Weekly Game Probabilities - Wildcard

Weekly game probabilities are available now at the nytimes.com Fifth Down. This week I focus on the one game where the model differs from the consensus favorite.

10 Responses to “Weekly Game Probabilities - Wildcard”

1. James says:

Couple of quick questions that you may have answered before but I don't remember.

1 - Do the predictions consider the specific game matchups, i.e. an above avg rushing offense vs a below avg rushing defense is an even bigger advantage?

2 - Do you factor in a team's run/pass ratio? I know it's biased some by winning teams running out the clock, but it seems that the Colts passing offense and the Chiefs rushing offense should be weighted more than their opposites.

2. Sampo says:

@James

I have hard time believing that either of those factors would bring more predictivity to the model. I'm no Brian though...

3. Anonymous says:

Brian has said EPA doesn't add much predictive power either, but that didn't stop him from relying on it in his article.

4. Brian Burke says:

EPA is not in the model, but it is fairly predictive. It was particularly useful because of Vick's unique abilities.

1-I've tested models that emphasize the particular match-up of offensive passing vs defensive passing efficiencies for each team, etc. But they were no more predictive than the much simpler method. It's still a good suggestion and probably worth a re-look with more data.

2. No, I don't include run/pass ratio. Good point about how some teams rely more on their run than others. But the backward-causation problem induces too much error in the results.

Recently there was discussion of the value of starting overtime with an onside kick. It seems to me that the new playoff OT rules would make this strategy even more valuable. If you succeed with the kick and score a FG, the other team has no chance to match. If you fail and they score a FG, you get the ball back. Any thoughts?

6. Brian Burke says:

I'm actually working on an article on that topic now. The new format is really hard to model. So many unknowns.

7. Anonymous says:

I don't have all of the numbers to crunch, so I won't, but I imagine that all of the talk and publicity it's getting is going to result in the surprise factor going way down, and so something like a 65% success rate in surprise situations isn't applicable, and you're looking at something closer to the 20% success rate or whatever it is. Not sure where the break even point would be, I'd guess it's even slightly below 50% SR though, which may not be attainable.

8. James says:

Thanks for the quick response, Brian.

Re: OT, my biggest question is how will (or should) the OT rules effect when it's worthwhile to kick?

Previously, some teams would attempt long OT field goals because it offered them a chance to win immediately. Now that the other team has a chance to respond, it devalues the gain of making such a kick. Similarly, near the endzone it becomes more worthwhile to go for the touchdown and end the game vs attempting the field goal and letting the other team respond (and possibly scoring a TD and winning outright).

Also, if you're the responding team and need a score this drive, at what point do you go for the winning TD vs the game-tying field goal? If it's 4th and goal from the 1 do you kick or go for it?

I suppose the better teams should take the low variance route and extend the games believing their superiority should carry them, while the underdog should go for the TD, but there must be a break-even point somewhere in between.

9. Anonymous says:

If you kick a FG on the initial possession of OT, should you then go for the win with a surprise onside kickoff? That would be the best finish ever!

10. Brian Burke says:

That would rock. Look for an article on that topic in Slate tomorrow.