FAYDPC

Dr. X again posts this from the Mean Regression Archives at Stanford (0-3) University:

"My life really hasn't been the same since the wildly enthusiastic response to my interception-adjusted yards-per-attempt metric. People talk about it on the Internets, and someone (not me) created this amusing graphic. It is gratifying to have made this small contributoin to western civilization. It so original that I am absolutely certain no other person could ever have thought of it.

"For many years I have ruminated on developing a similar metric for running backs. In some ways this should be easier. IAYPA can never evaluate a quarterback in isolation - quarterbacks rely on the line, the receivers, and the quality of the overall offensive scheme. So Montana and Young, throwing to Rice and playing in the West Coast Offense, get better scores than they might have had throwing to, say, Harold Carmichael and playing for Chuck Knox.

"A metric for running backs is purer - they rely almost exclusively on the line and their instincts. Sure, the coach 'diagrams' a 'play' for them, but if they call a trap play and there's a 350-lb defensive lineman standing where the hole is supposed to be... The running back is the only offensive player who has to improvise on virtually every play.

"But the technical challenges to developing such a metric are significant. Key issues include:

• Fumbles happen less often than interceptions (two or three times a season per RB, on average). This means that a given back's fumbles are less costly to the team than the QB's interceptions. We should also be careful about inferences from such tiny sample sizes. We should, but I don't plan to.
  
• Should we penalize the back for fumbles or fumbles lost? The back's agent would probably say the latter, since the number is smaller. But the metric I propose, Fumble-Adjusted Yards-per-Carry, penalizes fumbles rather than fumbles lost. My reasoning is that the recovery of a fumble has more to do with luck and team skill than anything about the running back. The central issue is that the ball is on the ground. So, we penalize the act of fumbling, not the eventual outcome.
• The size of the penalty needs to be adjusted, however. In IAYPA we penalize the QB 50 yards because an interception is, by definition, a loss of possession. The negative impact has a 100% chance of occurring. Fumblesby contrast, may or may not be lost. Last year about 1/2 of all fumbles turned into changes of possession. So, we incorporate a 50% probability of losing the fumble into our penalty. This cuts the penalty in half - from 50 to 25 yards.

"So, who was the greatest running back last year? Sean Alexander, of course. But he was not the FAYDPC champion. That was arch-enemy Tiki Barber, who ran for 1,860 yards and fumbled once for a FAYDPC of 5.1. Warrick Dunn (5.0), Tatum Bell (4.9), and Larry Johnson (4.7) also stand between Alexander (five fumbles) and the summit.

"But, really, this metric doesn't change much. The chart of backs with high yards-per-carry is about the same if you adjust for fumbles or if you don't.

"With one exception. A very important exception, as it turns out. The 2005 NFL leader in Yards-per-Carry was Michael Vick. But, after applying a penalty for his five fumbles in 102 attempts (Alexander had five fumbles in 370 attempts), Vick's ypc goes from 5.9 to a FAYDPC of 4.6 (NFL average=4.0), a huge tumble.

"I think this is a potentially important variable in our assessment of Vick - an assessment that has problematic for football analysts everywhere. We know he's not a good passer (below-average IAYPA of 4.7 last year). Now we know he's not a great runner either.

"Vick's team, the Atlanta Falcons, led the league in rushing in both 2004 and 2005. But they had never posted consecutive winning seasons before he arrived, and after an 8-8 record last year, they still haven't. It is too early to draw the curtain on this unique and masterful practitioner of the football arts. But these findings, along with the impact of age and accumulated injuries on his speed, suggest to me that we have already passed the peak.

"I hope I'm wrong about that - he is fun to watch."