Wednesday, September 17, 2008

Chasing the Points


Easy to forgive these things when they work out

So as I was watching the Redskins vs. Saints game on Sunday the role of Washington's failed two point conversion was central in my mind. The decision to go for it at that time, the failure to pick it up and the ensuing points disparity could easily have cost the Redskins that game; that it did not, that the team could overcome the error and still put in a convincing come from behind win, is no reason not to look at game management and conventional wisdom as re: the two point conversion.

Here is the scenario: trailing 17-9 on their first possession of the third quarter the Redskins drove 83 yards in 10 plays over 4:43 of game time culminating in a nine yard left side run by Clinton Portis for a touchdown. Now down by two 17-15, still with more than five minutes to go in the third quarter, Redskins line up and go for the two, Saints defensive end Charles Grant gets a hit on Jason Campbell and his throw to Ladell Betts is incomplete, score remains 17-15.

Saints punt away their next possession then the Redskins three and out, a poor line drive 38 yard punt by rookie punter Durant Brooks lands in Reggie Bush's hands and he runs it back around the left side for a touchdown, with the extra point Saints now lead 24-15, a nine point differential.

Granted, the eight point differential that would have been in place would still require a touchdown plus two point conversion to tie, not at all guaranteed. It still however, would only have been a one score delta, not a two score delta.

Washington punted on their fourth quarter opening possession, then the Saints punted then the Redskins started their next possession with a 12 yard sack and you could hear the groans of Redskins fans all over the world: like the Giants game the team would weaky it another possession and then get the ball back with about four minutes to go, down two scores and not in any big hurry.

Well we all know how it ended, Jason Campbell led the team on a 94 yard drive on five passes and two runs ending with another Clinton Portis touchdown. Still down two Washington held New Orleans to a three and out, scored on the one play drive and then Chris Horton sealed it with an interception, his second of the day.

=====

So, if it all worked out then what is there to review? Well for starters right at the failed two point play itself color commentator Brian Billick who normally annoys me very much noted that it was too early to go for two, that the conversion cards coaches keep around should only be brought out for the fourth quarter. Hogs Haven and Curly R reader drWNC expressed a similar sentiment.

Why? Because as a coach when you start chasing points it has a way of coming back to haunt you. Better to go for the automatic one point and put the game back into the hands of your defense then to play the odds and go for two. Obviously and as Brian Billick's point above implies everything is different for the fourth quarter. When a team needs to score it will and should do everything it can, there is very little a team is not willing to risk in order to gain a lead, look at the onsides kick, I don't know for sure but something tells me the conversion rate for recovering an onsides kick is probably somewhere between 10% and 20%. That is a low percentage to gamble to get the ball back in exchange for giving it back to the opponent who by definition already has the lead and now has dynomite field position.

The best example I can offer of this phenomenon of chasing points too early is the Redskins vs. Cowboys game at Redskins Stadium on 5 November 2006. After surrendering a safety and a field goal to the Redskins (there would have been no safety to surrender if Joe Gibbs had opted for the field goal instead of going for it on fourth down from inside the one, not making it and turning it over) Bill Parcells' Cowboys scored to make the game 6-5 Cowboys. Bill opted to go for two to make it 8-5,putting the Redskins down an even field goal but failed, leaving the score at 6-5 Cowboys. This was the first quarter.

The rest of the game went like this, scoring wise: 9-5 Cowboys, 12-9 Redskins, 12-12 tie, 19-12 Cowboys, 19-19 tie, 22-19 Redskins, this final Washington field goal coming after the blocked Mike Vanderjagt field goal, recovery by Sean Taylor, facemask by Kyle Kosier and untimed kick by Nick Novak.

That extra point the Cowboys gave up in favor of the try at two was by no means what cost Dallas the game, rather it changed the scoring dynamic and the two ties and two subsequent lead changes may never have happened if the margin to tie or go ahead was greater than a field goal.

Bill Parcells did not lose that game because he went for two in the first quarter; he lost that game because of errors and bad decisions of which that failed two pointer was one. That almost happened to Jim Zorn on Sunday.

=====

Will Allensworth at Hogs Haven ventures into the long form to discuss the two point conversion and why the logical choice is to go for two after touchdown. I agree generally with his mathematical and logic arguments, it is the central risk aversion factor that determines most touchdowns are destined for extra points and not two point conversions. I generally disagree that coaches start empty handed after a touchdown. The extra point is so heavily commoditized that it is virtually house money, picking the lower percentage play is choosing possibly to give away free money. Witness what happens when a kicker misses an extra point, and how the shortest field goals are characterized as virtually extra points.

Each touchdown places a two point range on the scoring outcome of that touchdown, zero points, one point or two points. If the rate of extra points is 98% per try and the rate of two point conversions is 55% per try (liberally, could be as low as 52%) then the rate of success for two extra point tries is 96% (98% * 2). For three extra point tries the number is 94%, four successful extra point tries is 92% and so on. For the relatively few touchdowns scored in any given NFL game the overall rate for converting all those points will remain above 90%.

For two point conversions on the other hand if the rate is 55% per try then the rate of two succeeding (ie four points) is 30%. For three it is 15%. For four it is nine percent.

So more than nine times out of ten four touchdowns will yield four extra points. And less than one time out of ten four touchdowns will yield eight conversion points.

The basic risk aversion theorem is right there in the previous three grafs. Like the flip of a coin each individual two point try may have a near 50-50 outcome. However to continue to land heads each time in a coin toss, your chances go down by half until they quickly become nil.

NFL teams assume they always will get the chance to score again and then plan for not. Never knowing what the outcome of the other team's next drive or your own next drive pushes coaches to go with the fatter outcome curve over time, that is to say the extra point. Since the sample set of opportunities (ie touchdowns) is always small for every game then there is a reasonable chance that a team will fall down on a two point conversion when they need one and not get a chance to play the odds again the next time to recover back to what would have been two extra points, for the game will have ended. The time and skill boundaries are subjective factors that influence decision making.

Factor in parity, the notion that most teams should play well against one another and scoring differentials should be small on a regular basis. If coaches feel as they logically should about their defense, that this time we can hold them to a punt or a field goal then that pushes coaches to the more certain outcome based on an expectation that the other team will not bee too far ahead or too far behind.

This is not a comprehensive rebuttal to Will, he clearly put a lot of research into this and knows what he is talking about. Go read it yourself. You too coach Zorn.



Reggie Bush forlorn at the end of the Redskins vs. Saints game on Sunday: Reuters Pictures from here.

0 comments: