Bent, TheJetsBlog.comOnce this year’s free agency signing period got underway, the Jets moved fast to secure their top target, wide receiver Eric Decker. The addition of Decker bolsters the Jets’ much-maligned receiving corps, as he was one of the most productive receivers in the NFL over the past two seasons. However, most experts agree that it’s going to be difficult for him to replicate that kind of production now that he won’t have future hall-of-famer Peyton Manning throwing him the ball.
So, we can probably expect some kind of drop-off in terms of his statistical production. However, can we use statistical data from previous seasons to try and quantify the scale of the drop-off?
The answer, of course, is no. There are far too many variables at play that any numerical data analysis can’t account for. Still, that doesn’t mean we can’t look at what these variables are and what will have the biggest influence upon how close he can get to the anticipated numbers. That’s what we’ll do after the jump.
Football Perspective Study
The best place to start is with this article from Chase Stuart on Football Perspective. You may have read this before, because I linked to it when I did my Eric Decker BGA scouting report (where I discussed similar themes in terms of free agent wideouts signed by the Jets in the “Potential Downside” section).
While Stuart’s article attempts to do what I’ve already conceded isn’t possible and project how successful Decker will be by comparing him to similar players who switched teams at the equivalent point of their career, it does give some sense of the probability of a move like this working out.
Stuart’s conclusion was that, while there were a couple of high profile receivers who give “receivers moving teams in their prime” a bad name (notably Peerless Price and Alvin Harper), Decker isn’t a close enough comparison to enough players from the past to draw any definitive conclusions. Ultimately, Stuart seemed pretty optimistic about Decker’s chances of continuing to be a productive receiver as he transitions into a number one receiver role and felt it was a good move that provided an “enormous upgrade” at the position.
If you missed this article first time around, make sure you check it out this time.
Taking this analysis one step further is this analysis article from ESPN-owned “data journalism” site FiveThirtyEight.com. Neil Paine writes, not about Eric Decker, but about DeSean Jackson and what we can expect from him in Washington. This is still relevant to Decker, though, because he meets the same criteria as Jackson and therefore the data and conclusions drawn therefrom can also be applied to Decker in a discussion of what we can expect from him with the Jets.
Paine and Stuart are actually former colleagues who used to write for the Pro Football Reference blog. Stuart’s site does use statistics from PFR, although his Decker analysis was just based on fantasy numbers. Paine’s analysis, on the other hand, uses PFR’s AV metric to try and quantify a player’s influence in the season before they switched teams and how much of a drop-off there was in the following season. We’ll discuss the AV metric in the next section, but Paine concludes that there is usually a drop-off, although not necessarily a large one.
Paine’s final conclusion says that the data suggests Jackson is unlikely to emulate his career-best performance and therefore, by extension, we can assume he would draw the same conclusions about Decker. Again, though, the data makes it difficult to quantify how far such numbers would typically drop.
Just by way of an explanation, here’s how the AV (Approximate Value) metric works. What AV aims to do is, as the name suggests, assign a value to each player’s statistical contribution over the course of their career. Broadly speaking, each team has a certain amount of points to divide between all the players on the roster and this is weighted so, for example, an above average offense will have more points to divide between its players than an inferior offense.
Since it works that way, it measures your contribution with regard to all 32 teams rather than your own team. So, if Decker’s 2014 numbers are identical to his 2013 numbers, but he’s making a much bigger contribution to the Jets offense in terms of his proportion of the total yardage than he was in Denver, that won’t affect his AV much. As an example, Dustin Keller’s production in 2009 was slightly below his 2008 numbers (three catches and 13 yards less) and as a result his AV dropped from 6 to 5. However, as a proportion of total passing yardage his production jumped from 16% to 22%. If you can say a player’s value to the offense as a whole increased (and that’s what we’d expect as Decker moved from being a second or third option in Denver to being the Jets’ primary option), then maybe a study based on numbers quantifying a player’s contribution to their team would provide a more interesting set of results.
Whether or not you agree with the AV metric, it does go beyond the metrics from sites like Football Outsiders or Pro Football Focus in terms of historical context, because it goes all the way back to the beginning. This makes it particularly instructive in terms of projecting a player’s Hall-of-Fame worthiness. However, in terms of being a projection tool, it doesn’t currently appear to be any more useful than just using yardage totals.
Interestingly, there is a discussion on the methodology page about introducing a touchdown bonus into the AV numbers. Had they done that, then Decker’s AV over the past two years (24 touchdowns) would be that much higher and if we did use AV to try to put a number on the amount by which his production might drop off, then including a touchdown bonus co-efficient would produce a better projection.
Mining the Comments
In the above links, there’s wisdom to be found beyond the original articles themselves, as the 538 commentariat in particular raise several issues which Paine didn’t address in his article. These are worth a read too.
Issues raised include age, having to learn a new system, chemistry with the new quarterback, recent sample size and injuries (although Paine makes a reasonable rebuttal to this one). Another big one – one which clearly applies in Decker’s case – is that the offense and/or quarterback for the new team often isn’t as good as the one that the player is leaving. The point is also well-made that even if a player’s statistical production might drop, they might make up for that with the intangible value of how much the other players on the team benefit from the defensive attention the new guy draws.
Maybe the best point of all is raised not in the comments, but by Paine himself during the article, where he admits that you would expect a drop-off from any group of players that were pre-selected based on past performance. Since the criteria they were looking for was good receivers that changed teams, you could expect the following year’s performance from that same group to regress to the mean on the whole. Add in the other issues above – everybody being one year older, in a new system and the potential for injuries to eat into statistical production and you suddenly have a raft of reasons for the expected drop-off with the new team, some of which will not apply in each individual case.
The Absolute Best and Worst Case Scenario
When I think back over the years to a receiver switching teams in his prime, two examples immediately spring to mind, neither of which are included in either of the above studies (although one is mentioned in the comments).
For the worst case scenario, consider David Boston. Boston just fell short of the AV threshold (he was at 9, but the data was based on players with an AV of 10 or above), but he’s arguably the biggest cautionary tale in terms of big money wide receiver signings. He was coming off a 70 catch, seven touchdown season where he would have exceeded 1,000 yards (and had an AV of 10 or above) if he didn’t miss a couple of games. Miami acquired him and he would play just five times for them, catching four passes.
This was certainly an unusual set of circumstances. Miami had only given up a sixth-round pick for Boston, but they inherited the last six years of a massive $47m, seven-year contract. Boston was suspended for steroid abuse, tore up his knee and missed the whole season. He was then cut, re-signed and was totally ineffective in season two as he had bulked up far too much and lost too much speed and agility.
If the spirit of these studies is to consider when a player moves from the team where he had initial success, then Boston had arguably already made his “prime years move” when he signed for San Diego. That came after a season where he missed eight games, but he had a 98-catch, 1,598 yard season in the year prior to that. Average out the production over those two seasons and his performance in San Diego was what you’d usually expect based on the rest of this data: A slight drop-off. (For the record, he was traded for his laziness in practice and “moody personality”. I think we can guess where that moodiness came from.)
The Boston case was such a unique set of circumstances that it holds little predictive use. However, it does serve to underscore how difficult it is to make any kind of definitive projection because you never know what’s going to affect the outcome.
For your best case scenario, consider Santana Moss. Moss went from the Jets to Washington and exploded for career highs in receptions (84) and yardage (1,483). Moss isn’t included in the 538 dataset because, like Boston, he had achieved an AV of over 10 in a previous season, but not in the one preceding his move. (You might recall me writing several times before about how disappointed I was with Moss’s performances and effort during the 2004 season). Moss’s improvement in AV from his last year in New York to his first year in Washington was far better than the best two cases in Paine’s dataset (Brandon Marshall and Vincent Jackson).
There were a few reasons for this, but the biggest of these was scheme fit. Moss, one of the league’s top deep threats was leaving a team where the head coach and offensive coordinator’s conservative nature and the lack of arm strength of their quarterback meant that they hardly ever threw deep. That’s another factor not taken into account above. If anything, that should give us more confidence that Decker’s performance won’t deviate too far from expectations, because he’s a solid route runner that should be capable of getting open in any system.
As I said, there were a few reasons for this, so just for completeness, the others as I see them were as follows: First of all, Moss was traded, so unlike most of the players in this study, his final year with his old team wasn’t a contract year for him. That might explain what I considered to be disappointing effort that season. Secondly, he was bothered by a hamstring injury in the middle of the 2004 season, so perhaps that hurt his production.
On the basis of these two studies, which use similar but not identical methodology and data sets, we can surmise that a player in Decker’s situation often sees their statistical production drop off in their first year with their new team. Based on the data sets analyzed by Paine and Stuart, the average drop off doesn’t seem to be so significant that the Jets would end up disappointed with their financial outlay.
However, the most important takeaway from this is that Decker’s production might drop off by more than the average or by less than the average. Or maybe it will increase. There’s no way of knowing and there are plenty of factors at play. While we can predict some of these things with reasonable accuracy, there are others where it is simply impossible to know what will happen.
While there are a few things you could do to take this type of analysis one step further, it’s probably not worthwhile in the grand scheme of things. The Jets may hope they got a good one in Decker, but the fates will decide whether or not this proves to be a good move. While history may suggest there’s a good chance, there are no guarantees either way.