SITE NEWS: We are moving all of our site and company news into a single blog for Sports-Reference.com. We'll tag all PFR content, so you can quickly and easily find the content you want.
Also, our existing PFR blog rss feed will be redirected to the new site's feed.
Pro-Football-Reference.com » Sports Reference
For more from Chase and Jason, check out their work at Football Perspective and The Big Lead.
ShareThisFifth-round picks
Yesterday I asked whether you, were you sitting in Bill Parcells' chair (and were such a thing not pure fantasy), would exchange the first overall pick in the draft for all 32 fifth round picks. Some good comments followed. As promised, I'll give a general overview of how often starters and stars emerge from the fifth round.
I looked at all fifth round picks between 1990 and 1999. That keeps the data somewhat recent while also leaving time for all the players being looked at to have finished at least most of their careers. Using some quick ratings based on my almost-finished-but-not-yet-released approximate value formula, here are the best fifth round picks of the 90s. 'St' is the number of seasons he was his team's main starter at his position and PB is the number of pro bowls he made.
Player St PB =========================== Mark Brunell 11 3 Zach Thomas 11 7 La'Roi Glover 10 6 Joe Horn 8 4 Bryan Cox 10 3 Rob Burnett 10 1 Gary Walker 10 2 Larry Centers 8 3 Merton Hanks 7 4 Fred Miller 10 0 Rodney Harrison 10 2 Ben Coates 7 5 Santana Dotson 9 0 Chad Bratzke 7 0 Ed McDaniel 7 1 Dorsey Levens 3 1 Sam Garnes 7 0 Benji Olson 9 0 Joe Bowden 5 0 Roderick Coleman 4 1 Eric Barton 5 0 Stephen Boyd 4 2 Torrance Small 3 0 Chris Gray 11 0 Adrian Murrell 5 0 Brian Mitchell 0 1 Marcus Coleman 7 0 Norman Hand 6 0 Tony Williams 7 0 Lee Flowers 5 0 Calvin Williams 6 0 Jeff Mitchell 8 0 Chris Villarrial 10 0 Barry Foster 4 2 Charles Mincy 3 0 John Wade 7 0 Ian Beckles 9 0 Rich Owens 3 0 Shane Burton 2 0 Jay Foreman 4 0 Steve Martin 3 0 John Holecek 4 0 Gene Williams 4 0 Mike Rosenthal 3 0 Jerry Azumah 3 1 Everett Lindsay 3 0 Corey Bradford 2 0 Darren Hambrick 4 0 Oliver Ross 3 0 Jermaine Lewis 2 2
On the other side of things, here are the 50 top-five picks from the same period, a group that should form a reasonable approximation of the range of possibilities for a first overall pick.
Player St PB =========================== Peyton Manning 10 8 Marshall Faulk 11 7 Edgerrin James 8 4 Steve McNair 10 3 Cortez Kennedy 9 8 Drew Bledsoe 12 4 Donovan McNabb 8 5 Junior Seau 16 12 Orlando Pace 10 7 Keyshawn Johnson 11 3 Kerry Collins 10 1 Simeon Rice 11 3 Willie McGinest 12 2 Garrison Hearst 7 2 Jonathan Ogden 12 11 Peter Boulware 7 4 Jeff George 9 0 Tony Boselli 6 5 Russell Maryland 10 1 Kevin Hardy 9 1 Charles Woodson 9 4 Ricky Williams 5 1 Marvin Jones 9 0 Sean Gilbert 8 1 Dan Wilkinson 12 0 Shawn Springs 9 1 Terrell Buckley 8 0 Todd Lyght 11 1 Eric Turner 7 2 Darrell Russell 5 2 Michael Westbrook 6 0 John Copeland 7 0 Mike Croel 6 0 Rick Mirer 5 0 Tim Couch 3 0 Quentin Coryatt 4 0 Bryant Westbrook 3 0 Cedric Jones 2 0 Keith McCants 2 0 Desmond Howard 1 1 Blair Thomas 3 0 Steve Emtman 1 0 Ki-Jana Carter 0 0 Andre Wadsworth 2 0 Curtis Enis 2 0 Heath Shuler 2 0 Trev Alberts 0 0 Bruce Pickens 0 0 Ryan Leaf 1 0 Akili Smith 1 0
Let's take a look at the best three fifth-round picks of each season of the 90s:
1990 ==== Rob Burnett 10 1 Larry Centers 8 3 Brian Mitchell 0 1 1991 ==== Bryan Cox 10 3 Merton Hanks 7 4 Ben Coates 7 5 1992 ==== Santana Dotson 9 0 Ed McDaniel 7 1 Joe Bowden 5 0 1993 ==== Mark Brunell 11 3 Chris Gray 11 0 Adrian Murrell 5 0 1994 ==== Rodney Harrison 10 2 Chad Bratzke 7 0 Dorsey Levens 3 1 1995 ==== Gary Walker 10 2 Stephen Boyd 4 2 Norman Hand 6 0 1996 ==== Zach Thomas 11 7 La'Roi Glover 10 6 Joe Horn 8 4 1997 ==== Sam Garnes 7 0 Tony Williams 7 0 Jeff Mitchell 8 0 1998 ==== Benji Olson 9 0 John Wade 7 0 Corey Bradford 2 0 1999 ==== Roderick Coleman 4 1 Eric Barton 5 0 Jay Foreman 4 0
Looking at these lists, it seems clear that if --- and this is a pretty big if, which is why I threw in the second of the two modifications in the last post --- you can actually identify the keepers in your group of 32, the fifth-rounders are a much better bet. Groups like Cox/Hanks/Coates and Thomas/Glover/Horn show that your group has as much upside as the first overall selection without the downside risk of Akili Smith or Ryan Leaf. At worst, you're going to get three players who can start for five years.
But there's more to it than that, as was detailed in an astute comment by JKL:
My equation is not simply #1 overall pick vs. 32 5th round picks.
It is #1 overall pick + [replacement level street free agents that would fill the roster spots otherwise occupied by 5th rounders] vs. 32 5th round picks.
He's right that the key issue is not raw value but marginal value. The value of a fifth-round pick is not the expected value of the player you'd get with that pick. It's the difference between the expected value of the player you'd get with that pick and the player you'd get off the street if you didn't have that pick.
To be more specific, the question isn't whether fifth-round pick Mark Brunell is better than nothing. It's whether he's better than undrafted Chris Leak. Or Brock Berlin. Or Mark Hartsell. Or Jason White. Or Tony Romo. Hmmmm.
However, I think JKL might be overestimating the quality of the street free agents you could acquire to replace your foregone 32 fifth-rounders. It is true that the top undrafted free agents are probably of nearly the same caliber as the top fifth-rounders. But the beauty of having the entire fifth round is that you would be locking down the exclusive rights to exactly the guys you wanted. Roughly speaking, the fifth round gives you players number 130 through 160. If you replace those guys with undrafted free agents, you don't get players 225 through 257. You'd get 1/32 of players 225 through 1180, or something along those lines. While the expected difference between player 130 and player 225 probably isn't huge, the difference between player number 150 and player 450 probably is, and the difference between player 160 and player 1180 definitely is. All those differences add up to quite a bit.
After all this, I still don't have an answer, but I think I'd lean toward the fifth-rounders, assuming I could have a year to evaluate them.
This entry was posted on Thursday, March 13th, 2008 at 8:18 am and is filed under Approximate Value, General, NFL Draft. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.
I didn't realize that we were expanding it to more than 32 so you get to keep both LaRoi Glover and Joe Horn from the same draft year.
In reviewing your list, doesn't the answer to your hypo also depend on team needs? Lots of linebackers, safeties, and defensive tackles, as well as interior offensive linemen can be had there, at least enough to say that it is non-trivial and you are likely to address these positions by throwing numbers at it.
At quarterback (Brunell and nothing else), both tackle positions (Miller and a bunch of T-G journeymen) and wide receiver (Horn, followed by some okay starters), not so much.
If you are drafting first, is it because you have vastly inferior guards, centers, and linebackers, or because your quarterback and tackles stink? But if I was fortunate enough to have a decent quarterback and my tackle position was set, I would probably switch my answer.
But three of the top four on the list of fifth rounders were with their original drafting team for at least a year as a backup, before moving to another team and starring. (Brunell, Glover, Horn).
I'm pretty sure I was player 1180 when I graduated in 1996 as the team cameraman, so there is definitely a dropoff. I did have upside to add to my frame to the tune of about 80 pounds though.
Let's look at this another way. First, let's assume that we are a competent GM: we're not going to pick a quarterback unless Lewin tells us he's going to be good, and we're not going to take a player whose position is insufficiently valuable to warrant first overall money, or which the evidence says cannot be evaluated well enough to justify a top five pick. We'll therefore look at the first quarterback, left tackle, wide receiver, or pass rusher taken in each draft, excluding quarterbacks that starts/completion percentage would lead us to reject. That gives us:
1999: Donovan McNabb
1998: Peyton Manning
1997: Orlando Pace
1996: Keyshawn Johnson
1995: Tony Boselli
1994: Willie McGinest
1993: John Copeland
1992: Desmond Howard
1991: No suitable player in top 5
1990: No suitable player in top 5
NFL teams just had a very different (and mistaken) idea about what to do with high draft picks in the early nineties.
I'm going to (subjectively) rank the 10 5th round combos and 8 suggested first overall picks.
1. Peyton Manning
2. Orlando Pace
3. 1996 5th round
4. Donovan McNabb
5. 1991 5th round
6. Tony Boselli
7. Willie McGinnest
8. 1993 5th round
9. Keyshawn Johnson
10. 1995 5th round
In other words, 75% of the pseudo-#1 picks were better than at least 60% of the 5th rounds. 25% were better than any of them. None of the fifth rounds included a franchise quarterback. The #1s included two, and two franchise left tackles.
I would also note that by going back before the inception of the cap you cease to be comparing apples with apples, and that even if your team does acquire 32 UDFAs, other teams won't, so you'll only be going down to around player 450, not 1180.
The realistic problem, of course, is the fact that you can't evaluate 32 fifth round picks in the offseason. You can barely have 32 fifth round picks on your team.
Even more importantly, you can't play them all during the season. How many 5th round quarterbacks does it take to get a first round quarterback? Well, since you can only play 1 QB at a time, if it takes 2 5th round QBs to equal 1 first round QB, and it takes you 1 year to evaluate that QB, it'll take you two years to evaluate both.
Functionally, this is why the studies like the Massey & Thaler paper are completely and utterly wrong. Teams go into the draft with a limited number of slots available on their team. They have to maximize the talent that will be produced by those open slots. If it's a low-number position (with few on the team - like QB, or RB) you have to go with the player with the absolute best possible potential, because you can't grab multiple and pick the best (because you have no space to do so).
While the expected difference between player 130 and player 225 probably isn’t huge, the difference between player number 150 and player 450 probably is, and the difference between player 160 and player 1180 definitely is
Not really, according to the Massey and Thaler paper. The value of undrafted picks vs. fifth round picks, based on their average free market value when they hit the open market (including those out of football) is pretty similar.
"While the expected difference between player 130 and player 225 probably isn’t huge, the difference between player number 150 and player 450 probably is, and the difference between player 160 and player 1180 definitely is."
Only in the relative sense, and even then, I'm unconvinced. And in absolute terms, they're both pretty unlikely to be good.
It's like when my friends were going to put $40 together to buy lottery tickets. I was like, "Guys, why are you wasting your money on lottery tickets? I have almost the same chance of winning the lottery as you do, and I don't even buy lottery tickets!" Of course, they protested that they were much, much more likely to win than I was. So I answered, "Not really. My chances of winning the lottery are zero. Your chances of winning the lottery are virtually zero. It's about the same."
I think, that if in the 32 picks, you were to draft a single player who would go on to have the greatest player at his position of his era, you would certainly be able to tell in a camp setting that you really had a great player on your hands.
There is no objective measure that leads the team with the first overall pick to take the best player in the draft. However, there are objective measures to "check" whether or not the player the team wants to take will be a good prospect at the NFL level.
Even with the good point that JKL made about marginal value, you are still more likely to pick that great talent with 32 5th round selections than you are with the 1st overall pick. The Colts did get Peyton Manning and the Rams did get Orlando Pace, but Brady, Montana, Thomas, and Harrison were all mid-late round picks, and even when you have to cut through all the fluff, it's hard for me to see how the value of multiple players, when at least one is likely to be as good if not better than the first overall selection is not greater than the "sure thing".
Would be interesting to expand the approximate ratings to 5th round and less. I'd guess that the player ranking in the 5th round is much more variable than the 1st few rounds-- players that reach the 5th have significant question marks, and the staffs will have very different answers to those questions. With 32 picks, you might be willing to take a few more of the high variability slots and get the Marcus Colston and Derek Andersons.
I think, that if in the 32 picks, you were to draft a single player who would go on to have the greatest player at his position of his era, you would certainly be able to tell in a camp setting that you really had a great player on your hands.
Exactly. That's why Brett Favre totally beat out Chris Miller and Billy Joe Tolliver for the starting QB spot after his rookie year, instead of getting traded to the Packers...oh wait.
Favre notwithstanding, even if you were able to tell if you had a great player on your hands, you would only get the chance to do that when a really, really great player gets drafted in the 5th round, and that doesn't happen every year, especially not at QB.
Since 1995, the best QB to be drafted in the 5th round was...AJ Feeley? Maybe Troy Smith, depending on how he does? I'm not even sure, but of the 17 QBs drafted in the 5th round in the last decade or so, none have been above average starters for a significant amount of time. That's not good if you're trying to find a QB, which most teams that get the #1 overall pick need to do.
And you can't just say, "Oh, some of the 32 players you pick are bound to be really great." Some years, yes. Some years, no. And if you don't get any great players, your task becomes separating the above average/good players from the below average/bad players, which is almost impossible to do in one year, much less one training camp.
Well, since you can only play 1 QB at a time, if it takes 2 5th round QBs to equal 1 first round QB, and it takes you 1 year to evaluate that QB, it’ll take you two years to evaluate both.
Or, if you're the Bears, it'll take you 5 years to evaluate one, and three years to evaluate the other, and you'll still have no idea which one to start. Seriously, is there even a single above average QB in the entire NFC North now that Favre's retired? Is there any other division in the NFL that's so well stocked at every other position, or so utterly lacking at QB?
I think, that if in the 32 picks, you were to draft a single player who would go on to have the greatest player at his position of his era, you would certainly be able to tell in a camp setting that you really had a great player on your hands.
No way. Absolutely no way.
Suppose it's a quarterback. Where are you going to play him at camp? First string? Second string? Third string? If he comes in at first string and struggles, is it because you threw him in too high over his head? If he comes in at third string and excels, is it because his competition is weak? If he struggles at third string, is it because the other players are poor?
Heck, if you've got 32 of those players in camp, you don't really have enough room for players who don't suck to evaluate them against the 32.
Plus, in the end, you still won't have seen those 32 players against anyone other than the other 48 in training camp.
Suppose you've got a DE that looks great in training camp. Maybe he just has the OT on your team's number - in the regular season, he could just fizzle out. Whereas the DE who didn't look quite as stellar (but still decent) might be a monster because he's far more capable against the OTs you don't have.
Favre is, of course, the best example, but he's certainly not the only one. And the chance that you'll get the "greatest player at his position of his era" in 32 5th round picks is essentially zero.
I think, that if in the 32 picks, you were to draft a single player who would go on to have the greatest player at his position of his era, you would certainly be able to tell in a camp setting that you really had a great player on your hands.
In addition to the problem of identifying good players who are playing poorly in camp, there's also the problem of false positives. After all, sometimes terrible players look great in camp, and even through parts of the regular season, until other teams figure out how to deal with them. So how do you know which of the "great talents" to hold onto? You can't keep 6 QBs on your roster and just wait until one of them distinguishes himself in the regular season. Half the NFC North has been trying this the last couple years, and it just doesn't work.
"Bears fans are crazy about Rex Grossman."
"What do you mean, crazy?"
"...Crazy!"