In this week's Varsity Numbers, Bill Connelly takes a page out of baseball's playbook and attempts to isolate power from efficiency.
21 Mar 2012
by Danny Tuccitto
Welcome back for the second half of our 2011 Wisdom of Crowds (heretofore "WoC") review. If you didn't get a chance to read last week's review of quarterback and running back projections, it's here.
Today, I'll be looking at how well our Twitter followers fared in projecting fantasy stats for wide receivers. If you need background on how the process works, or want to see how I discussed the projections at the time, you can read the original articles here, here, and here. The important details to keep in mind today are that all projections assumed a 16-game season, and that I purposefully chose players whose fantasy prospects were reasonably difficult to judge back in August.
Below, I've listed each player's crowdsourced projection (with error range) and his actual performance. For those players who missed games, I'll also list performance prorated to a 16-game season. Compared to the uncanny accuracy of projections for last week's group, wide receiver projections ended up being more of a mixed bag. Most results showed once again that WoC participants are a Carnac collective; others suggested they were the besotted sidekick cackling nearby. (That reminds me: Check your local Ticketmaster outlets for Carnac Collective's upcoming post-prog tour; Besotted Sidekicks opening.)
Projected: 73 ± 3 receptions, 953 ± 34 yards, 7 ± 0 TDs
Actual: 87 receptions, 967 yards, 6 TDs
Really, the only misfire related to Harvin's projection was a two-yard overestimation of his average reception length: 13.1 projected yards per catch versus 11.1 actual. Based on his 12.2 average in 2010, we can infer that the crowd thought it would increase by a full yard in 2011. Instead it decreased by that amount. It turns out that this was actually the second season in a row that Harvin's receiving average dropped by about one yard. For the purposes of evaluating his future fantasy prospects, that's potentially a red flag, so it's helpful to get at potential explanations.
As is often the case outside of ancient guys and a post-juicing David Boston, the best explanation for change in wideout performance is change in quarterback performance. Harvin's downward career trajectory with respect to yards per catch is a vivid illustration. Specifically, take a gander at the table below, which shows his receiving average broken down by quarterback and year (except for two receptions from Tarvaris Jackson in 2009).
Harvin was better with 2009 Favre than he was with 2010 Favre, he was better with 2010 Favre than he was with 2010 Webb, and he was better with 2011 Ponder than he was with 2011 McNabb or 2011 Webb. As his quarterbacks go, so goes Percy Harvin.
Also boding well for Harvin is that he finally played a full season last year, and he scored over 80 percent of his fantasy points in Games 9-16 (i.e., shortly after Ponder became the starter).
Projected: 69 ± 3 receptions, 991 ± 42 yards, 8 ± 1 TDs
Actual: 70 receptions, 966 yards, 5 TDs
Prorated: 74 receptions, 1,030 yards, 5 TDs
Even though it looks spot-on at first glance, evaluating the accuracy of this projection is somewhat difficult, mainly because the assumption that Lloyd would play the entire season with the Broncos did not hold. We could just ignore the trade, and simply prorate his four-game Denver stats to a full season. However, his departure coincided with, and was probably a direct result of, Tebowgenesis, so prorating 2011 stats he accumulated with Kyle Orton doesn't make sense.
Even withholding judgment in terms of the overall projection, I can say the crowd astutely recognized the unsustainability of Lloyd's receiving average in 2010. The projection worked out to 14.4 yards per reception, and Lloyd ended the season at 13.9. If we split it between teams, his average was significantly higher with Denver (14.9) than with St. Louis (13.4), but the important thing is that both were a far cry from the 18.8 yards per reception he posted the previous season.
Given that Lloyd's receiving average has consistently hovered in the 14-to-15 yards per catch range his entire non-2010 career, it's reasonable to think that's the most likely outcome going forward. Maybe we can boost that by a yard or two because of his McDaniels-aided move to New England, but I doubt we'll see a return to Mike Wallace territory. On the other hand, it's almost inconceivable that Lloyd will be limited to five touchdowns in that offense. Expect Lloyd's projection to be significantly higher going into next season.
Projected: 67 ± 3 receptions, 897 ± 35 yards, 6 ± 0 TDs
Actual: 15 receptions, 276 yards, 1 TDs
Prorated: 16 receptions, 294 yards, 1 TDs
This is the first of two laughably bad projections exemplifying that wide receiver stats are more unpredictable than those of other offensive skill positions. Ochocinco's fantasy season ended up being such an underachievement that he barely made it halfway to the crowd's worst-case scenario statline (35 receptions, 500 yards, 3 TDs).
Some offered the explanation that he fell into Bill Belichick's doghouse, but I prefer the tried and true, "One day the guy just wakes up, and can't get it done anymore." To see what the future might hold, I ran Ochocinco's career trajectory over the past three years through our similarity scores system, and found that 32 of the 50 most similar three-year stretches of early-30s decline involved wideouts who never played another NFL game. Of the 18 whose careers did continue, 14 only played one more season. In other words, history says there's about a 95 percent chance that Ochocinco's out of the league by 2014.
Projected: 73 ± 2 receptions, 980 ± 26 yards, 8 ± 0 TDs
Actual: 63 receptions, 928 yards, 9 TDs
Prorated: 67 receptions, 989 yards, 9 TDs
It's outcomes like this that make me want to just crowdsource every player on my fantasy draft board, and cut you guys in on the profits.
Projected: 76 ± 2 receptions, 1,101 ± 30 yards, 10 ± 1 TDs
Actual: 65 receptions, 771 yards, 3 TDs
This season was definitely a disappointment for anyone who drafted Williams. He showed flashes of 2010 at times (e.g., 28 points in Weeks 10 and 11), but was a fantasy disaster most of the time. In trying to figure out what happened, it's tempting to play the worse-quarterback-play card like we did with Harvin. However, as I discussed in our review of WoC quarterbacks, although Josh Freeman was much worse in 2011 from an advanced stats perspective, there wasn't as big of a decline in his standard stats, and -- last I checked -- fantasy football platforms have yet to offer DVOA-based scoring systems. Therefore, Williams' 42 percent decline in fantasy points can't be blamed primarily on quarterback play when Freeman's point total only dropped by 18 percent.
Instead, my guess is that there were multiple causes. First, he just isn't as good as his standard stats in 2010 suggested. Lost in the hoopla of a No. 11 fantasy ranking during his rookie season was that he ranked 56th among wideouts in DYAR and 65th in DVOA. In addition, his 11 touchdowns as a rookie was the first double-digit total since Randy Moss circa 1998. Unlike Moss, however, no one would mistake Williams for a transcendent red zone and vertical talent.
Second, he was probably affected by increased attention in coverage. Finally, it's been reported that the poor work ethic leading to Williams' draft-day free-fall re-emerged last season. If those reports are true, then they represent a giant red flag for his fantasy future. Also representing a giant red flag for his fantasy future: Vincent Jackson.
Projected: 64 ± 2 receptions, 908 ± 29 yards, 7 ± 0 TDs
Actual: 32 receptions, 484 yards, 2 TDs
Prorated: 56 receptions, 860 yards, 3 TDs
Tweeting Rice's name to our followers was almost unfair. Trying to predict his stats under the assumption that he doesn't miss a game is like trying to predict Mike Tanier's alcohol intake during an Eagles game under the assumption that he doesn't change the channel in disgust before it's over. But hey, I was just curious what people were thinking.
Amazingly, even with that impossible constraint, the crowd still came reasonably close to Rice's prorated stats, especially with respect to receptions and yardage.
Looking ahead to next season, I think it's safe to say at this point that only divine intervention of some sort will allow Rice to play all 16 games. Suffering three concussions in a calendar year and having surgery on both shoulders earlier this offseason certainly doesn't help matters. What does potentially help matters is Seattle's upgrade from Tarvaris Jackson to Matt Flynn at starting quarterback. The jury's still out on Flynn, but it seems unlikely that he'll end up being worse than Jackson. Rice's injury risk is too high to rely on him heavily for fantasy purposes, but if he can manage to play something like 12 games with Flynn throwing passes, he wouldn't be a horrible option at WR4.
Projected: 56 ± 2 receptions, 746 ± 30 yards, 5 ± 0 TDs
Actual: 54 receptions, 959 yards, 8 TDs
Prorated: 66 receptions, 1,180 yards, 9 TDs
Projected: 49 ± 2 receptions, 651 ± 46 yards, 4 ± 0 TDs
Actual: 65 receptions, 1,057 yards, 7 TDs
Prorated: 69 receptions, 1,127 yards, 7 TDs
I grouped these two together because crowdsourcing failed similarly for both. Jones and Green were highly drafted rookie wide receivers, and we all know how that usually turns out. It's tough to be enthusiastic when faced with such a poor historical track record, so I'll cut you guys some slack here.
What's interesting to me about their stats is how similar they were despite entering entirely different team situations. Jones joined a playoff team with a good running game, an established above-average quarterback, a legitimate No. 1 wide receiver already on the roster, and future Hall of Famer Benjamin Button at tight end. In contrast, Green got drafted by a four-win team with an offense that included the least valuable 300-carry back in the league, a fellow rookie at quarterback, and a fellow wideout whose main career goal is to top Nate Newton's NFL-record 213-pound pot bust.
Because of this, I'm inclined to view Green's 2011 season as far more impressive than Jones'. As a side note, both DVOA and DYAR agree with me for more quantitative reasons.
Projected: 73 ± 3 receptions, 969 ± 38 yards, 7 ± 0 TDs
Actual: 76 receptions, 1,004 yards, 7 TDs
Steve Johnson is our Francis Galton Hat-Tip of the Week. Like Colt McCoy before him, the crowd apparently knows Johnson far more intimately then he's probably comfortable with. Or perhaps there's just something about Bills and Browns football that makes people say, "C'mon, we already know how this one ends."
Projected: 63 ± 3 receptions, 796 ± 52 yards, 6 ± 1 TDs
Actual: 51 receptions, 586 yards, 2 TDs
Prior to the season, Roberts was on everyone's list of sleepers -- or maybe just mine. With Steve Breaston leaving for Kansas City, Roberts and Early Doucet had a training camp competition to see who would start across from Larry Fitzgerald. That's some valuable NFL real estate, so Roberts' fantasy stock appreciated considerably when he initially won the job. Unfortunately, that meant a lot of people ended up buying high only to sell low a few weeks later; you know, kind of like what Colts season ticket holders were doing last fall.
In terms of his prospects for 2012, the signs are mostly positive. Roberts' age, experience, production last season (albeit mediocre), and potential opportunity fit Matt Waldman's reasonably predictive profile for breakout fantasy wide receivers. In addition, he was nearly three times as productive from a fantasy perspective during the second half of the season (51 points) than the first (18 points). That improvement coincided with John Skelton becoming the starter, but I think it was more of unit-wide phenomenon than a quarterback-specific one. Why? Because Roberts had his best game of the season (six catches for 111 yards) with Kevin Kolb throwing the passes in Week 13.
I know there's the whole "fool me twice" danger here, but Roberts is worth keeping on your radar, especially if Arizona doesn't use a high draft pick on a wideout.
Projected: 61 ± 3 receptions, 853 ± 55 yards, 7 ± 1 TDs
Actual: 12 receptions, 150 yards, 0 TDs
Prorated: 32 receptions, 400 yards, 0 TDs
You may see this result and think it's not one of the crowd's shining moments. What you should really be thinking is that it wasn't one of mine. Behold this quote of mine from August while discussing KUBIAK and the crowd's projection: "If my life depended on it, I'd guess he beats them both, finishing somewhere between the average and best-case-scenario projections: something like 75-1,000-9." Your Football Outsiders fantasy guy, ladies and gentlemen.
You know how bad this prediction was for all involved? It's a waste of space to even bother discussing Sims-Walker's fantasy prospects in 2012.
Projected: 51 ± 4 receptions, 716 ± 50 yards, 6 ± 1 TDs
Actual: 45 receptions, 612 yards, 8 TDs
Let's end on a positive note. For all the unknowns about Burress heading into 2011, the crowd's projection ended up being in the ballpark of reality. His yardage total ended up being higher than anticipated, but his 13.6 yards per catch was well within the margin of error (12.1 to 16.2).
At this point, Burress isn't anything more than a slow-moving, tall guy who will probably help a real football team (e.g., as a red zone specialist) before he ever helps a fantasy football team again.
That's it for this year's WoC review. Now that I'm officially a veteran of the process, my sophomore season figures to be bigger and better -- unlike Mike Williams. Maybe I'll implement that idea about populating a WoC-based draft board. Maybe in next year's reviews, we can name an all-WoC fantasy team based on the crowd's most accurate projections. One thing I do know for sure, though: No more hyperbolic assertions about marginal receivers with hyphenated last names. I'm meekly glancing in your general vicinity, Darrius Heyward-Bey.
5 comments, Last at 22 Mar 2012, 2:00am by The Ninjalectual