09/14/2010 11:17 AM ET
By the Numbers: Luck or skill?
How much randomness is in a quarterback's numbers?
By Michael Salfino / SNY.tv
If you're relying on Alex Smith to carry your fantasy team at quarterback, your season may already be lost. (AP)

Last week, we demonstrated to my satisfaction and we hope yours that yards per rush (YPR) numbers far above or below the league average had a large element of randomness to them.

We anticipated that, in comparing YPR leaders and trailers to what they averaged the next year, we'd reasonably be able to attribute about 75 percent of a running back's output to randomness or team factors beyond his control. The result was that 77 percent of the success or failure of a running back was random, and just 23 percent was the result of "true skill." The takeaway last week was to pick up Matt Forte and Darren McFadden, which worked out really well for owners on Sunday, at least.

But Rotowire's Chris Liss challenged me when I was a guest on his fantasy football show on XM (Channel 147, where you can hear me at 1:30 p.m. ET every Friday) to test those running back results against something he knows I believe is less random and more inherently reliant on true skill -- quarterback yards per pass attempt (YPA).

Because I believe that YPA is a more player-dependent than team-dependent stat, my expectation was that YPA results would prove to be less influenced by factors outside the quarterback's control. Let's check the results after noting that the league average for YPA in 2008 and 2009 was about 6.6.

Rank
QB
YPA 2008/09
1Philip Rivers8.39/8.8
2Matt Schaub8.01/8.2
3Drew Brees7.98/8.5
4Jake Delhomme7.94/6.3
5Matt Ryan7.93/6.5
6Chad Pennington7.67/DNQ
7Kurt Warner7.66/7.3
8Tony Romo7.66/8.2
9Aaron Rodgers7.53/8.2
10Jay Cutler7.35/6.6

Four regressed, five improved. Their average YPA (average of their averages, not their combined passes) was 7.8 in '08. Last year, it was 7.6. Calculating the the same way we did for the running backs, the vast majority of the difference above the average seems sustainable and thus can more reasonably be attributed to true skill. About 83 percent of above-average YPA seems to be true skill. That's about the reverse of YPR for the backs, for whom true skill accounted for 29 percent of the difference.

Now the trailers:

Rank
QB
YPA 2008/09
23JaMarcus Russell 6.58/5.2
24Kerry Collins 6.45/5.7
25Jason Campbell 6.41/7.1
26Kyle Orton 6.39/7.0
27Dan Orlovsky 6.34/DNQ
28Seneca Wallace 6.33/5.8
29Tyler Thigpen 6.21/DNQ
30Marc Bulger 6.18/5.9
31Derek Anderson 5.71/4.9
32Ryan Fitzpatrick 5.12/6.3

Five got worse and three got better. Their average YPA was 6.2 in 2008 and 6.0 last year. No need to crunch the numbers. The guys who stink appear simply to stink relative to NFL quarterbacks. The population went from bad to worse. There seems to be very little that is random about the trailing group's 2008 or 2009 performance.

So while running-back YPR performance for leaders and trailers worked out to 77 percent random and 23 percent true skill, quarterback YPA performance seems about 10 percent random and 90 percent skill when you combine the results of the leaders and trailers.

Takeaway: the only way to beat the market on quarterbacks is to target YPA leaders who did not translate that efficiency into prodigious touchdown totals. Also avoid quarterbacks like Alex Smith who defied a poor YPA with their touchdown efficiency (which we suspect is much more random).

Targets this year are/were Donovan McNabb (seventh in YPA last year) and Eli Manning (ninth).

The one bad value that jumps out at me is Joe Flacco (14th), who was drafted as a 12-team starter just about everywhere. Yes, he is a developing player with better skill talent around him. But there's more skill-level risk with Flacco than his Average Draft Placement indicates.

Michael Salfino writes for the Wall Street Journal and Yahoo! and is a regular contributor to SNY.tv.
Write a Comment! Post a Comment