_VAL by OLD_VAL in order to find SURPLUS_VAL for each draft class.

 

NEW_VAL/OLD_VAL = SURPLUS_VAL

 

Any idea what you think happened before I share the results?

 

DISCLAIMER: I have not included UDFA because the workload requires significantly more manual work. I also don't believe the UDFA process can be likened to the draft process because the process in acquiring players is drastically different than the draft. If I were to conduct an analysis on team's performance in identifying and gaining value from UDFA, it would likely need to be it's own separate analysis. UDFA is an open market bidding war, whereas the draft is simply that, a draft. Player's have very little leverage in the draft and teams have more leverage when it comes to acquiring the players they want but they're at the mercy of other teams taking the players that they want or to move around and acquire the players they covet. 

 

 

 

 

Edited January 27, 2021 by JGMcD2

[JGMcD2]

Alternate Surplus Value Method as suggested by @Billl

 

I took every draft from 2017 to 2020 and matched up each player picked with the assigned draft points for that slot. This is what I called OLD_VAL. 

 

I used the draft points from Meers' Harvard Draft Value Chart. My rationale for that can be found in the then italicized section below. 

 

The Draft Value Chart (DVC) gained prominence in the 90’s after the Cowboys overhauled their roster through the draft, winning three Super Bowls that decade with a modest payroll. The actual usefulness of the chart for pre-draft trades has been debated for years afterwards, and in 2011 Harvard economics student Kevin Meers sought to improve the chart, explaining how the point values from the original basically were created arbitrarily and not based on any analysis or statistics.

 

I used AV to objectively "re-draft" each draft class from 2017 to 2020. Each player in the re-draft was then matched up with corresponding points Meers' Harvard Draft Value Chart based on their new draft spot. This is what I called NEW_VAL.

 

AV tie-breakers were awarded by giving the team who took a player at a lower draft slot in the real draft the higher draft slot in the re-draft. For example, Elijah McGuire and Trent Taylor both have the same DrAV with the Jets and 49ers respectively. Elijah McGuire is awarded the value for pick 117 in the redraft while Taylor is awarded the value for pick 118 in the redraft. This is because McGuire was selected in the 6th round (188) in 2017, while Taylor was selected in the 5th round (177). 

 

After finding both OLD_VAL and NEW_VAL I divided the NEW_VAL by OLD_VAL in order to find SURPLUS_VAL for each draft class.

 

NEW_VAL/OLD_VAL = SURPLUS_VAL

 

Here are the results for each draft class since 2017. 

 

 

 

 

 

 

 

[JGMcD2]

These are the overall results from 2017 to 2020.

 

8 out of 10 teams that were in the top in my analysis were also in the top 10 in this analysis. 8 out of 10 teams that were in the bottom 10 in my analysis were also in the bottom 10 of this analysis. The biggest riser was Dallas up 12 spots from 25 to 13 and the biggest faller was San Francisco down from 8 to 18. The average change in position between my analysis and this analysis was +/- 3 spots.

 

 

[Billl]

Well we’ve learned that my way sucks just as bad or worse. 🤣

 

Theres just no way the Bears who traded up for Trubisky had the 11th best draft while the Browns who drafted Garrett, Peppers, Njoku, and Ohunjobi were 32nd.  Kansas City drafted Mahomes, Kareem Hunt, and Tanoh Kpassignoh ranked 6th.

 

That said, the combined one certainly trends much better than the single season does.  8 of the top 9 teams made the playoffs, so that’s a strong indicator.  Tampa being 14th makes sense because they didn’t draft Brady, Barrett, Gronk, Fournette, etc.

Edited January 28, 2021 by Billl

[JGMcD2]

22 minutes ago, Billl said:

Well we’ve learned that my way sucks just as bad or worse. 🤣

 

Theres just no way the Bears who traded up for Trubisky had the 11th best draft while the Browns who drafted Garrett, Peppers, Nokia, 

It’s not the best draft! It’s the most value... most bang for your buck. If you can separate the two it’ll change your perception. They’re not necessarily the same thing. 
 

The Bears didn’t necessarily have the 11th best draft (there are complex factors I’d have to add to figure that out) but they were the 11th best team at finding value with their picks. Penalized for Trubisky in RD1, rewarded for Eddie Jackson and Tarik Cohen In RD4. Njoku objectively has not been good for Cleveland. Peppers only played 2 seasons for them. Garrett is less valuable than guys drafted after him like Mahomes. 
 

If you don’t mind me asking, out of curiosity, what’s your profession? You’re really bright. 

Edited January 28, 2021 by JGMcD2