Free-Agency Cost-Benefit Analysis

The original Bill James definition was "readily available talent", like street free agents. For DYAR, Aaron uses average backups as the standard. I'm not sure exactly what Andrew is using here.

(The reason for using replacement level is that average seasons have value. An average player for ten years is more valuable than a slightly above-average player for one year, but "value over average" doesn't reflect that.)

OTOH, the Dolphins have 22m in dead money this year, 22 + 6 = 28. So it's possible Miami will be able to survive in not much worse shape as this year.

------
Who, me?

I see it as risk/reward. Those negative AV numbers represent both high risk but also often high reward. Taking a lot of risk with little reward seems the worst. You are taking a great risk signing a high AV guy long term since he might get hurt and you are overpaying. Sure over time usually the cap goes up and a high contract this year will often become a reasonable value in five years do to the rising cap. Of course players also have a high injury rate in the NFL so a long term high price deal has the risk that the player will have zero value. Teams also void/ renegotiate late terms in contracts all the time meaning the future of long term deals can be murky. New England has chosen a low risk, but low reward path this year. They signed players for short term meaning they know exactly what they are getting and what the cap will look like, but they are also not likely getting more than a league average starter. We all know risk contracts that have bombed out in the NFL and risk contract that haven't, but I don't know how often they bomb. The number present here also doesn't include several players Miami has signed that aren't high risk high reward... The analysis as it stands left a lot out. If it isn't indexed to the rising cap numbers and contract escalation which most GMs factor into contracts then I'm not sure it has a lot of value or not.

In a way, the numbers don't tell us anything much about the Bradford trade.

They sure do tell us something. Specifically, they tell us that unless a guy who's been average or worse for five years suddenly becomes Peyton Manning, we're screwed.

If Kelly is right about Bradford's ability to stay healthy and play effectively, it's a great move.

Not really. I mean, yeah, the team would benefit in some way from the trade if Bradford were great, sure. But strictly speaking, that'd be more accurately characterized as a lucky move, given what we know about the players involved.

Put it to you this way: if I went to a casino and put my entire paycheck on 00, it would pay out 35 to 1 if I won. By all accounts, that seems like a great move if you just look at the result. But there's a reason financial advisers don't tell people to save for retirement by playing roulette. And it's the same reason that a rational NFL team would not make the trade that the Eagles did for Bradford: the downside is far too high, and the upside far too unlikely, for the decision to be justified under any reasonable appraisal of the evidence.

what trade WOULDN'T be great if you get a player who wildly exceeds the performance he demonstrated over the course of half a decade in the NFL? Every trade can be judged a "success" if you think of it in terms of "he's been below average so far... but if he's suddenly among the best in the league, the trade will DEFINITELY be worth its horrible, horrible value."

Exactly! Sure, if you look at everything in the best possible light, and make favorable assumptions about the future that have no basis in reality, then of course the trade looks good.

Everyone says that the jury is still out on all of Chip Kelly's personnel moves. But honestly, if two freight trains are barreling down the same track towards each other at 60 mph, I don't need to wait until after they crash to be angry at the idiot who sent them hurtling towards each other.

Any player *could* turn out to be amazing. Bradford has been in the league 6 years with 0 probowls. How many 3+ probowl QBs did that? He has spent about half his time injured, and the other half looking middling.

Foles has spent about 1/3 of his time on the bench, 1/3 looking like Aaron Rodgers, and 1/3 looking middling.

The risk of every trade has to be factored into the cost. What odds would you place on Bradford being voted to 3 probowls?

If we follow your logic to it's reductio ad absurdum, a team should literally trade anything asked for any semi-promising QB since you have to get that QB to compete for a superbowl. Of course, the team will then have literally nothing left when they find their great QB (and like Elway or Manning will probably refuse to play for them).

This is the kind of insane, utterly magical thinking I've seen everywhere about this trade. There is zero AND ONLY ZERO evidence so far that Sam Bradford is "elite." Zero. His highest ranking by DVOA was 14th. His other rankings range from "Not Elite" to "Absolutely, terribly, horribly utterly Not Elite" to put it in the ridiculous terms of "eliteness." Now, I'll admit that now that he's surrounded by talent like Riley Cooper, Zach Ertz and Josh Huff he might finally have the weapons to break out!

There is an 8 game stretch where Foles was not only "elite" but as good as any HOF QB of the last 15 years. I mean... what people are suggesting about this trade is just SO preposterous, I literally can't imagine what folks wouldn't be ok with Kelly doing. If he had made the same trade for... I don't even know who. So few QB's have as much cap-space downside, as extensive a history of mediocrity and as many injury problems so you can't even find another QB as bad as Bradford to have done this with. Even Johnny Manziel or Blaine Gabbert would be more reasonable because they're younger, cheaper, have less of a history of injury, were drafted highly and are more unproven (that is to say, we KNOW what Bradford is - we have half a decade of the NFL to look at, so less proven players you might be able to talk upside)

There's this magic idea that "Kelly sees something" and has found his QB so whatever kind of trade he makes is justified. It's bonkers. Utterly bonkers.

Without Nick Foles as his QB, Kelly has never fielded a passing offense in the top half of the league. Vick, Barkley and Sanchez all led terrible, NY Jets level passing offenses. There is no reason to trust Kelly's talent evaluation or the ability of mediocre QB's to perform transcendently under Kelly.

The only argument Kelly has in his favor was the performance of Nick Foles. And Foles is gone, for a second round pick and about a $12 million difference in cap space. That's an utter disaster if you're not an insane person.

Elite quarterbacks are extremely valuable.

Agreed.

Elite quarterbacks are hard to get in the draft

True (although they are even harder to get outside the draft).

Elite quarterbacks rarely change teams

Also true.

So if you have a shot at a potentially elite QB, and it's not ridiculously expensive, then you take it.

With you so far.

If Kelly thinks that Bradford has a decent shot at being elite,

Then he’s out of his fucking mind! Seriously, he’d have to be crazy. Why do you think we’re all so upset? I mean, how many QBs in the last 40 years have spent the first five years of their career sucking before they finally got their shit together and became elite? Because that’s what would have to happen here for this trade to be good for the Eagles. Rich Gannon, maybe? That’s the only one that I can think of.

So, yeah, if Bradford is Rich Gannon 2.0, then the trade would be okay. In the much, much more likely event that he’s not a once in a lifetime outlier, then the trade is a disaster.

Now, I know what you’re thinking, and you’re right. “Chip Kelly is a highly paid professional coach who spends a large portion of his time breaking down film of NFL players, and you’re just a fan whose only experience watching game film comes from drunkenly yelling at the TV at your local sports bar. Isn’t it reasonable to believe that Kelly just knows way more than you do about evaluating QBs?”

And it totally is. He’s probably forgotten more about the subject than I will ever know. But here’s the thing – I’m not the only one whose evaluation of Bradford is radically different than Kelly’s. Every other team that would’ve benefited from an elite QB, including several that have seen Bradford play against them, also passed on him*. So, not only would Kelly have to be much better than I am at evaluating QBs (which wouldn’t take much, frankly), he would also have to be much better than the highly paid professionals in charge of a couple dozen other NFL teams. And I see no reason to assume that.

For one thing, the only time he had any legitimate competition for the starting QB spot, he made the wrong call. Otherwise, Foles would’ve been the starter coming out of training camp in 2013, instead of getting the spot only after a 2-3 start.

*Except the Cleveland Browns, who offered a first round pick for Bradford. Oddly enough, I think this actually strengthens my case, given that the last three QBs the Browns have spent a first rounder on are Johnny Manziel, Brandon Weeden, and Brady Quinn.

Thinking out loud, and unwilling to do the work:

There's presumably a relationship between a team's total AV and their record. The next question is for teams with the same total AV, do different distributions correspond to different records? Do teams with high-AV quarterbacks have better records? Do teams with more AV concentrated in 2 (or 3, or whatever) players have better records? Etc. This doesn't need to be a thought experiment; p-f-r should have the data.

The problem is there's clearly a breaking point there. You can't have twenty 2 AV guys. For something like what you're doing here, that could be accounted for by having 2 baselines - one for "projected starters" and one for "projected bench player." There's subjectivity there (and human time cost), of course, which is the problem.

But in addition, the "18 AV" guy doesn't become available... basically ever, but the 8 AV guys probably show up every other year or so. As a GM, your goal isn't to maximize the AV contribution of each player - it's to maximize the AV contribution of the entire team. Which means the "18 AV + 2x replacement" is more valuable than the "3x 8 AV" situation (provided you have the 18 AV guy for longer than the current year) because you can do the "replacement -> 8 AV" upgrade more frequently than the "8 AV -> 18 AV".

I think without a great player to dictate to the other team you have a much harder time of things generally speaking.

Well, the NWE answer is clearly option 1; find a HOF QB and then get multiple B players, rather than a couple of A's, and a bunch of C's. Seems to work.
Unfortunately, the Raiders have been trying the second alternative, and even worse, failing to get the A's.

Yeah, that's obviously something that's missing from that graph as well, which is why age-adjusting things like Andrew's done here is really important.

To be honest, that graph is literally the easiest thing you can do: look at net AV change with free agents versus SRS, and both AV and SRS are probably the easiest measures of player and team strength you can possibly make. And it still does clearly show a correlation. A weak one, sure: but given that both the measures are incredibly poor, and the assumptions are clearly limited (that is, expecting previous performance to equate to future performance), the underlying effect would have to be quite strong to overcome that.