OT-ish: Galaxy Brain Game Theory by Tennessee Titan's HC Mike Vrabel
I know OT season ended but this is football related and I think it was some beautiful game theory. Vrabel sent a 12th man out on 2nd and 1 to intentially give up the first down Houston was likely to get anyway and keep them from running ~40 seconds off the clock.
The Titans won the game on TD scored with 0:04 left on the clock.
https://broadwaysportsmedia.com/how-mike-vrabels-intentional-penalty-saved-the-day-for-the-titans/
October 19th, 2020 at 8:52 AM ^
Pretty interesting game theory, always a fan of this kinda stuff
October 19th, 2020 at 8:54 AM ^
Correction: The Titans tied the game and won in OT. But the point still stands that the only reason they were in that position was the time/time-out saved here.
October 19th, 2020 at 8:54 AM ^
Wow, that's some brave coaching with smarts behind it!
What's next for Vrable? He gonna start going for it on all 4th downs? Which I think is an awesome idea, especially at 5 yards or less to gain...
October 19th, 2020 at 8:57 AM ^
Very interesting. I hate to give a Buckeye credit, but Vrabel does appear to be a pretty sharp coach.
October 19th, 2020 at 9:10 AM ^
I'd rather say he learned that from Belichick.
I also don't mind Romeo Crennel going for two to try and make it a two possession game either. If they make the conversion, the game is essentially over and we're probably talking about him instead as the genius.
October 19th, 2020 at 9:27 AM ^
He never coached under Belichick. He coached under Jim Tressell, Urban Meyer and Bill O'Brien. He played under BB but never coached under him.
October 19th, 2020 at 9:41 AM ^
Can you not learn a thing or two from your coach that would carry over to your time coaching?
I'm not saying he did get this from BB, but he certainly could, and probably did, that at least some of what BB does as a coach and uses it himself.
October 19th, 2020 at 9:56 AM ^
There's something to be said to the fact the most successful (thus far) HC off the Belicheck tree never actually coached under but, but instead played under him.
Probably easier to impose "the patriot way" to players when you've been in their shoes.
(It also helps to have Derrick Henry.)
October 19th, 2020 at 9:46 AM ^
I thought going for two was an incredibly bad decision. Kick the point and then if they score they have to go for two just to get to OT.
October 19th, 2020 at 10:46 AM ^
Bad decision in hindsight, but not an INCREDIBLY bad decision to try & seal the win to go up 9 with a couple minutes left.
October 19th, 2020 at 11:04 AM ^
I don't agree. The success rate of two point conversions is about 50%. So, the Texans flipped a coin to win the game or go up seven. If the Texans kicked the extra point (98% conversion rate), the Titans would need a successful coin flip to tie the game. By kicking the extra point, the Texans would take on the 2% risk of missing the extra point rather than pushing that on the Titans. Texans 2 point conversion = 50% coupled with a 98% Titans kick if the 2pt try was unsuccessful; Texans 1 point conversion = 98% kick with 50% Titan try.
Add in the fact that the Texans offense was doing quite well and had the Titans defense on its heels, and this becomes a no-brainer.
October 19th, 2020 at 11:52 AM ^
Wow, I also thought that the point that the 2-point conversion was a bad idea, but re-evaluated after reading your comment and you're exactly right. From my back-of-the-envelope math, it looks like going for 2 is a good idea if you convert with 45.6% odds or higher, which the NFL average is above.
Assuming both teams score a touchdown, which is the only time the decision is relevant:
- If the Texans go for 2, they win if: they make it -or- they miss it and the Titans miss their 1-point conversion.
- If the Texans go for 1, they win if: they make it and the Titans don't make their 2-point conversion -or- both teams miss their 1-point conversion.
The 1-point conversion is actually around 94% (98% was before it was moved back), so using that and plugging in various 2-point conversion values, these two scenarios give equal odds at about a 45.6% rate for the 2-point conversion. If the rate is above that, which it has been for the last few years, the Texans are more likely to win outright if they go for 2 than if they go for 1.
That's without looking up how both teams' kickers are doing and their own separate 2-point conversion rates, so it's just an estimate, but I would not have expected the 2-point conversion to be a good idea.
October 19th, 2020 at 12:22 PM ^
I do think you left out one important variable though: if the Texans go for two and miss, they will lose if the Titans go for two and convert. Obviously that ended up not applying in this case (the Titans went for the tie), but it should probably be factored into the original decision making.
If my math is right, a 45.6% conversion rate equates to a roughly 24.8% chance of losing to a 2-point conversion if the Texans attempt a 2-point conversion, compared to a 2.74% chance of losing to a 2-point conversion if the Texans attempt an extra point.
October 19th, 2020 at 1:45 PM ^
I think that you can safely assume the chance of a 2pt attempt down 1 to be close to zero. We know how risk averse NFL coaches are in general, so I think all play for the tie and go to overtime. In this game in particular, the Titans are pretty clearly a better team than the Texans, so they are even more likely to play for overtime instead of risking the loss on a coinflip-ish 2pt try. So you can discount to basically zero the chance of a loss on missed 2pt (Texans) and made 2pt (Titans) in your end of game scenario matrix.
Very interesting that there were two big game-theory items at the end of an otherwise non-descript AFC South (the NFL's lamest division) game.
October 19th, 2020 at 3:45 PM ^
That the Titans (clearly the better team) would almost definitely play for the tie is a good point. I'm curious though, if teams were flipped, do you think it changes the decision making at all?
Personally, I think that if it was the Titans deciding whether to kick the extra point or go for two the best play would be to kick the extra point. I absolutely think the Texans would be right to go for two and the win if presented with the opportunity because of a missed 2-point conversion, as they would be less likely to pull out a win in overtime.
October 19th, 2020 at 1:51 PM ^
75% chance of winning the game > 25% chance of losing the game. 50/50 Texans win with their 2pt conversion. If they fail, it is another 50/50 they win if the Titans go for 2. Two coin flips have to go wrong for the Texans to lose; if either go right, they win. You play to win the game.
October 19th, 2020 at 4:19 PM ^
While I completely agree that, in this instance, the Texans made the correct decision considering they were playing the Titans, I don't think the decision making is that simple in every case.
Specifically, as Sambojangles pointed out, the Titans are pretty clearly the better team. As such, they had much less to fear from overtime than the Texans did (which was proven out). For that reason, the Texans playing to win was the right call, even if it didn't work out.
But flip the teams and I think the decision making changes. The Titans would be looking at the following outcomes (using 45.6% for the 2-point conversion, 94% for extra point, and assuming the Texans would go for the win given the opportunity). A 2-point conversion means there's a 75.2% chance of winning and a 24.8% chance of losing. An extra point means there's a 54.4% chance of winning, a 42.9% chance of tying (OT), and only a 2.7% chance of losing.
What do you think the odds are that the Titans would win should a game against the Texans go into OT? Because anything at or above 48.5% means the odds of the Titans winning would be as good or better if they were to kick the extra point. And, even though it shouldn't matter, the optics of the better team losing in regulation because of an aggressive play call are significantly worse than the optics of losing in OT because of a conservative play call.
Obviously this is all hypothetical, and the numbers will change as you play around with the 2-point conversion percentage, but I think it's an interesting conversation to have.
October 19th, 2020 at 2:23 PM ^
You're right that I left that out. I did that because I thought it would depend on the odds of making the decision: if the Titans would definitely go for 2, that'd be one calculation, if they would definitely go for 1, that'd be a different one, and any percent likelihood in between would be a separate calculation. But I think I got bailed out by some luck: the likelihood of winning in each scenario is basically equal, so it wouldn't matter which one the Titans chose.
The only scenarios that impact this calculation are that the Texans miss their attempt (whether it's for 1 point or 2), then the Titans score a touchdown and need to decide whether to go for 1 or 2.
At that point, if the Titans kick the extra point, there's an 8% chance of missing, and 92% of the time you go to overtime, which is roughly 50-50: so the Texans win 54% of the time (8% + half of 92%).
If the Titans go for 2 to win, they convert and win 46% of the time (using the break-even point from my previous calculation), but if they don't convert the the Texans win. That results in the Texans winning 54% of the time in this scenario, too.
Since the 2-point conversion percentage is actually higher than 46%, it might actually have benefitted the Titans to go for 2 in that scenario, but in the overall picture a higher conversion percentage would benefit the Texans original decision to go for 2 even more, so going for 2 would still be the correct call.
October 19th, 2020 at 12:52 PM ^
Normally, I'd agree. But those defenses were bad. The Titans were going to score again and I liked their odds with Derrick Henry on the 2-yard line. At that point, I thought the gamble to make it a two-score game was worth it.
October 19th, 2020 at 12:00 PM ^
Eh - he's a Buckeye AND a Patriot - a combination of the two worst things you can be!!
October 19th, 2020 at 9:16 AM ^
#STRATEGERY
#BRINGBACKOPPONENTWATCH
October 19th, 2020 at 9:38 AM ^
I guess you will need to do your own opponent watch diary or thread.
October 19th, 2020 at 10:58 AM ^
Never.
Too much like work.
October 19th, 2020 at 9:30 AM ^
Great stuff.
It surprises me that Harbaugh is not all over critical end of game tactics and clock management as a coach who seems to take advantage of every potential grey area or loophole.
Pay one person to be full time end of game scenarios coach if you need to.
October 19th, 2020 at 9:40 AM ^
Learned from Belichick one of the modern masters of game theory.
October 19th, 2020 at 9:53 AM ^
That isn’t game theory. That’s just being smart.
October 19th, 2020 at 10:04 AM ^
What I don’t understand is why Houston went for 2 when they scored the touchdown - they were up 30-29, and 7 would’ve put them up 8. I guess 9 would’ve been insurmountable, but margin of 8 is significantly more difficult than 7, no?
October 19th, 2020 at 10:42 AM ^
Agreed, make them score again and make them go for 2 after the TD to tie.
October 19th, 2020 at 11:25 AM ^
In either scenario, TEN would need a TD so take that out of the equation.
So the question is, which do you trust more: Watson converting a 2pt conversion to win OR your defense stopping a 2pt conversion with Henry and Tannehill.
I don't blame a 1-4 HOU with a lame duck interim coach for putting the ball in your big contract, franchise QB's hands for the kill shot.
October 19th, 2020 at 11:55 AM ^
I thought the same, but when Westside Wolverine disagreed with a similar comment above, I did the math--the details are above, but if I did them right, generically the odds of winning are actually better if you go for 2.
October 19th, 2020 at 10:51 AM ^
Have to admit that I am impressed with Vrabel's in-game coaching chops.
October 19th, 2020 at 11:11 AM ^
He got the idea from Buddy Ryan’s “Polish Goalline defense”
http://smartfootball.com/defense/buddy-ryans-polish-goalline-tactic#sthash.G7phiI8W.dpbs
October 19th, 2020 at 12:22 PM ^
I wish Harbaugh had read that prior to his first MSU game.
October 19th, 2020 at 12:28 PM ^
Was thinking about this yesterday. I'm surprised that penalties aren't seen as more of a resource in football. Every so often you see stuff like this: intentional kicks out of bounds, delay of game when trying to draw opponents offsides or to give a little more space for a punt, etc. IIRC, Belichick took two "false starts" on a punt to run the clock...or was it one delay of game and one false start??
Basketball definitely has foul strategies, especially at the end of games.
October 19th, 2020 at 5:54 PM ^
Yeah, Belichick and the Jets last year was hilarious. New England takes a delay of game on a punt, but the Jets decline the penalty. So the clock continues to run. Belichick then says, OK, how about a deliberate false start. The Jets AGAIN decline the penalty, so the clock still runs.
The hilarious thing was the Jets trying to troll Belichick --- in a game they were losing 33-0 at the time. OK Jets.
October 19th, 2020 at 5:02 PM ^
I've always been an advocate for this rule change:
(1) if there is a penalty committed within the last 10 minutes of the game, and
(2) that penalty is accepted, and
(3) the clock was running prior to the play, then
(4) the unpenalized team gets to decide whether the clock should run or not on placement of the ball.
---
I specifically recall a Purdue/PSU game back in 2004. Purdue is winning by 1 score, 6:00 left, has the ball, clock is running. Purdue has a false start, 5 yard penalty. Then the same thing happened a couple plays later on their next set of downs. They weren't cheating by any means, and it was PSU defense's fault for letting them get a new set of downs. But that's also 1:00 of game clock they got to run off to their advantage.
October 19th, 2020 at 6:48 PM ^
...and to think. The Lions could have easily hired Vrabel instead of the pencil pillsbury doughboy.
Quinn got his way and the Fords are still Edsels.