The odds of success for a draft pick, part 3
April 15, 2020
by Steve Thomas
This is my continuing, and probably final, effort in my recent draft pick analysis series. Once again, I apologize for burdening everyone with more statistics for a third week in a row, but if there were ever a time to take an analytical look at draft success, it’s late March and early April. In the first column, I examined the “success” and “bust” rate of every draft pick selected in the top 10 between 2000 – 2019 (click here to read). Part 2 expanded the analysis to picks 11 – 20 and 21 – 32 of round 1 and round 2 (click here to read). Now, in part 3, I’ll make an attempt to examine the success of top 10 picks by team and compare those results to winning percentage. I promise this will be the last edition of these columns, unless it isn’t, in which case I will withdraw my promise. And don’t worry: I’m going to do everything possible to make sure this column isn’t so incredibly data-intensive as was the last column. With that having been said, let’s get into it.
For this exercise, I pulled every draft pick in the top 10 between 2000 – 2019 (again), only this time sorted them by team, and then charted the same data as I did for the first and second parts of this (admittedly now massive) effort: number of players drafted who (1) earned multiple 1st Team All Pro selections, (2) earned at least one 1st Team All Pro selection, (3) earned at least one Pro Bowl nomination, and, in an attempt to determine the number of obvious, massive busts, (4) the number who were only full time starters for one or zero seasons. As to the fourth group, the “bust rate”, I have excluded all players drafted in 2019.
The goal here is to see if we can determine which teams have been better and worse at unearthing elite talent, above average talent, and busts, but also then to see if a correlation between successful and poor drafting and winning and losing exists.
One major limitation of this effort is that some teams have had as little as 2 top 10 selections in the evaluated time period, which isn’t enough of a statistically relevant sample to establish any sort of scientifically valid statistical conclusion for each team. Also, a myriad of other inputs are involved in causing a team to win, lose, or be mediocre, not just draft picks, and I don’t claim that this is the be-all, end-all of draft evaluation. This is just a quick look at some data to see what we can see.
The data
This first chart shows all 32 teams, the number of top 10 draft picks they made between 2000 – 2019, and each team’s respective winning percentage, sorted by winning percentage:
| Rank | Team | # pks | Win % | Rank | Team | # pks | Win % | |
| 1 | Patriots | 2 | 0.741 | 17 | Panthers | 5 | 0.492 | |
| 2 | Steelers | 2 | 0.644 | 18 | Bears | 7 | 0.491 | |
| 3 | Packers | 3 | 0.619 | 19 | Dolphins | 5 | 0.466 | |
| 4 | Colts | 2 | 0.616 | 20 | 49ers | 9 | 0.464 | |
| 5 | Eagles | 3 | 0.595 | 21 | Jets | 9 | 0.463 | |
| 6 | Ravens | 4 | 0.594 | 22 | Bengals | 8 | 0.458 | |
| 7 | Saints | 3 | 0.572 | 23 | Texans | 6 | 0.455 | |
| 8 | Seahawks | 3 | 0.57 | 24 | Rams | 7 | 0.452 | |
| 9 | Broncos | 2 | 0.569 | 25 | Cardinals | 10 | 0.431 | |
| 10 | Cowboys | 5 | 0.525 | 26 | Buccaneers | 7 | 0.431 | |
| 11 | Chiefs | 6 | 0.519 | 27 | Bills | 8 | 0.428 | |
| 12 | Vikings | 6 | 0.519 | 28 | Redskins | 8 | 0.414 | |
| 13 | Falcons | 7 | 0.514 | 29 | Jaguars | 14 | 0.397 | |
| 14 | Chargers | 5 | 0.506 | 30 | Raiders | 9 | 0.391 | |
| 15 | Titans | 7 | 0.5 | 31 | Lions | 11 | 0.358 | |
| 16 | Giants | 5 | 0.494 | 32 | Browns | 12 | 0.311 |
The correlation coefficient[1] between winning percentage and number of Top 10 picks is -0.878, which is a very strong negative correlation – in English, for the most part, the higher the overall winning percentage, the fewer Top 10 draft picks were made. That’s a very logical outcome and not at all surprising. I didn’t bring you here to prove that hypothesis. I’m going somewhere with this, trust me.
The second chart shows each team, their number of picks in the top 10 between 2000 – 2019, the total number of players drafted by each team who earned multiple first team All Pro honors, and the percentage of players drafted who achieved such honors, sorted by winning percentage.
| Team | # pks | # Mult 1st tm AP | % Mult 1st tm AP | Team | # pks | # Mult 1st tm AP | % Mult 1st tm AP | |
| Patriots | 2 | 1 | 50.00% | Panthers | 5 | 2 | 40.00% | |
| Steelers | 2 | 0 | 0.00% | Bears | 7 | 1 | 14.29% | |
| Packers | 3 | 0 | 0.00% | Dolphins | 5 | 0 | 0.00% | |
| Colts | 2 | 1 | 50.00% | 49ers | 9 | 0 | 0.00% | |
| Eagles | 3 | 0 | 0.00% | Jets | 9 | 0 | 0.00% | |
| Ravens | 4 | 0 | 0.00% | Bengals | 8 | 0 | 0.00% | |
| Saints | 3 | 0 | 0.00% | Texans | 6 | 1 | 16.67% | |
| Seahawks | 3 | 0 | 0.00% | Rams | 7 | 1 | 14.29% | |
| Broncos | 2 | 1 | 50.00% | Cardinals | 10 | 1 | 10.00% | |
| Cowboys | 5 | 1 | 20.00% | Buccaneers | 7 | 0 | 0.00% | |
| Chiefs | 6 | 1 | 16.67% | Bills | 8 | 1 | 12.50% | |
| Vikings | 6 | 2 | 33.33% | Redskins | 8 | 0 | 0.00% | |
| Falcons | 7 | 1 | 14.29% | Jaguars | 14 | 0 | 0.00% | |
| Chargers | 5 | 1 | 20.00% | Raiders | 9 | 1 | 11.11% | |
| Titans | 7 | 0 | 0.00% | Lions | 11 | 2 | 18.18% | |
| Giants | 5 | 0 | 0.00% | Browns | 12 | 1 | 8.33% |
The correlation coefficient between winning percentage and the percentage of players selected by each team who earned multiple first team All Pro honors is 0.318, which indicates a small positive correlation between the two data sets. This means that while the better teams have done a better job unearthing franchise-level talent at the top of the draft, it is only slightly so overall. Note that the range of results is between 0% (i.e., some teams drafted no players who earned multiple first team All Pro honors; including the Redskins) and 50% (Patriots, Colts, and Broncos). One limitation with this data is that such a small number of picks per team, such as the Patriots, Steelers, and Broncos with just 2 each, limits the statistical relevance of the information. Nevertheless, this is what we have. Better teams are slightly better than lesser teams in finding elite talent in the top 10, but not substantially so.
The third chart shows each team, their number of picks in the top 10 between 2000 – 2019, the total number of players drafted by each team who earned at least one first team All Pro team honor, and the percentage of players drafted by each team who achieved such honors, sorted by winning percentage:
| Team | # pks | # one 1st tm AP | % one 1st tm AP | Team | # pks | # one 1st tm AP | % one 1st tm AP | |
| Patriots | 2 | 2 | 100.00% | Panthers | 5 | 5 | 100.00% | |
| Steelers | 2 | 0 | 0.00% | Bears | 7 | 1 | 14.29% | |
| Packers | 3 | 0 | 0.00% | Dolphins | 5 | 1 | 20.00% | |
| Colts | 2 | 1 | 50.00% | 49ers | 9 | 1 | 11.11% | |
| Eagles | 3 | 1 | 33.33% | Jets | 9 | 1 | 11.11% | |
| Ravens | 4 | 3 | 75.00% | Bengals | 8 | 1 | 12.50% | |
| Saints | 3 | 0 | 0.00% | Texans | 6 | 2 | 33.33% | |
| Seahawks | 3 | 0 | 0.00% | Rams | 7 | 1 | 14.29% | |
| Broncos | 2 | 1 | 50.00% | Cardinals | 10 | 2 | 20.00% | |
| Cowboys | 5 | 3 | 60.00% | Buccaneers | 7 | 1 | 14.29% | |
| Chiefs | 6 | 2 | 33.33% | Bills | 8 | 2 | 25.00% | |
| Vikings | 6 | 2 | 33.33% | Redskins | 8 | 0 | 0.00% | |
| Falcons | 7 | 3 | 42.86% | Jaguars | 14 | 1 | 7.14% | |
| Chargers | 5 | 1 | 20.00% | Raiders | 9 | 1 | 11.11% | |
| Titans | 7 | 2 | 28.57% | Lions | 11 | 2 | 18.18% | |
| Giants | 5 | 0 | 0.00% | Browns | 12 | 1 | 8.33% |
The correlation between winning percentage and the percentage of players selected by each team who earned at least one first team All Pro honors is 0.423, which is slightly stronger than that for the multiple first team All Pro honors data set, but nonetheless still fairly weak. This again indicates that the better teams were more successful at identifying this level of talent, but not tremendously so. As the chart shows, the range of results is between 0% (Redskins and 5 other teams) and 100% (Panthers and Patriots). The same limitation as I mentioned with the prior chart also applies here: the small number of draft picks for some of the better teams limits the statistical validity of the correlation calculation, but, again, this is what we have.
The fourth chart shows each team, their number of picks in the top 10 between 2000 – 2019, the total number of players drafted who earned at least one Pro Bowl selection, and the percentage of players drafted who achieved such honors, sorted by winning percentage:
| Team | # pks | # with one PB | % with one PB | Team | # pks | # with one PB | % with one PB | |
| Patriots | 2 | 2 | 100.00% | Panthers | 5 | 5 | 100.00% | |
| Steelers | 2 | 0 | 0.00% | Bears | 7 | 2 | 28.57% | |
| Packers | 3 | 1 | 33.33% | Dolphins | 5 | 3 | 60.00% | |
| Colts | 2 | 2 | 100.00% | 49ers | 9 | 6 | 66.67% | |
| Eagles | 3 | 3 | 100.00% | Jets | 9 | 3 | 33.33% | |
| Ravens | 4 | 3 | 75.00% | Bengals | 8 | 3 | 37.50% | |
| Saints | 3 | 0 | 0.00% | Texans | 6 | 3 | 50.00% | |
| Seahawks | 3 | 2 | 66.67% | Rams | 7 | 2 | 28.57% | |
| Broncos | 2 | 1 | 50.00% | Cardinals | 10 | 5 | 50.00% | |
| Cowboys | 5 | 4 | 80.00% | Buccaneers | 7 | 3 | 42.86% | |
| Chiefs | 6 | 3 | 50.00% | Bills | 8 | 4 | 50.00% | |
| Vikings | 6 | 5 | 83.33% | Redskins | 8 | 8 | 100.00% | |
| Falcons | 7 | 6 | 85.71% | Jaguars | 14 | 3 | 21.43% | |
| Chargers | 5 | 3 | 60.00% | Raiders | 9 | 2 | 22.22% | |
| Titans | 7 | 2 | 28.57% | Lions | 11 | 6 | 54.55% | |
| Giants | 5 | 2 | 40.00% | Browns | 12 | 6 | 50.00% |
There is a positive correlation between winning percentage and the percentage of players selected to at least one Pro Bowl, but it is a very weak correlation, just 0.224. This means that finding a draft choice in the top 10 who is at least a Pro Bowl-level player is barely more than a crapshoot overall between the best and worst teams. As the chart shows, the range is between 0% (Saints) and 100% (Redskins and 4 other teams). For example, the Cleveland Browns, which has been the NFL’s worst overall franchise over the past 20 years in almost every conceivable way one could measure such a thing, found 6 Pro Bowlers in 12 selections, whereas the Steelers struck out on their 2 selections. The Redskins hit on all 8 of their selections, and they have the 28th-ranked winning percentage.
The final chart shows each team, their number of picks in the top 10 between 2000 – 2019, the total number of players drafted who only played either one or zero years as a full-time starter (except for the 2019 draft class), and the percentage of players falling into that category (the “bust rate”), sorted by winning percentage:
| Team | # pks | # started 0/1 yrs
|
Bust rate | Team | # pks | # started 0/1 yrs | Bust rate | |
| Patriots | 2 | 0 | 0.00% | Panthers | 5 | 0 | 0.00% | |
| Steelers | 2 | 1 | 50.00% | Bears | 7 | 2 | 28.57% | |
| Packers | 3 | 1 | 33.33% | Dolphins | 5 | 1 | 20.00% | |
| Colts | 2 | 0 | 0.00% | 49ers | 9 | 0 | 0.00% | |
| Eagles | 3 | 0 | 0.00% | Jets | 9 | 2 | 22.22% | |
| Ravens | 4 | 0 | 0.00% | Bengals | 8 | 0 | 0.00% | |
| Saints | 3 | 1 | 33.33% | Texans | 6 | 0 | 0.00% | |
| Seahawks | 3 | 0 | 0.00% | Rams | 7 | 1 | 14.29% | |
| Broncos | 2 | 1 | 50.00% | Cardinals | 10 | 2 | 20.00% | |
| Cowboys | 5 | 0 | 0.00% | Buccaneers | 7 | 0 | 0.00% | |
| Chiefs | 6 | 0 | 0.00% | Bills | 8 | 0 | 0.00% | |
| Vikings | 6 | 0 | 0.00% | Redskins | 8 | 0 | 0.00% | |
| Falcons | 7 | 0 | 0.00% | Jaguars | 14 | 2 | 14.29% | |
| Chargers | 5 | 1 | 20.00% | Raiders | 9 | 0 | 0.00% | |
| Titans | 7 | 1 | 14.29% | Lions | 11 | 1 | 9.09% | |
| Giants | 5 | 0 | 0.00% | Browns | 12 | 1 | 8.33% |
The correlation between overall winning percentage and my “bust rate” was slightly positive – 0.215, which is almost the same as it was for the Pro Bowl data. As is the case for each of the five categories of data, the small sample size for some teams limits the statistical relevance of the results. Regardless, this weak correlation tells us that suffering a clear and obvious bust of a draft pick doesn’t have much relationship to winning. In other words, the good teams can overcome their bad picks, or haven’t had enough of them to matter, in a broad, overall sense.
Analysis and conclusions
The fact that the only strong correlation with winning percentage and any of the other categories of data was the number of picks in the Top 10 validates two broad concepts: (1) the draft is truly a roll of the dice, even at the top of the draft, and (2) teams with strong ownership and culture can overcome bad draft picks, whereas teams that have historically been a mess (hello, Cleveland, nice to see you again) can’t overcome their mistakes. Many things go into becoming a consistent winner, and the draft is but one variable in the calculus. The franchises that have consistently won over the years like the Patriots, Steelers, Packers, and Colts have done so without many Top 10 picks, and without uncovering many more franchise-level picks on a percentage basis than the worst teams like the Jaguars, Raiders, Lions, and Browns. Clearly, just having alot of very high draft picks isn’t going to solve a team’s problems unless that team is also able to overcome its other structural problems.
What isn’t addressed here is the number of picks in each position group selected by each team, primarily because that would dramatically expand the scope and length of this column. Obviously, the Patriots, Packers, and Colts, in particular, have had very strong stability and elite performance at quarterback, whereas bad teams like the Redskins, Browns, etc., have all been through the ringer at the position. Good coaching, a positive culture established by the front office and the head coach, and the effect of free agency are critical variables that go into establishing and maintaining a consistent winner.
If you only take one thing out of all three of these columns, it should be that the draft isn’t a panacea: very high draft picks are probably not going to turn out the way teams hope, and even if they do, it won’t affect winning much, and acquiring many picks over the years isn’t going to help much either unless teams fix the other things that are wrong with the organization.
Do you have any thoughts on this information? If so, please leave a comment in the comment section below.
[1] The correlation coefficient shows the relationship between two sets of data – whether there is a positive relationship, an inverse relationship, or no relationship at all. It is measured between -1 and 1, with 0 indicating no correlation, negative numbers indicating an inverse correlation, and positive numbers indicating a positive correlation. In other words, if the correlation coefficient is higher than zero, then the two data sets are “linked”, meaning the higher or lower in one data set should equate to a higher or lower number in the other data set. The inverse is true for a negative correlation coefficient: the higher number in one data set generally means a lower number on the other data set.