Why does your win-odds chart start with Michigan below 50%?
"I love it that Ivy League coaches are coming to our camp and Big Ten coaches are coming to our camp. South Florida is coming. We've got about 70 schools that are coming to our camp."
Pictured, the closest defenders Jeremy Gallon saw all day [Upchurch]
|Field Pos||Early Conv||Bonus Yds||Avg 3rd Dist||Adj 3rd Conv||Red Zone|
So, that was a lot of bonus yards. Michigan’s 385 yards beyond the first down mark was the highest mark in college football this year (largely because the game was close and all second half plays counted, unlike say every Baylor game this year). It was the 11th highest total since 2003 and narrowly edged the 2010 Indiana game (by 2 yards) for the highest mark for Michigan in the time period. For comparison, Purdue has 399 bonus yards on the season so far.
Add in Indiana’s 270 yards and this was the highest total for any game this season and the 13th highest total since 2003. There were a lot of big plays in this game. Also, not a lot of third downs, especially long ones and the ones that did occur were converted at a high rate, even after accounting for their short distance. The offenses may have done OK in this one. The important thing is that Michigan’s offense was better than Indiana at virtually all of these things, even if Indiana did pretty well themselves.
Devin Gardner: +29.9 EV+, +118% WPA (1st)
The top quarterback score of the season, even after adjusting for what the average offense does against Indiana. This game finished as the #10 QB performance since 2003 and the second best B1G QB score behind CJ Bacher against Michigan State in 2007
Fitzgerald Toussaint: –1.8 (+5.3 before opp. adjustment), +11% (90th)
On the one hand, it was great to see some success from Toussaint, on the other hand, after adjusting for the Indiana factor, it was actually below average. At this point, my expectations are pretty low for the traditional running game. Toussaint had some nice success but given how bad Indiana was and that he still was only at 4.0 YPC until the final run, I don’t see things getting significantly better for him down the brutal November stretch.
Jeremy Gallon: +29.5, +84% (1st, obvs)
So this was a nice game for #21. This blew the doors off of any other receiving performance in the last 11 seasons. Gallon broke Stedman Bailey’s record set last year against Baylor by nearly 15%. I put together a quick chart of Gallon’s projected yards throughout the game.
After the first 70 yarder the pace shot up to nearly 600 yards and never dropped below 300 from then on. It was pretty amazing to see all the On Pace For jokes throughout the day and see that in the end, some of them undershot the final total.
Devin Funchess: +6.8, +15% (49th)
Tre Roberson: +15.5, +54% (4th)
Nate Sudfeld: –0.3, –6% (63rd)
Tevin Coleman: +5.6, +14% (11th)
Kofi Hughes: +9.3, 29%, (22nd)
Cody Latimer: +7.4, +19% (39th)
The six biggest plays of the game:
6. 11.1% Roberson runs for a 15 yard TD/Gardner hits Gallon for a 50 yard TD
5. –11.5% Tempo hits Michigan for the opening 59 yard TD pass
4. +14.2% Gordon picks off Sudfeld
3. +17.1% Gardner to
Roundtree Gallon for 70 yards
2. –18.3% Michigan fumbles the snap at the doorstep of the goal line
1. –20.3% Roberson to Hughes (through Stribling) for 67 yards to bring Indiana within a two point conversion
Last week featured a couple of coaches who punted to win. Both Butch Jones at Tennessee and Bronco Mendenhall at BYU called for punts while trailing with under 4 minutes left in the game. Those made not have been the optimal choices, but both coaches managed to get the stops and convert the stops into game winning scores.
Brian Kelly went for the punt on 4th and 3 from the USC 38.
Northern Illinois and CMU will share this week’s award. Northern Illinois punted on 4th and 3 from the CMU 35 yard line. This happened in a game where Jordan Lynch set the single game rushing record and averaged nearly 10 yards a carry on 32 carries! CMU shares the award by punting away to Northern Illinois with 5 minutes left while trailing by 13 points. Somewhere Gary Andersen is nodding in approval.
Bonus Dumb FG of the week: Arizona State took it to Washington early, and Washington was facing a 4th and goal from the 9 midway through the third trailing by 20 points. Rather than push to truly get back in the game, the Huskies opted for the chip shot field goal to cut it from a three score lead to a smaller three score lead. Going for in on 4th and Goal from the 9 is usually not a no-brainer, but when trailing by 20 in the second half, there isn’t a great case for the field goal.
The Six Factors for all teams can still be found here
With no game this week, we’ll take a look at the season projection for Michigan. My numbers are more optimistic than most on the prospects for the remainder. I have Michigan as a favorite in all remaining games, with @NW, Nebraska and @Iowa listed as reasonable ~60% odds. My numbers aren’t big fans of either Michigan State or Ohio State, largely due to schedule concerns for both and offensive woes for Michigan State.
Next week I’ll dig more into the MSU numbers, but for now I have listed the original numbers as yellow and optimistic projections and a more realistic assessment (noted here as Pessimistic) of Michigan’s chances versus its two main rivals. Depending on your take on the games, this season is projecting to 9-10 wins, but with all the remaining games competitive, a lot of different outcomes are still on the table.
Why does your win-odds chart start with Michigan below 50%?
I think I can answer this.
Simply possessing the ball provides an immediate, slight advantage at kickoff.
then wouldn't teams want to open the first half with the ball, when in fact they almost always want the ball to open the second half?
But if you notice, we got a little bump immediately after halftime, too, so those initial possessions balance out.
It doesnt make sense that Michigan would get a bump immediately after halftime unless the probability was being done incorrectly...it's not like it was a surprise that Michigan got the ball to start the second half so that fact would have been known when the teams walked into the locker room at halftime.
It would make sense only if the probabilities being calculated ignored the fact that Michigan was going to get the ball to start the second half..in that case, the halftime probability being calculated based on 28-17 lead with neither team having the ball even though it was known that Michigan was going to get the ball to start the second half.
I guess if we were going to get down to real small probability differences then the fact that Michigan didnt fumble on the kickoff improves their chances by tiny tiny amount relative to going into the halftime.
Technically, there wasn't 100% certainty that Michigan would get the ball, since we deferred our decision until the second half - we could have elected to kick.
Also, there's the chance of an onside kick or fumble. In fact, I would imagine that actually getting the ball back at any point in the game results in an immediate, yet small, bump in your chance at winning the game.
Interesting question. I'm guessing the other team gets an identical advantage push when they receive the kickoff in the 2nd half. In that case it wouldn't matter. But, this is assuming your odds of scoring on the first possession of the game and the first possession of the second half are the same. Perhaps they're different.
This is a spot on observation...unless the probability calculation is completely ignoring the fact that Michigan will start with the ball in the second half.
Also... how much risk did Fitz put us in at the end of the game by scoring a touchdown (taking us from a 9 to a 16 point lead, but forcing us to give the ball back with a full minute left) rather than going down after picking up a first down, which would have allowed us to kneel the game away? Not very often a touchdown decreases your odds of victory, but that play must have done it.
Not sure I agree. Unless you're saying that IU was guaranteed at least 7 points on their next possession, how could that TD have possibly hurt M's odds to win?
We could have kneeled it out at that point (or run the clock out). So, aside from a fumble, that would be a 100% chance of winning. Scoring to go up 16 with under a minute left still leaves the possibility of Indiana scoring two touchdowns and getting both 2pt conversions to send it to overtime. Our chance of winning probably dropped by 1%.
yup, that's my point exactly. To my knowledge no team has ever fumbled away the ball away while in victory formation. FItz scoring a touchdown gives Indiana a very small chance of winning (less than 0.1% I'd guess), but it's still more of a chance than they would have had if we got to kneel-down mode.
There was a game in which a team fumbled the snap on a kneeldown - 2010 Oklahoma State vs. Troy. OkSU fumbled the snap and Troy recovered. OkSU ended up winning though, so it was mostly forgotten.
That's the only time I know of it happening, though, so a team's odds of winning in the victory formation have to be about 99.9%.
Interesting! Thanks for mentioning it... found it on youtube after you provided the details.
....but it cost me a game in my pool. I took Indiana +10.5 and it was in the bag until Fitz took it to the house. I got over it pretty quickly.
I guess you're right. If he takes a knee, the game is essentially over (but I think that IU had one TO left). By giving them the ball, they have a chance at a TD, 2 point conversion, onside kick recovery, TD and 2 point conversion to tie the game. Stranger things have happened (I know because I was in Happy Valley two weeks ago).
Way too optimistic Mathlete re season W/L projections. That may be what the numbers indicate. Let's not let one game where we certainly didn't decisively beat Indiana (they could have taken the lead in the 4th before Gordon's pick) cloud PSU, Akron and UCONN...
This team is an underdog vs MSU and Ohio, and will be a coin flip with Nebraska and NW. 8-4 will still be what most expect.
"My numbers are more optimistic than most on the prospects for the remainder."
He did mention that his numbers are optimistic. I think you are being a little too pessimistic at this point, though. I think it's really going to depend on how many turnovers we have in each game. The Akron and UConn games would have been blowouts had it not been for turnovers and PSU was a win if not for 3 missed FGs. Granted, turnovers seems to be the MO for this team, but it's not like that's something that can't correct itself midseason.
Mathlete - how do you get all the data about games from around the country? I assume you have some formula that uploads it all automatically but I'm curious how it's done.
Still see 8 wins as the ceiling. This team is not capable of winning on the road.