The Statistical Reality of Drafting for Needs

Feb 1, 2023 - 8:00 PM

Washington Commanders" src="https://cdn.vox-cdn.com/thumbor/t4co-lrk07EpjR7ydg0lyW77pO0=/0x313:6000x3688/1920x1080/cdn.vox-cdn.com/uploads/chorus_image/image/71930692/1440700849.0.jpg" />

Mock draft season has come to Hogs Haven once again and with it comes the inevitable draft for need vs. best player available debate. The idea that NFL teams should direct premium draft capital to positions of need is prevalent on internet discussion forums and in certain corners of sports media. On the other hand, when NFL GMs are asked about their approach to the draft, they usually stress the primary importance of picking the best players available.

There are two very good reasons for that. One is the inevitable opportunity cost of passing over premium talent to select a less highly rated prospect at a position of need. The second problem with the concept of filling immediate needs in the draft is the limited availability of NFL-ready players after about the first round of the draft.

It is not uncommon in draft discussions on Hogs Haven to hear that the Commanders must use their first three picks in April’s draft to address their top three needs for the 2023 season. In this article, I will attempt to put some numbers around the likelihood of that approach working out for them.

Contrasting Draft Philosophies

NFL GMs tend to approach drafting players with a horizon of two to three years, because that is about time it takes most draftees to become regular contributors, if they work out. The usual view about the place of need in the selection process is expressed in this quote by Ravens’ GM, Eric DeCosta:

The idea would be that you are not going to pass up on a fantastic player to take a good player. We’re never going to do that. But when you’ve got two players close together and one players at a need position and the other players not at a need position, if they’re that close, of course you’re always going to address the need player with a need-based pick.

The first consideration is overall talent and potential contribution to the team over the next few years. Need only enters the equation when the players near the top the team’s draft board are rated similarly on those characteristics. If the top player has a significantly higher scouting grade than the player after him, that’s who they’ll take, regardless of whether he addresses an immediate need or not.

The Draft for Need philosophy dictates that a team’s premium draft capital must be used to address their biggest needs. This line of thinking is illustrated by Bleacher Report’s assessment of the Seahawks’ selection of Russell Wilson with the third round of the 2012 draft. In the words of draft analyst Matt Miller:

Grade: D

Summary

The Seahawks received our lowest Round 1 grade for their reach to grab Bruce Irvin at No. 15 overall. They messed up again in Round 3 with Russell Wilson after having signed Matt Flynn this offseason. Neither pick makes much sense to us.

Bobby Wagner and Robert Turbin saved the Seattle draft, but two of its first three picks were just bad.

Heading into this draft, Bleacher Report identified Seattle’s biggest needs as DE/OLB and WR, with a less obvious need at DT. In Miller’s opinion, Seattle’s use of a third-round pick on another quarterback, instead of addressing the bigger needs at WR or possibly DT, was a waste of draft capital. At least he gave them a nod for their other non-need based pick in the second round.

According to a recent article by ESPN, the Seahawks gained the most benefit from the draft of any NFL team in the decade from 2012 to 2021, and the 2012 draft was the high point of their dream run. The Wilson/Carrol Seahawks went on to win a Super Bowl, two NFC Championships and only missed the playoffs twice in the next decade. The next WR off the board after Wilson was Mohamed Sanu, who started 3 games in 2012 and went on to have a good, if not spectacular career. The Seahawks might have done a bit better if they had a higher grade on the next WR selected, TY Hilton, but almost certainly not as well as they did with the “luxury pick” they spent on Russell Wilson.

Brad Mills-USA TODAY Sports

Commanders’ Biggest Needs

From reading the comments sections on Hogs Haven, I sense a general consensus among those advocating to draft for need that the top three priorities should be OT, IOL and LB. Cornerback comes up fairly frequently too. From a pure needs-based drafting perspective, however, it is not as big a need as the first three because the Commanders have three players under contract for 2023 who are barely competent (Bobby McCain playing slot corner) or better. However, I agree that it’s a major need, so I’ll include it.

Quarterback is undoubtedly a major need, but it doesn’t come up much as a priority for the first round of the 2023 draft. I’m going to leave it out, because I agree with the prevailing view that it’s unlikely the Commanders will take a QB early. Therefore, the Commanders top four needs to be addressed in this analysis are:

1. Offensive Tackle
2. Interior Offensive Line
3. Linebacker
4. Cornerback

Estimating Probability of Drafting First-Year Starters

In order to address any of the top four needs for the 2023 season, the Commanders will need to draft a player who is able to start in his first year. Setting a criterion to identify first-year starters is tricky. Players who become full-time starters as rookies may still take a few games to crack the starting lineup, or miss a few games due to injury. To account for these possibilities, I set the criterion for a first-year starter at players who started a minimum of 10 games in their rookie season.

To estimate the probability of drafting first-year starters at positions of need I had to assume the perspective of an NFL GM seeking to fill a position of need with his next draft pick. Unless he gets lucky, and the next player on his draft board plays the position he is targeting, he will need to scan down his board for the next-highest rated player at that position. The big question is how far down his board he is willing to reach to fill a need.

To approximate that selection process, at each of the Commanders’ draft picks, I counted all of the players at each need position who were drafted at that pick number or the next n picks in each of the past 10 drafts (2013 to 2022). The number n can be thought of as the reach limit. Different teams might rate the next few players taken at the position differently (e.g., in the example above, if Seattle had drafted for need, they might have selected TY Hilton instead of Mohamed Sanu), so I counted all of the players drafted up to the reach limit, rather than only counting the next player taken after the Commanders’ pick. This also helped get the sample size up, for greater reliability.

The reach limit, n, was set at 16 picks in the first round and 32 picks in rounds 2 thorough 4. I don’t know how far a desperate NFL GM is willing to reach for a player. But these numbers are well within the limits of NFL teams’ ability to accurately rank prospects that I have calculated previously.

Finally, to estimate the probability of drafting a first-year starter at each position of need, I simply determined the proportion of players drafted at that position from the Commanders’ pick to the Commanders’ pick number + n who started 10 or more games in their rookie seasons. This is essentially the hit rate for drafting first-year starters in the vicinity of the Commander’s pick.

A few technical notes for the technically minded:

The Pro Football Reference database which was used for this exercise has some limitations which constrained my searches for players matching criteria as follows:

OT – The database does not distinguish between LTs and RTs. Therefore, I searched for players listed as OT who met criteria. This is not really a problem, because the Commanders need new starters at LT and RT.

IOL – The database lists some players as G and C and others as just OL. The latter group tends to include players with unclear NFL position projection in the draft, as well as players that spend time at different positions. Most of these players end up playing guard in the NFL. Therefore, I searched for players listed as G, C or OL.

LB – The database does not provide clean separation of off-ball linebackers from 3-4 outside linebackers who are likely to be primary pass rushers. The comparison cohort therefore includes players with draft position listed as ILB, OLB (to include 4-3 OLBs), and just LB.

CB – The database includes players listed as CB and DB. The latter includes a mix of CBs and safeties. To keep this one clean I limited the search to players listed as CB. This will result in undercounting CBs, but that should not affect the estimates of starting probabilities.

Photo by Scott Taetsch/Getty Images

Probabilities of Drafting Early Starters at OL, LB and CB

Round 1, Pick 16

OT

Probability of Starting Year 1: 8 out of 14 drafted players = 0.571

Left tackle is one of the premium positions in the draft for a reason. Plug and play LT prospects are rare commodities, second only to QBs, and prospects who show obvious day one starting potential tend to get taken off the board in the first few picks.

In the past ten drafts, there have still been enough NFL-ready OTs on the board at the Commanders’ pick to support a hit rate of nearly 60% for drafting first-year starters. More than half of the plug-and-play OTs drafted from picks 16 to 32 played RT in their rookie seasons (4 RT, Ja’Wuan James split time at RT/LT), while three started at LT. The good news for the Commanders is that they either would be good, because they could use upgrades at LT and RT.

IOL

Probability of Starting Year 1: 24 out of 32 drafted players = 0.750

The player cohort used for this probability estimate includes guards, centers and players listed in the Pro Football Reference as “Offensive Lineman.” The latter is likely to consist mainly of IOL and prospects who played OT in college, but have questions about their best position in the NFL. Only eight of the 24 early starters (33%) in this category were listed as playing OT in their first years, so the cross-contamination of OT’s is not a huge issue.

The probability of finding a day-one starter at IOL (with some chance to start at OT) in the second half of the first round is very high. This is no surprise, since IOL is not regarded as a premium position in the draft. IOL prospects who rise first to round consideration are generally elite players at their positions in college with easily-scoutable starting potential.

LB

Probability of Starting Year 1: 15 out of 25 drafted players = 0.600

The historical draft cohort does include some edge rushers. However, only three of the 15 players (20%) that met criteria as first-year starters were edge defenders: TJ Watt, Montez Sweat and Emmanuel Ogbah. Therefore, cross-contamination does not appear to be a major issue.

Like IOL, off-ball linebacker is not generally regarded as a premium position in the draft, so the prospects who rise to first-round consideration tend to be a fairly elite group. Consequently, the probability of finding day-1 starters at Pick #16 is high.

CB

Probability of Starting Year 1: 12 out of 18 drafted players = 0.667

The Pro Football Reference database does not distinguish between boundary and slot corners. I don’t think that’s a problem, because, if the Commanders do pick a CB 16^th, they will probably be targeting a starting boundary CB. Doing so successfully might address the need at slot CB indirectly, by allowing one of their existing CBs to move to the slot.

Of the Commanders four major needs, CB has the second highest position value after LT. It also has the second highest chance of drafting an immediate starter. A GM seeking to optimize the value and chance of success with the Commanders’ first-round pick, while also addressing a need, might lean toward targeting cornerbacks. Prioritizing CB would also align with the strength of this draft, which appears to be particularly stacked at the position in the first round. I am not sure whether those considerations would be important to GM who prioritizes need, however, since there is a much greater immediate need at OL than CB.

Pick #16 Round Up

The recent draft data suggests that the Commanders have between a 57% and 75% chance of addressing one of their top needs for the 2023 draft, depending on which position they target. Their top two needs, aside from quarterback, have the lowest and highest hit rates. Among their top needs, OT is the most premium position, so it is not surprising that it has the lowest hit rate by the middle of the first round. Conversely, the position group with the highest hit rate, IOL, has the lowest position value.

If the Commanders wanted to optimize the value of their first-round pick, they might do well to prioritize CB, which has the best combination of hit rate and position value, and is also stacked in the first round of this draft. On the other hand, it is the least of the four major needs. From a pure needs perspective, they should probably wait to address it, even if an elite prospect is pushed down the board to Pick #16.

Round 2, Pick 47

OT

Probability of Starting Year 1: 7 out of 20 drafted players = 0.350

The estimated probability of drafting a first-year starting OT with the Commanders’ second-round pick is close to 40% lower than in the first round. The mix of drafted OTs playing left and right tackle here is about the same as at 16, with four of the seven first-year starters playing RT as rookies.

IOL

Probability of Starting Year 1: 29 out of 59 drafted players = 0.492

The probability of drafting first-year starters at IOL remains high in the second round. This is where teams really start picking guards and centers in numbers. The interior linemen picked in this range that don’t start as rookies still have a very good chance of becoming eventual starters.

LB

Probability of Starting Year 1: 13 out of 46 drafted players = 0.283

LB has the steepest drop in estimated probability of drafting first-year starters between the first and second rounds. The chance of drafting a first year starter at Pick #47 is more than 50% lower than at #16. It is common to hear that you can find plug-and-play starting linebacker in the middle rounds. The numbers don’t seem to back that up, since the hit rate for finding first-year starters is already below 30% in the middle of the second round.

CB

Probability of Starting Year 1: 10 out of 30 drafted players = 0.333

The estimated probability of drafting a first-year starting CB remains reasonable in the second round, although it is not high enough to count on to fill a roster vacancy in 2023. Cornerback remains a good value proposition for the Commanders in the second round, since they have needs for a starter (slot CB/nickel) and for improved depth. Therefore, an eventual starter addresses a need for 2023.

Pick #47 Round Up

There is some chance of addressing one of the Commanders’ top needs in the second round. However, the probability of doing so tops out at 0.492 for IOL, which is not really high enough to rely on to fill a starting roster vacancy for the 2023 season. It is lower at the other need positions. This is why teams generally try to fill roster vacancies with functional starters in free agency ahead of the draft.

Among the positions of greatest need, CB and OT offer the best value at this pick, since position value and is high and the past hit rates offer some hope of landing a day-one starter. However, by the second round, the Commanders are more likely to draft an eventual starter than an immediate starter. Doing so has value for the team’s future but, of course, would fail to address their biggest immediate needs for the 2023 season.

Round 3, Pick 103 (Projected Comp Pick per Over the Cap)

OT

Probability of Starting Year 1: 3 out of 17 drafted players = 0.176

By the end of the third round, the chance of drafting a first-year starter at OT is below 20%. The probability of drafting eventual starting tackles by this point in the draft isn’t any better, either. Of 19 OTs drafted from picks 103 to 135 between 2011 and 2020, only two started 10 or more games in their third season at OT. One of those was two time All Pro LT David Bakhtiari. A third OT drafted here, Alex Lewis, became a borderline starter at guard for the Ravens and the Jets.

IOL

Probability of Starting Year 1: 8 out of 52 drafted players = 0.154

So much for the theory that you can draft a plug-and-play guard in the fourth round. Well, you can, 15.4% of the time.

LB

Probability of Starting Year 1: 3 out of 39 drafted players = 0.077

If you thought the chance of drafting a first-year starter at IOL was bad, the chance of finding a day-one starter at LB this late in the draft is even worse. In fact, it is half as likely.

CB

Probability of Starting Year 1: 2 out of 22 drafted players = 0.091

There is less than a 10% chance of drafting a first-year starter at CB by the end of the third round. This should surprise no one.

Pick #103 Round Up

The probability of landing a first-year starter to address any of the Commanders’ major needs with the 103^rd pick tops out at 0.176. At LB and CB, the chance of hitting on a first-year starter this late in the draft is much lower than that. If Ron and the Martys hope to address any of the top four major needs in the draft, they had better do it with their first two picks. Any GM whose plan to address a major roster need for the following season is a late Day 2 draft pick should think about another line of work.

Round 4, Pick 117

The Commanders’ 4^th round pick is expected to move back 2 places due to addition of 2 3^rd round comp picks.

OT

Probability of Starting Year 1: 1 out of 20 drafted players = 0.050

The Commanders’ 4^th round pick is only fourteen places behind the projected 3^rd round comp pick. Nevertheless, the estimated probability for drafting first-year starting OTs with this pick is even lower.

IOL

Probability of Starting Year 1: 5 out of 51 drafted players = 0.098

Yeah. About those plug-and-play guards in the fourth round. They exist, but the chance of finding one is around 10%.

LB

Probability of Starting Year 1: 3 out of 40 drafted players = 0.075

The hit rate for drafting first-year starting LBs at pick #117 is about the same as the previous pick. At around 7.5% it is far too low to rely on to address the need at linebacker in 2023.

CB

Probability of Starting Year 1: 2 out of 26 drafted players = 0.077

The chance of landing a first-year starting CB at Pick #117 is slightly lower than at Pick #103. A pick in this range could address the need for better depth at this position.

Pick #104 Round Up

By Day 3 of the draft, the chance of finding a first-year starter at any position is below 10%. By this point in the draft, the Commanders should be looking for depth players with eventual starting potential.

Photo by Eakin Howard/Getty Images

Conclusions

The idea of addressing the Commanders’ top three needs other than QB with their first three picks might look good on paper. The only problem is that it is not based in reality.

The recent draft data shows that the Commanders have between a 57% and 75% chance of addressing one of their top four needs for the 2023 season in the first round. That might be comparable to the hit rate that Rivera’s Commanders have achieved with free agent signings, if not higher.

However, at their second-round pick, the chance of drafting a second first-year starter is around half of what it was in the first round. If Ron Rivera were to address the two biggest needs (other than QB) by picking OT and IOL with his first two picks, his chance of success would be 0.571 x 0.492 = 0.281. If he was willing to be a bit more relaxed about level of need, he could give himself the highest chance of addressing two needs for 2023 by drafting CB in the first and IOL in the second. In that case, the estimated probability becomes 0.667 x 0.492 = 0.328. That only improves his chances by around 4%.

Taking that one step further, he would have the highest chance of addressing a third need by taking an OT in the third round. Therefore, the maximum probability of addressing three top needs with his first three picks is 0.667 x 0.492 x 0.176 = 0.058, or 5.8%. It’s not impossible, but you wouldn’t want to plan your offseason around it.

In summary, the Commanders have a good chance of addressing one need for the 2023 season in the first round. After the first round, the chance of picking a first-year starter at any position is too low to rely on to address a need for the upcoming season. The probabilities in the second round are high enough to offer hope, but not enough to roll with without a capable starter in place for security. By the third round, the chance of hitting on a first-year starter to address any of the Commanders’ top needs is below 18%, which should not be any sane person’s idea of a reasonable plan.

These numbers illustrate why NFL teams usually sign veterans to fill roster vacancies ahead of the draft and why the good ones, at least, don’t wait until the last minute to draft players to replace departing starters.

The best use of the first round pick is debatable. From Round 2 onward, draft capital is better used to address depth for the future than attempting to chase immediate roster needs. Since roster needs are only partially predictable a year or two in advance, immediate needs add less and less value to the decision-making process the later in the draft we get.

Acknowledgement: Edited by James Dorsett

---