Friday, October 12, 2012

You Got to Know When to Hold ’Em, Know When to Punt

That chart to the left is from a website called Advanced NFL Stats, one of many number-crunching sites that produce all sorts of breakdowns and analyses of NFL players, teams, and games.

For every game, Advanced NFL Stats creates a graph such as this one that reflects the chance a team will win the game based on historical comparisons of the situation at any given moment in a game. They call this Win Probability Added (WPA) because what they are measuring is how each additional play increases or decreases a team’s probability of winning the game. (Click here for a full explanation.)

It’s kind of a generic tool, really, that doesn’t take into account the actual teams or players or specific match-ups, but rather just looks at every situation -- the score, the time left in the game, the present down, yardage needed for a first down, and field position -- and compares that to how games have historically gone for teams in similar situations. I’m not sure, but I assume they’re using a database of thousands of previous NFL games to produce these percentages.

I sometimes like to take a look at these graphs after a particularly exciting game, just to see how the line came to resemble a seismograph measuring an big earthquake or some wacky EKG readout.

For example, the particular graph shows how the probability of the Pittsburgh Steelers winning last night’s game was 84% when there was 1:52 left and they had a first down on the Tennessee 39-yard-line with the score tied 23-23.

The Steelers subsequently saw their drive stall at the 36-yard-line (and their WP dip to about 50%), at which point they made a risky decision to have their field goal kicker, Shaun Suisham, try a long field goal -- a 54-yarder, in fact, longer than any Suisham had hit in his entire career.

Suisham missed the try, and immediately the Steelers’ probability of winning dipped to 33% as Tennessee took over possession with 54 seconds left. Meanwhile the Titans’ probability had improved to 67%, and sure enough they drove down the field and kicked a winning field goal as time ran out to win 26-23.

If Pittsburgh had punted rather than try the field goal, their win probability would have also decreased although not as severely. And in fact, if they had been able to pin the Titans back inside their own 20 with the punt, they would’ve likely gotten the game to overtime, at the start of which both teams’ WP would have been exactly 50%.

All of which is to say, Pittsburgh found itself in a tricky spot near game’s end, ultimately deciding to take a chance on Suisham hitting a career-long field goal rather than accept what was essentially a “coin-flip” situation. Literally, in part, as overtime would have begun with a flip of a coin.

Suisham had just hit a long field goal (52 yards, tying his career best) earlier in the fourth quarter, an event that clearly affected the decision to let him try another long one. His chances of hitting that second one are hard to estimate. Career-wise, he was 5 for 12 when it came to attempts of 50 yards or more, although as noted he’d never hit one as long as 54 yards. Probably safe to say it was less than a 50% chance he’d make that attempt last night, although even if Suisham had made it that still wouldn’t have ensured the Steelers a victory.

The whole situation uncannily resembled that of a poker player at the final table of a tournament who after enjoying some recent success -- say, he catches several good hands to chip up and into the lead -- decides to “ride the rush” and take an unnecessary risk that if successful will improve his chances of winning the tournament considerably, but if it fails will unequivocally reduce the likelihood of victory.

Like I say, I find the WPA graphs over at Advanced NFL Stats diverting, probably because of they way they can make analogies between poker and football seem more apparent. So Pittsburgh was 84% to win with a couple of minutes to go last night (probability-wise). The probability of pocket aces beating 8-5-offsuit is about the same.

But sometimes the aces get cracked.

Of course, in the Steelers’ case it wasn’t just a matter of getting unlucky. They failed to make a first down, and their pocket aces were reduced to Q-7-offsuit (or worse). Then they decided to push anyway, and when their hand didn’t hold, they soon found themselves all in and dominated after the Titans drove to set up a game-winning field goal of their own.

Or, to put it another way, Pittsburgh decided to play the role of a punter rather than punt.

Labels: , ,


Blogger Robert Dudek said...

I'm not so sure it was a bad gamble.

Let's say if Suisham makes it, then the Steelers win 85% of the time. So we would need to plug it into an equation to find the break even point.

First: FG try:


Now suppose that the punt would have resulted in an overall 50% chance to win - the break even point thus becomes:

.85x+.33(1-x) = .5

Solving for x leads to x being approximately .327 - s Suisham needed at least a 32.7% chance of making the field to make the attempt at least EV neutral.

If you change the assumptions a little in favor of the attempt - e.g. a made field raises chances to 95% and a put could go relatively poorly and give the Steelers only a 45% chance of winning - the breakeven point drops to 19.35% - quite a difference.

It's safe to say that this particular gamble was defensible and might have even been the correct play.

10/12/2012 9:04 PM  

Post a Comment

<< Home

Newer Posts
Older Posts

Copyright © 2006-2017 Hard-Boiled Poker.
All Rights Reserved.