What is the best strategy for approaching draft moves? Trade up, trade down, or stay put? I've pulled together a few sources to try and frame up the answer. I did this for my own education. There are many people on the site that already know all this, and more, but there are also so many senseless and uninteresting debates about draft tactics and such that it seemed worth sharing.
Draft Expectations based on Draft Position
First, what can we expect from a draft selection? Draftmetrics did a study of NFL drafts from 1995-2004, and then updated it from 1993-2012, looking at results relative to draft placement. They found that there were groups of players that tended to have the same value, and they named them Value Groups (VG). There are 7 VGs, but they do not match the 7 rounds of the draft. They are as follows:
|Group||Draft Position||% Chance 3YR Starter||% Chance Pro Bowl||Round|
|VG1||1-13||74%||44%||Top of 1|
|VG2||14-40||61%||19%||Mid to Lower 1, Top 2|
|VG3||41-66||44%||13%||Bulk of R2, Very Top 3|
|VG4||67-86||30%||6%||Top to Mid 3|
|VG5||87-149||19%||4%||Bottom 3, R4, Top Half 5|
|VG6||150-189||12%||3%||Bottom 5, R6|
|VG7||190+||8%||2%||Bulk of R7|
How to read this? Any draft will have a dozen or so elite players (avg = 13). After that, the remainder of the 1st round talent is about the same through the top of round 2. Other than the top of round 2, the rest of the round 2 is about equal. Round 3 is the most variable, with the top 3 players in VG3, and the bottom 10 players in VG5. Finally, round 6 is slightly more valuable than round 7, but not by much.
These numbers are important when evaluating draft picks. For instance, Marcus Lattimore was drafted with the 131st pick of last year's draft. Historically, this is smack dab in the middle of VG5, with a 19%, or 1 in 5, chance of drafting a 3 year starter. So, we drafted a player that we know we will lose 1 season from, but we are getting potential first round talent with a pick that historically would only hit once every 5 years. This seems like a no-brainer. Even if Lattimore doesn't work out, the logic is completely value driven.
We employed this same strategy with Tank Carradine, our 40th pick. This is the bottom of value group 2. A first round pick costs anywhere from 6.7M to 22M over 4 years. We will be paying Tank ~5M for 3 years of service, with the option that Tank would be a restricted free agent when his contract ends. This allows us to potentially pick up a 5th year at a reasonable price, or get a high draft pick from the team that steals him from us. This pick is not clearly the same "slam dunk" as above. When you think we might only be getting 3 years of service, we are essentially paying first round money. Tank has all the measurables that our FO is looking for at DT/DE, and ACL injuries are very well understood, so we will have to see how this plays out.
Playing the percentages, Tank and Vance's draft slots have a combined 105% chance of producing a single 3 year starter. If they both fail, then major fail. If they both become starters, then great draft. If one of them becomes a starter, then that is about average.
The real value in looking at this data is to understand those lower round picks. By the end of the second round, or VG3, the odds are AGAINST a player being a 3 year starter in the league. If one of Corey Lemonier, Quinton Patton, and Marcus Lattimore becomes a 3 year starter in this league we are hitting the league average. If two of them hit, we've blown the average out of the water. Combined, those 3 draft positions have a 12% chance of going to 1 pro bowl. If any of these guys makes a pro bowl we've basically hit a home run.
The irony is with all these 7th round picks that we've accumulated over the years. They've never been worth much, and now that our roster is stacked and teams are scrutinizing our waiver wire the picks are almost useless. As much as you may pine for players like BJ Daniels, he will most likely never start in this league. Marcus Cooper has already defied the odds, and in the metrics above he would count as a positive for us as the analysis does not differentiate which team the player starts for. It is no coincidence that this offseason the FO brought in a large number of free agents. They kept the numbers balanced so that we would get one high comp pick, but after that we are better off getting the free agent talent than rolling the dice on another late 7th rounder.
An average draft will produce 2 "3 year starters." That's it. If we can produce an average of 3 starters per year, or continue to draft players like Navarro Bowman who become all pros, we are doing great.
Best Value in the Draft
Success rates define the risk with a draft pick, but to get the value we need the cost also. The rookie salaries here are from overthecap.com for 2013. I plotted the ratio of salary against the #1 pick.
The a-axis is draft pick, the vertical lines separating the rounds. The blue line is the % chance of drafting a 3 year starter. The brown line is the salary of the draft pick relative to pick #1, using 2013 numbers. While not exact, generally speaking where the gap between the blue line and brown line is the biggest represents where the most value sits. Note, it does not mean anything when the brown line is above the blue line. I just didn't normalize the draft yield rates.
VG2 and VG3 represent the best values in any draft. A decent chance of success, and at less than half the cost of the top picks. You lose the ability to draft any player you want, but as long as you can find the right players, you will get the most value from your picks. Even if you don't buy into Draftmetrics Value Groups, multiple other sources have identified the second round as the best value of the draft.
There are additional nuances to drafting in the first or second round. 1st round picks have a 5th year option. For picks 1-10, it is an average of the top 5 salaries for the position from the previous year. For picks 11-32, it is the average of the top 25 salaries. For example, Aldon Smith was the #7 selection. We will most likely exercise our option for 2015, at a cost of slightly less then $10M. If Aldon had been the #11 pick, that number would drop to around $7M (overthecap estimate).
The downside of first round contracts is that they are guaranteed. Some second round contracts have some guarantees, but not to the same extent as first rounders. AJ and John Baldwin are examples of this. I think both players had the first 3 years of their contracts guaranteed, which is a large part of the reason that we did the trade. We were able to give Kansas City cap room in 2013 and in exchange they took on AJ's guaranteed salary this year. Picks at the beginning of the second round have 2 years guaranteed. By the end of the second round there are no guarantees other than signing bonuses.
Following this draft chart, the best spots to pick in any draft are probably #11-13, and #30-35. With the 11th pick, you have the highest chance of success at half the cost of the #1 pick and the lower cost for the 5th year option. Pick 32 has the lowest salary with the first round benefits. Pick 33 has the same chance of success as pick 32, without the 5th year option, but also with less guaranteed money.
Next up would be the draft picks at the end of round 2 and beginning of round 3 (#63-66). Your still in VG3 at the lowest cost in the rookie wage scale.
The value groups listed here are averages. They represent the best spot to sit if you had to sit in the same draft spot for the next 100 years. There is variation draft to draft, so the value groups can be larger or smaller for any given year. It would be a great exercise, if its not already done, to use the mock drafts to determine where the value groups sit for 2014. It would help us predict trades. For example, 2014 is said to be a deep draft. If the value groups are 25% larger then average, than our 3rd round pick, #77, would fall into VG3 instead of VG4. This makes the pick more valuable, and if it at the edge of a value group we may be able to trade it for a premium.
Draft Trade Value
Given the above, in most years I would advocate trading up into the middle of the first round to get a player we've targeted. Since the value group is the same, we can hopefully get a trade at a discount and have more control over the player selected. We have a stacked roster, so there is value to us being able to target a position versus a team like Oakland that can take BPA at any position and most likely upgrade their roster. Our trade with Dallas last year for Eric Reid is an exact example of this. People said we won the deal but it was really a win-win. We were able to target one of the guys we wanted. Dallas was able to stay in VG2 and pick up a 3rd round pick. If you believe in the value groups concepts the actual point differential does not matter.
There are many trade charts that people reference. The traditional chart was developed by Jimmy Johnson, and is an exponential drop from the first pick (worth 3000) to the last pick (worth 1). More recent charts recognize that the early picks are overvalued in Jimmy Johnson's chart. For example, the draftmetrics data above suggests the 1st pick and the 13th pick, on average, have the same value. In 2013, Miami traded with Oakland to move from the #12 to the #3 slot. They gave Oakland pick 42. Using the draft chart, Oakland's pick was worth 2200, while Miami's two picks were worth 1680 (1200 + 480). Oakland gave Miami a 24% discount, or 520 points. In reality, Oakland stayed within their value group (3 to 12) and picked up an extra VG3 pick. For that particular draft, the winner will eventually be decided by they players each team picked up. If they do that trade 100 times, Oakland wins in the long run.
I believe most teams do create value groups, but just give them different labels such as A and B grades.
When it comes to trades, the value for a pick is in terms of draft points instead of salary. Probably not coincidentally, rookie wages track against the traditional trade chart, so plotting value group yield versus trade value looks about the same:
These values are both normalized against the number 1 pick. Again, the end of VG2 represents the best value (red arrow). The question is, do teams still use this chart? The 49ers, for example, have said that they created their own trade chart.
I took a look at the last 3 years drafts using the traditional draft chart. I took out trades for future year picks, just to simplify the analysis, and plotted the actual points received versus the expected points received for each trade.
The Y-axis is difference in draft points. A positive number means to move to that draft spot you had to pay a premium from the traditional draft chart. A negative value represents a discount.
Trade values in the beginning of the draft are fairly volatile. In 2011, Atlanta gave up 5 picks to Cleveland to move from 27 to 6, but otherwise most moves at the top of the draft are less than 7 spots. You may be able to get a discount, but you may also have to pay a small premium. A 100 point difference in the top 15 picks is less than 10%.
To trade up into VG2 expect to pay full price. You might get a discount trading up within the value group, such as our trade with Dallas referenced above.
If you can wait for VG3 or VG4, you can probably get a discount. Teams don't appear to value this draft range. I would estimate teams trading up can do so for about 10% less than the traditional trade chart dictates.
Once you get into VG5, or the end of round 3, you are back to paying full price again.
Finally, cost is relative. If you are trading up a value group ( i.e. 3 to 2) expect to pay more than if you are trading within a value group.
Draft Strategies - Bundling Late Round Picks
Bundling picks to make a large jump in draft position does happen, but unfortunately it usually only happens in the early part of the draft. If you are willing to give up multiple draft picks, it usually means there is a player that you are really exited about. Those players are at the beginning of the draft. If we think of it in terms of percentages, a sixth and two seventh round draft choices, depending on where they fall, would yield a 28% chance of drafting a 3 year starter. Those same 3 picks, bundled together, would get you into the middle of VG5, with a 19% chance of drafting a 3 year starter. You are gambling on a player that, quite frankly, has issues or he would have already been drafted sooner. Teams would rather play the percentages at this point in the draft.
In 2012, Green Bay gave New England picks 197, 224, and 235 for pick 163. New England was the draft points loser, but I bet they do this trade every time.
Draft Strategies - Trading into Future Years
Teams generally request a 1 round improvement if they are giving away a current draft pick for a pick in a future year. If you want someone's 3rd, you give them your 2nd in the following year. Who wins in these scenarios ends up depending on how the teams do the following year. For example, we got Kansas City's second round pick this year as part of the Alex Smith trade. That pick could have ended up being worth anywhere from 270 to 580 draft points, a huge swing depending on Kansas City's performance this year. The worst part about this is that the team that trades away a future pick and then does poorly loses out even more, as the draft pick they gave away becomes worth even more at a time when they clearly could use it (e.g. Washington).
[Update: As pointed out in the posts, the Alex Smith trade was for two picks, and the 2014 pick was actually a conditional pick. If the Chiefs had stunk, the pick would have become a 3rd rounder. This creates a larger swing in potential value, but allowed the Chiefs to hedge their bets slightly against the phenomena I was illustrating and highlights the value of making future picks conditional.]
Trading away future picks seems like a win now strategy that is short-sighted. Teams that do this are clearly in love with a player that is still available. Even a slam dunk, like RG3, might now work out how you thought because of injury, capability, or circumstances.
In 2013 we gave away pick #34 to Tennessee for pick #40 and a 3rd rounder this year. In terms of value, we clearly destroyed Tennessee with this trade. We stayed in VG2 and got the player most people think we wanted anyway, Tank Carradine. Tennessee got WR Justin Hunter. He may end up being great, but Tennessee gave up a lot to get him, and then he was limited in 2013 when they lost their starting QB early in the season.
There is no such thing as a can't miss draft prospect, so why limit your future year's growth when even in the best of circumstances you are still gambling?
Also, teams do not give up future 1st or 2nd round picks on draft day unless it is one of those blockbuster trades. Realistically, the best you will get for a future round pick is a 3rd rounder. There are exceptions, so never say never, but most likely no one is going to give up next years 1 for your 2nd pick.
Outside of the first round:
In 2011, Oakland traded away a 2nd and TB traded away a 4th.
In 2012, Caroline traded away a 3rd, Detroit a 4th, Indy a 5th, and Miami a 6th.
In 2013, three teams gave away two 3rds and a 4th.
Less than 1% of picks are traded for future year picks and a more realistic scenario is moving back in the first or second round to pick up a 3rd to 5th round future pick. Actually, the number of future picks the 49ers have acquired is quite amazing.
Draft Strategies - Trading Players for Picks
Mike Iupati and LaMichael James names seem to come up all the time as trade candidates. Teams will make these kinds of trades, but usually for skill players, not linemen. QBs, WRs, and CBs seem to be the only positions worth a trade in the early rounds. The only exception to this the last 3 years is Richard Seymour, who was traded for a first, but unfortunately the maverick behind that trade, Al Davis, is no longer with us.
Other players recently traded for 1sts: Darrelle Revis (CB), Percy Harvin (WR), Carson Palmer (QB)
For 2nds: Alex Smith (QB), Vontae Davis (CB), Kevin Kolb (QB), Antonio Cromartie (CB)
For 3rds: Brandon Marshall - twice (WR), Greg Olsen (TE), DeMeco Ryans (ILB), Jammal Brown (OT, but this was with Donavan McNabb)
If history is our guide, the best we should hope for in an Iupati trade is a 5th rounder. Again, never say never, but if you think we can do better you should have a pretty strong rationale.
As far as running backs go, last year Chris Ivory was traded to the Jets for a 4th round pick. This offseason, Darren Sproles was traded to the Eagles for a 5th round pick. Both these players have more stature than LMJ, so I can't see a scenario where we would get any value for him in a trade. He is still too unproven.
Most draft pundits say you should always draft the Best Player Available, or BPA. Everything here supports that notion. The key is recognizing that draft picks are valued in groups, not force ranked. The 49ers will identify a group of players, for example, that they believe are the best of the best. If one of them falls to us at pick 30, we will take that person regardless of position. Most likely, all those VG1 players will be gone by the time pick 30 comes up, so we will draft from the next group of players. If you want a cornerback, but all the cornerbacks on your board are in a lower group, then you try and trade down. You don't reach for a player because of his position.
The BPA concept drives a lot of the trades that happen on draft day. You have a position of need, and you don't want to reach on a player, so you move around to get the player that fits both your value group and your needs.
While it is fun to produce lists of players, we should really put them into groups to best reflect this value group concept. The notion that a player may score a grade of 95 and then another player, playing a completely different position, scores a 94, and is therefore inferior, is just not practical. The numbers are reflecting a degree of precision in draft evaluations that just doesn't exist.
There are exceptions to BPA that are worth noting. For example, we have two of the three best inside linebackers in football on our roster. If a great ILB fell to us in the first round it might not make sense to pick him. Willis and Bowman are on long term contracts, so we would essentially be paying $2M per year for a backup. The new CBA has drastically reduced rookie wages, but they are still sizable in the early rounds. If you are maximizing the total value of your draft picks then you don't want your early picks on the bench. Depth is important, but are you overloading at one position and neglecting another?
As long as you draft from the highest group of players available to you then you are following the BPA concept. I don't know that anything more draconian than that is necessary.
I've been guilty of saying Harbaugh likes to "red shirt" his draft picks, sitting them out a year before putting them in the line up. That is not exactly true. Harbaugh actually wants to get each guy on the field as soon as possible. He just runs a "meritocracy." You don't get thrown out there to learn as you go. You have to earn playing time in practice. Bruce Miller started before the end of his first year. Aldon and Vance were key contributors their first year. Eric Reid started from game 1 his first season. In each of these cases, the players simply had to earn their playing time. If we draft a player, and they beat out a veteran incumbent, then they will get the start, but it won't be handed to them.
Players cannot be resigned to contract extensions until after their 3rd season in the league. One way to maximize value for some players is to sign them to contract extensions BEFORE they become starters. This relegates them to backup duty for 3 years, until they become eligible for extension, but then allows a second contract at a reduced rate. Since there is not a lot of tape on a player, we are in the best position to offer a reasonable but fair contract. We did this with Kilgore this off-season, and may try to do this with Looney next year. Dial is another player that may fit this model. Assuming Kilgore gets the start at center this year, we will have paid a total of $7M for 7 years of service, 3 as a backup and 4 as a starter. Depending on how he performs, this could end up being a great bargain.
This only works with the later round picks as having the early picks relegated to backup roles is cost prohibitive. If you see a draft pick that is good, but you know is injured and will miss a season, don't expect us to draft them before the 4th round. Tank Carradine, our 2nd pick from 2013, ended up missing his first season, but I don't think that was expected. We tried to activate him last season, but he had complications with his ACL that kept him out longer than expected.
So, nation, what do you think? What should our strategy be this year? How do you see the value groups for 2014 and how do we maximize our draft? When do we veer off these charts, and what should drive our decisions when we do?
Also, if you made it this far, thanks for reading. If you have additional information to provide, or a different perspective, please add to this. I will be happy to edit this if I totally butchered something. I know there is nothing new here, and I've simplified a few things, but I enjoyed checking assumptions and validating what is out there, and if you haven't figured it out, I like charts.