Up 7 - go for one or two?
Haven't seen this decision discussed. Heading to the end of regulation Michigan was up 1 and scored a TD to go up 7. Why not go for 2? Would win the game most likely as now you are up 9 and Rutgers would have to score twice. If you don't get the 2 you are still up 7 and most likely if Rutgers scored a TD they would go for 1 to tie and go for OT.
November 22nd, 2020 at 9:48 PM ^
In this scenario, a team cannot lose by going for one point.
November 22nd, 2020 at 10:29 PM ^
I’m not saying going for 2 is the right answer, but your logic is wrong. I’m t doesn’t matter if you lose in regulation or OT. A loss is a loss
November 22nd, 2020 at 11:35 PM ^
What makes you think Rutgers would go for one? Look at Indiana and PSU early this year.
November 23rd, 2020 at 11:40 AM ^
Not only is this is a loser's mentality, its not even right lmfao.
November 23rd, 2020 at 2:48 PM ^
I assume you meant that "going for 1" is not the same as "getting 1." Sure, it's more likely that getting 2, but it's still not a guarantee. That would have to be part of any analysis of the best strategy.
November 22nd, 2020 at 9:48 PM ^
Go for 1 - make it 98%. Cannot lose in regulation. Greater than 50% in overtime if you have more talent. It essentially erases the chance the other team can beat you without overtime (like last night).
November 22nd, 2020 at 10:59 PM ^
I mostly agree, but when you factor in Nordin's XP problems in the past, the long snapping issues, and Nordin's 0-2 FGs at that point, the XP doesn't seem anywhere close to automatic either.
November 22nd, 2020 at 11:37 PM ^
I debated this one as well, and I think I lean going for 1- depending on opponent. I think schiano likely goes for the tie but then again who knows. If we were playing ohio state though, and the circumstances were the same, I'd say go for 2 and make it mathematically highly unlikely they can come back from a 2 possession game with just a couple minutes left.
November 23rd, 2020 at 9:28 AM ^
Ah to imagine a world in which we're in a position to take a two possession lead on osu late in the 4th quarter... that's a heady brew my friend
November 23rd, 2020 at 9:58 AM ^
Yes, it is heady brew. I logged in to support you. I will also add that I fantasize about beating OSU 101-100 just to counter their 100 points comment.
November 22nd, 2020 at 9:51 PM ^
In a situation where Rutgers scores at the end of regulation like they did, if they are the considerable underdog, they would have gone for 2 to win the game. And they would have gotten it. And Michigan would have lost.
Michigan did the same thing against OSU in 2013.
November 23rd, 2020 at 7:26 AM ^
Michigan went for 2 because Gardner had a broken foot and couldn’t play much longer.
November 23rd, 2020 at 10:40 AM ^
And Rutgers would go for 2 because they had 1 play to win the game against a team they shouldnt have had a chance against. Underdogs do it all the time.
November 23rd, 2020 at 1:08 PM ^
And Borges called the same damn play after OSU called time out. The OSU players couldn't believe their luck.
November 22nd, 2020 at 9:52 PM ^
I like being aggressive. Any opportunity to add a possession to your lead, take it.
November 22nd, 2020 at 10:01 PM ^
I would not make a good coach, but I generally go with momentum. Our offense was rolling and the clock was on our side. I would have felt good going for 2 this time.
November 22nd, 2020 at 10:31 PM ^
Don Brown says hold my beer in regards to the clock
November 22nd, 2020 at 10:35 PM ^
I agree with you. The way offense was playing felt like Michigan would have better than 60% chance of picking up conversion and icing game, assuming they were better prepared for an end of game onside kick. My guess is if we missed rutgers would have kicked extra point for overtime
November 23rd, 2020 at 9:03 AM ^
Our offense was certainly doing well, but for some reason our playcallers have become convinced that running the QB out of the gun is the thing to do in short yardage. I wouldn't love our odds of converting that.
November 23rd, 2020 at 9:07 AM ^
I disagree on the clock. 4 minutes and change plus 3 timeouts is an eternity in CFB. Rutgers only used 85% of it because they kept running the ball. If they went 3 and out fairly quickly they could easily get another possession even giving up a first down.
Take the XP all day. In the other hand, cover a fucking slant on third or fourth and long on the ensuing possession. We gave up 3 of those on that last possession. If anything cover the easy routes and make them beat you deep, if they score quickly you have plenty of time (and timeouts) to score.
Playcalling and time management was awful once we got the ball back but that is pretty par for the course for the last 6 years. Have we ever really executed perfectly at the end of a game in the entire Harbaugh era? Even times when we’ve gotten a critical late score we’ve left a lot of time on the clock.
November 23rd, 2020 at 10:02 AM ^
Yes, Starting with FSU game. The Wolverines have had no ability to speed up and maintain composure. I long for Gary Moeller and having scripted starts to a game with no huddle and no break in the action. A good 2-minute drill would be heavenly.
November 23rd, 2020 at 9:33 AM ^
But then JH calls the QB draw from the shotgun, with an average-sized QB, and does nothing to disguise the play so RU knows exactly what is coming.
Are those variables in the model?
November 23rd, 2020 at 9:38 AM ^
Not even a QB draw (which involves an element of misdirection), just a straight QB run.
November 23rd, 2020 at 5:16 AM ^
You say that, but had we gone for two and NOT gotten it, Rutgers drives down the field and scores, goes for two and GETS it (which they did do) then we’re down 1 with ~20 seconds left and we lose. Everyone and their mother would be bitching about the decision and calling for Harbaugh’s head, more-so than we already are.
November 22nd, 2020 at 9:54 PM ^
I have no statistics to back this up, but I have thought about it always thought kicking was the right choice. While it is true that getting the conversion makes it more than one possession, Rutgers would have approached their next possession very differently in that case, and the risk you take is that is a near-certain tie if you don't get it and they score. (or you expose yourself to losing on a two-point)
I think the 8-point lead gives you better overall value, because it makes certain your opponent cannot beat you with only one more possession, and you have one-play-take-all chance to keep your lead if you should give up a touchdown. In other words, they are forced to score twice regardless, albeit once only from the three, so it gives you some of the positive value of a 9-point lead anyway.
And let's be honest, if Rutgers had scored faster because they knew they had to and left enough time to kick it deep... we're probably running three straight times and punting it back to them with decent field position anyway.
November 22nd, 2020 at 10:08 PM ^
In that situation, I like the idea of trying for the 2 points in order to end the game. It's a little risky though if the other team gets a TD and goes for 2 to win the game... but going to OT because the other team scores and gets the 2 point conversion to tie the game is a little risky too.
November 22nd, 2020 at 10:17 PM ^
It's a tough enough question under normal circumstances, but with our long-snapper/placekicking game being in tatters right now, I go for two as long as Cade is QB.
Someone else in this thread cited a figure of 98% probability of successfully kicking the PAT; for Michigan Football 2020 I'd contend that the percentages are far less favorable than that.
November 23rd, 2020 at 3:16 AM ^
I know the purpose of your comment is not statistical accuracy, but Quinn Nordin is 100% on XP this year and 97.56% on them in his career.
November 23rd, 2020 at 6:43 PM ^
While his stats look good, his placekicking this season doesn't exactly inspire a lot of trust.
November 22nd, 2020 at 10:18 PM ^
This was the exact situation that Houston faced in an NFL game against Tennessee earlier this year. They went up 7 with a late touchdown, went for 2 to make the lead 9, failed, and Tennessee tied it at the end of regulation and won in overtime. Even though the 2PA failed, I supported it at the time. Houston was 1-4 at the time, and an underdog in the game, so that 2PA was basically their chance to seal the win. They also have Deshaun Watson, one of the QBs most able to convert the attempt.
In this case, I think Michigan has enough talent that you would rather force Rutgers to score 8 and then win in overtime. When you're the favorite, it's best to extend the game and give your superior talent the chance to prevail. So without running the numbers, I think it's close enough that neither option is wrong, and I'm fine with what we did.
November 22nd, 2020 at 10:19 PM ^
Because you force the other team to make the conversion with 1 point, otherwise you give them the tie and an opportunity to win with 2.way too high risk to go for 2 up 7
November 22nd, 2020 at 10:41 PM ^
If offense is playing well and defense playing poorly I like going for two. If offense struggling (relatively for team that just scored) and if I like my defenses chances of stopping 2 pt conversion I'd go for 1. Yesterday would have been in favor of 2, in 1997 would have always gone for 1.
November 22nd, 2020 at 10:41 PM ^
I was thinking about this last night as well. I think the convention is to go up 8 because the odds are significantly lower that a team will get the 2 pt conversion. Of course we still gave it up, but I bet Rutgers would have went for 2 if they were down 1.
November 22nd, 2020 at 10:43 PM ^
I found an interesting article that did a statistical analysis on when to really go for two or one (based on NFL probabilities of converting the PAT or 2 Pt conversion). It says that it is a push from a statistical perspective when you go up by 7.
Some pretty interesting scenarios where teams should go for two such as if you score a touchdown to cut a lead from 14 to 8 points, a team should go for two.
https://fivethirtyeight.com/features/when-to-go-for-2-for-real/
November 22nd, 2020 at 11:03 PM ^
Another interesting graph that shows variance on your team's chances of converting the 2-point conversion (40% at the lower end of the band and 55% at the upper) compared to the time left in the game on the x-axis. Orange means go for 2 and purple means kick the PAT. This would indicate the PAT was probably the better option because there were still a few minutes left when they scored and it's probably safe to assume Michigan would be closer to 40% on a 2 pt conversion than 55%.
November 22nd, 2020 at 10:46 PM ^
Mathematically I think it’s about equal?
Going for it wins the game in regulation 50% of the time (assuming 50% 2 point conversion) and wins it in overtime 25% of the time (assuming OT is a toss up).
Kicking will force the other team to go for 2, in which case you still win in regulation 50% of the time and win in OT 25% of the time.
If you thought you had a higher chance of stopping them or converting for 2 yourself that could decide it.
November 22nd, 2020 at 10:47 PM ^
It might be a worthwhile thing to consider if our o-line didn't primarily consist of backups and first year starters. When our RBs picked up yards, they were breaking tackles and bouncing around like pin-balls. As we have seen with other short-yardage goal line situations this year, we cannot reliably pick up a yard, let alone two. That makes a 2-point conversion a big risk.
A 2-point conversion may have iced the game, but unfortunately for us the odds of actually making that conversion were pretty low. If our odds of making that conversion were good, we probably wouldn't have been in the situation to begin with. Going for 1 meant the best they could do was overtime (which our defense should not have allowed).
November 22nd, 2020 at 10:51 PM ^
I did a rudimentary analysis in excel and my topline conclusion is that it is a slightly better decision to go for 1. It increased your win probability by 1 to 4 percent depending on the parameters.
Overtime is a 50/50 proposition. P5 kickers make XP at a 97% rate and 2pt attempts are successful at a 42% rate. What makes the analysis hard is estimating the aggressiveness of the opposing coach and the likelihood the other team will score.
November 23rd, 2020 at 9:42 AM ^
Overtime is theoretically 50/50 but in fact, the more talented team tends to end up winning. Our winning percentage in OT games is like .700.
Regardless, I think the math favors the PAT.
Kick the PAT and the underdog, Rutgers, must 1) score a TD; 2) convert a 2-pointer and 3) win in OT. The cumulative odds of doing all that can't be great, probably 10-20 %. Of course they nearly pulled it off - but ultimately didn't.
November 23rd, 2020 at 10:24 AM ^
If you are so much more talented that you have a better than 50% of winning in OT, why didn't you win in regulation? Calling OT a 50/50 proposition is reasonable.
In my analysis I made 2 initial assumptions
1. There is only enough time left in the game for your opponent to have one possession
2. Your opponent is rational and will not choose to go for 1 while down by 2
From there I came up with 5 tunable parameters
1. 1 PT XP success rate -> used 97% as that was the 2013 SEC conversion rate per a quick google search
2. 2 PT XP Success rate -> used 42% which was the overall NCAA rate for the '00s per a quick google search
3. OT Win % -> used 50% which seems reasonable to me
4. Opponent scores a TD on their last remaining drive % -> This one is a wild guess
5. Chance your opponent goes for 2 when only down 1 % (Aggressiveness) -> This is another wild guess
With all that you can make tree diagrams and calculate the cumulative odds that you win. Going for 1PT in this scenario seems to be a 1 to 4 % better choice depending on the input parameters.
November 23rd, 2020 at 10:34 AM ^
Seriously, look up teams' overall OT records. The more talented programs generally have good OT win percentages.
Overtime eliminates three significant game factors: punting, kicking off, and the clock. It's just lining up and performing in the red zone. The better team tends to win.
Looking it up, Michigan is 13-3 (.813) all-time in OT.
November 23rd, 2020 at 11:20 AM ^
It is interesting but I don't think 16 games is large enough sample size to project a better win percentage in OT. Not to mention what impact does OT results from 20 years ago have to do with the current team? Previous coaching staffs and players who aren't on the roster don't have impact on the current team.
I guess I view OT as an independent event.
Plus it doesn't really change the results of the analysis. If OT has better than 50/50 odds for you then for sure you kick the 1pt XP.
November 22nd, 2020 at 11:03 PM ^
Math time (rough bullshit modeling ahead)!
Kick: 98% likely you get the point. Opponent needs to score a TD and make a 2 point conversion and beat you in OT. Say OT is 50/50, 2 point is 43.5% NCAA wide. So chance of losing is .98*.435*.5*(percent of opponent drives that end in TDs). I don't want to look that up because it would make me sad, but let's say it's like 60%. Would give a 12.8% chance of losing if you kick. Plus the small edge case of the misses. Which is like .3% more.
Go for two: 43.5% chance you take a two score lead and let's just consider those all wins. For the other 56.5%, the opponent has to score a TD (60%), make the PAT (98%), and win in OT (50%) OR score a TD (60%) and get the two point try (43.5%).
That's 16.6% or 14.7% depending on the tendency of the opposing coach. And those are all linear so if your defense is better than ours is it shouldn't really matter, I think it's slightly better to kick the PAT if your 2 point success rate is expected to be NCAA average.
What does matter is how good you think your two point plays are. If you think your opponent is NCAA average and the opposing coach would go for 2 down 1, it looks like you want your 2 point play to be successful at least 47% of the time for it to be a good idea. You need 50% success rate for it to be a good idea if you think the opposing coach is conservative.
November 22nd, 2020 at 11:17 PM ^
Of course this is Michigan we are talking about and we all know that a 12.8% chance of losing is really closer to 92.8%!
November 22nd, 2020 at 11:10 PM ^
I thought about this last night and would’ve went for 2, just to put the game more out of reach.
November 22nd, 2020 at 11:15 PM ^
Force them to take the higher risk play to tie the game. Your odds of winning go way up with an eight point lead and an extra point is high percentage.
Now if we call it a field goal.....fuck.
November 23rd, 2020 at 12:25 AM ^
I would go for 2
November 23rd, 2020 at 12:29 AM ^
I thought they should’ve gone for two in that scenario, but I don’t get paid to make those decisions. Coaches live or die by that stuff. If they would’ve scored the 2PC, everyone would’ve been saying Harbaugh was a genius. If they didn’t score it people would be calling for his head. Kicking the extra point was the neutral between.
November 23rd, 2020 at 1:39 AM ^
This is kind of insane: I've stopped following football completely for a couple of years, haven't posted here in ages, came to the site to see if there was any Michigan basketball news, and saw this at the top of the Board. As it turns out, I developed an (extraordinarily basic) economic model to answer this question 11 years ago right here on this very blog:
https://mgoblog.com/diaries/7-late-game-1-or-2-simple-economic-model
Perhaps someone will find it useful.
EDIT: I posted this before reading the comments. Looks like others have posted information far more useful than mine. :)
November 23rd, 2020 at 3:21 AM ^
What a resource this place is sometimes.