Over-reyling on analytics? (not going for 2)

[Anonymous]

My thinking when we were driving for the tying TD was to go for 2 thinking should it go to OT, OSU would have been massively favored (I think 80%+) considering our QB situation and complete inability to stop their offense. Going for 2 would’ve been at worst 50-50 for us to at least take the lead and force OSU to score – I think much better odds for us potentially winning than if going to OT or even getting the ball back (explained below). Ultimately the way things played out with OSU easily storming down the field on the final drive, perhaps they would have won anyway even if we converted 2 (although who knows how things might have played out differently), but I think it’s still worth questioning the rationale of not going for 2.

In the post-game press conference, Whit explained that he would have gone for 2 if there was under one minute left, but with two minutes the analytics indicated to go for the tie. I question simply going with the analytics in this case when you factor the above mentioned QB and impotent defense situation. Even if we managed to force a 3-and-out and got the ball back, we would have likely still needed to drive about 50 yards with a minimal-experience QB just to get to FG range with a bad kicker – I think still lower odds of winning in this scenario than going for 2. Was Whit too simplistic in his assessment and over-relying on conventional analytics when there were other unique, significant factors to consider?

[belli1976]

Anyone that is super successful in any career minimizes emotional decisions while allowing their extensive experience to allow for flexibility. This is why Whitt has a $4M salary. And no one in this group is likely to have that salary. (There is still the possibility that OnlyU is Whitt 🤪).

Maybe it is just me, but why question Whitt. He is one of the top 5 coaches in the country.

The only thing I wonder is if he will be able to coach 5* athletes. Those athletes are like dealing with NFL players. IMO. Managing the egos is much different then the buying into the system and family approach that Whitt has created. 5* want the sexy.

[Anonymous]

“Anyone that is super successful in any career minimizes emotional decisions while allowing their extensive experience to allow for flexibility. This is why Whitt has a $4M salary. And no one in this group is likely to have that salary. (There is still the possibility that OnlyU is Whitt 🤪).”

Sorry, but this is very flawed reasoning as a rebuttal to my specific point. Sustained success indicates high net-positive decision-making; an extremely competent/successful person can/will still be wrong and make mistakes at times. I’m not at all questioning Whitt’s overall competence, I’m simply thinking analytically about one specific decision.

[kazute]

I agreed with Whitt then and I still do.

Here is the thinking. You go for 2 to try to get the “instant” win and avoid overtime, because you don’t like your chances in overtime. But you also take your chances for an “instant” loss if you don’t get the two.

In this case there was no “instant” win or loss. There were nearly two minutes left for an offense that had scored like 3 touchdowns on 3 pass plays during the game. You can only hope for a stop and get a chance to score again during regulation or go to overtime and go for 2 then.

They can easily get into field goal range with 2 min left and a one point lead still doesn’t help you.

This is what went through Whitt’s mind and my mind. He said that the analytics confirmed this thinking (too much time) not that it changed his mind. So without analytics you get the same decision.

He also said they would have slowed the clock with Rising, but with Barnes, they didn’t want to add that extra pressure when they absolutely needed the score.

I have second guessed a couple of Whitt’s decisions (like the time out during the Washington game when Washington was happy to run out the clock for overtime), but this isn’t one of them for me.

[mfaulk80]

Yeah, going for 2 wouldn’t have done much.  An onside kick could’ve been interesting, but even that’s really risky.  The only reason it could make sense is because our kicking game was bad (they returned it to the 42 anyway) and it was hard to imagine that we could trust the defense to get a stop.  I also think Whitt could’ve used some timeouts, but I understand why he didn’t.  When it comes down to it, it’s hard to win late with a backup QB and a defense that can’t get a stop.  

[Anonymous]

Yes, we were unfavored no matter what considering our QB situation and floundering defense, but you still have to go with your best possible chance to win, which I think would have been to go for 2. See my breakdown of scenarios in my other reply.

[Crazyforu]

Yep. At the game I felt the exact same way, still do. Too early to go for 2, also the D couldn’t stop them so using the TO’s didn’t really help. Its too bad Utah couldn’t slow play the clock on their last posession, had to take what they could with Barnes.

There were big plays on both sides that could have changed the game, I think the result reflects what I saw in the game (although a W would have obviously been preferred).

[Anonymous]

So here were the most likely possible scenarios to weigh:

1. Go for 1:
> 1a. OSU scores with no/nominal time left; or OSU scores TD quickly and Utah gets one more possession with minimal time left, having to go the length of the field with a minimal-experience QB.
> 1b. Utah defense holds and forces OT, in which OSU would then be heavily favored.
> 1c. Utah forces 3-and-out (likely using all timeouts in the process) and gets the ball back with likely <1min and likely needing to drive about 50 yards with a minimal-experience QB just to get into FG range with a bad kicker. Can settle for OT, but OSU would still then be heavily favored.

2. Go for 2:
> 2a. Fails and OSU runs out the clock or at best Utah gets the ball back with nominal time.
> 2b. Converts and OSU is forced to score with some time pressure and only 1 timeout. OSU scores and wins or Utah defense holds on a few plays and wins.

Assess the the win-probability of each of these possible scenarios – I think we are highly unfavored in every scenario going for 1 and going for 2 gives us the best possible chance to win. The longer you extend the game, the more it favors the better team, which was clearly OSU in this case. I think going for 2 (I think at least 50-50 chance of converting), taking the lead, forcing OSU to score with some time pressure (and only 1 timeout), and hoping the defense is able to make a few plays to hold was our best possible chance of winning.

[mfaulk80]

“Assess the win-probability of each of these possible scenarios”….

Isn’t that exactly what the analytics are doing?  And your initial post implies that the analytics told Whitt that going for 1 was more favorable.  

Honestly, the defense has to get a stop in either scenario or else it doesn’t matter.  The only option that removes the defense from the equation is an onside kick.

 

 

[Anonymous]

“Isn’t that exactly what the analytics are doing? And your initial post implies that the analytics told Whitt that going for 1 was more favorable.”

The computational analytics don’t/can’t account for every possible significant factor, and there were several significant extenuating factors in this case (e.g. game trend, QB situation, floundering defense, bad kicker). The computational analytics are useful as baseline guide, but they have their limitations and there are going to be situations where substantial subjective human assessment and judgement is appropriate.

[krindor]

This is an interesting case where the analytics say to do something specifically because most coaches don’t follow the analytics.

Let’s lay aside all the external factors (QB injury, Ohio State talent level) and take a simplified look at the analytics.

If you go for 1, there’s a (for simplicity) near 100% chance you make it to tie the game.
Then, with that amount of time left and the score tied, the opposing offense is less likely to go into complete hurry-up (only passes, going for it on 4th down, etc) and more likely to let the clock run out. So they score ~25% of the time.
That means a 75% chance at overtime, which (in the abstract) is a 50/50 shot. So 37.5% chance of winning.

If you go for 2, that’s loosely a 50/50 shot. If you don’t make it, then game is effectively over. So let’s talk what happens if you do make it.
Now the opponent is against a wall and so are more likely to play aggressively, pulling out all the stops and more likely to get into FG range. So there’s now a 50% chance they get a FG. So now your odds of winning are 0.50 (chance of getting the 2) x 0.50 (chance of stopping the FG) for only a 25% chance.

The key difference is in how the other team responds being down 1 vs being tied, which is why the analytics there suggest going for 1. Not saying I always agree, and there were certainly other extenuating factors, but that’s the big reason for it in this case

(This is all obviously a much simplified version, but explains the story well)

[Anonymous]

I get the value of base analytics, but I think there are clearly cases when there are unique extenuating factors that are significant and the analytics can’t account for. As such, the base analytics have their limitations and there are cases when subjective human assessment and override would be appropriate. I think more subjective assessment was warranted in this case.

For example, I don’t think the abstract 50/50 probability of OT outcome is at all realistic in this case considering the game trend and extenuating factors with Utah’s QB situation, floundering defense, bad kicker, etc. OSU would have been heavily favored had it gone to OT, I think moreso than Utah going for 2 (realistic 50%+ chance of converting) and condensing the game to a situation where Utah has the lead and OSU is forced to score under time pressure. As I said before, extending the game favors the superior team, which was clearly OSU in this case. An inferior team’s chances are better when the game is shortened, some higher-than-normal risks are taken to try to “steal” an advantaged position, and some luck.

[2008 National Champ]

I’m surprised because I question nearly all of Whitt’s decisions and had no problem with kicking. Why add degree of difficulty when you’ve got around a 95% chance of tying the game?

Going for two has less than a 50% of success, otherwise coaches would always go for two. Additionally, if you do not make it, now you are relying on an onside-kick which has less than 20% chance of success. If you do not recover the onside, now you are stuck with the same defense you didn’t trust in the first place with the ball on your side of midfield. The problem was that Utah was not able to kick the ball where Ohio State could not return it, and then gave up a return to the 42.

Now if you want to question a decision from that game, the 4th down pass to Kuithe was the one that I had a problem with. 4th and 3 at the Ohio State 31, up 7. Granted, no one besides Redding’s mom has confidence that he can hit a 48 yarder but a field goal at that point puts you back up two scores. Don’t think your kicker can make it? I’m okay with running a play but it can’t be a pass behind the line of scrimmage. Zone read, speed option, Fotheringham/Kincaid/Covey over the middle.,, anything but a flat pass to your gimpy TE. Rising owned the game to that point, let him make a play.

[Anonymous]

“Why add degree of difficulty when you’ve got around a 95% chance of tying the game?”

Because extending the game greatly favors the superior team, which was clearly OSU at that point. With the way the play was trending, Utah was essentially toast after losing Rising and OSU scoring the go-ahead TD. Mad respect for Barnes, but realistically that TD was an extreme improbability. Utah was very fortunate to get that TD and it gave them a glimmer of hope of still winning by trying for 2 and potentially going up with the game condensed to two minutes. An outmatched team’s best chance is to shorten the game, take some higher-than-normal risk to try to “steal” an advantaged position, and some luck. At that point there was no way you could reasonably expect Utah to be able to match or outplay OSU in an extended game. I think taking a chance to convert 2, stealing the lead, putting OSU in a position where they had to score with some time pressure, and hoping your defense could provide enough resistance for two minutes was their best chance.

[2008 National Champ]

I’m fine with you disagreeing, this place would be boring if we all agreed on everything and gave thumbs up to the right reverend OnlyU 🙂

I think we can all agree that a failed two-point conversion when down one forces you to rely on converting the onside kick to get the ball back. But let’s assume Whitt went for it and converted. Utah is up one, kicking off. The job is still to keep Ohio State from getting in to position to make the same game winning field goal. I can’t say that Utah would have played better D with a one point lead but I think it is fair to assume, based on what did happen, that it would probably have ended up the same way. So Utah probably loses 48-46 anyway.

Now let’s assume that Utah doesn’t convert. Even if they recover the onside, they still have to score again. Would I rather Utah have the ball down 1 with two minutes left instead of the opponent in a tie? Normally, yes. With a walk-on who has thrown 2 career passes? I’m still going to put the onus on my defense to make a play. We know what did happen, it is all speculation on what might have happened.

Ultimately, not going for two is not what cost Utah the game. There were a number of plays which had as much or more of an effect on the outcome. That decision just happened to be the last of the game so it is the easiest to disagree with. I just don’t think I would have done it any different than Whitt did in that moment because the slim % of success does not outweigh the potential disaster. Either I trust my D in overtime or in regulation. They were going to have to stop Ohio State to get the win no matter what happened on a potential two point conversion.

[Author]

Posts