Turnover margin doesn't explain losses
So out of boredom and nostalgia for better days, I looked through M's turnover margin over the games:
WMU: +2 (W)
ND: 0 (W)
EMU: 0 (W)
Indiana: -1 (W)
MSU: +1 (L by 6, OT)
Iowa: -4 (L by 2)
DSU: -1 (W)
PSU: -4 (L by 25)
Illinois: -3 (L by 25)
For all the fuss over the disastrous special teams and fumbles, this really can't explain any of M's losses except for Iowa. Unless each turnover was a net swing of seven points for the PSU game (by generating a PSU touchdown or stopping what would have been a sure M touchdown), they didn't make the difference, and the Illinois game was just an abomination (which I saw in person) in which even seven points per turnover wouldn't have made up the difference.
Unless you assume that M should have a positive turnover margin, it's really more of a systemic problem across the board (but mainly the defense). The general lack of hands across the team could explain it, mainly because that kills drives as well, but it's more than just fumbling the ball.
November 2nd, 2009 at 2:53 PM ^
Someone borrowed (and never returned)my Diaries of a Michigan Fan book but IIRC Craig Ross studied the effect of turnovers by type, as in fumbles v. interceptions and found that fumbles appear to harm a team to a much greater extent than interceptions. IT would be interesting to see T/O margin as it relates specifically to fumbles.
BTW, I think it is time for a diary of a michigan fan part II.
November 2nd, 2009 at 2:58 PM ^
Could it be that fumbles are more costly simply because they occur where the ball is located on the field rather than down the field like an interception? Just from a field position perspective, it seems clear that a fumble at the line would be much more costly than a pick 15-20 yards downfield (assuming no return).
November 2nd, 2009 at 3:10 PM ^
I wish I had my copy of the book so I could tell you whether or not that was taken into account. In theory though, where the turnover happens probably makes a difference. A fumble on the opponent's 1 yard line is probably less costly than an interception thrown inside your own 10 (this is assuming your defense can prevent a 99 yard drive . . .)
November 2nd, 2009 at 3:46 PM ^
Well, I'd actually say that turnovers closest to no man's land are the least important. If you fumble on the one yard line on first and goal, you're losing close to seven points, since you'd typically get a touchdown. Not scoring seven points that you would have gotten is the same as your opponent scoring seven extra points on you. It's really the opportunity cost of what you were going to do; turnovers on 1st down are worse than turnovers on 3rd down, assuming your offense is usually non-negative.
This is one of the reasons that turnovers did matter in the Iowa game; each team got seven points off of turnovers, but the other four turnovers likely prevented at least some Michigan points. Still, that's the only game where it made a difference out of the four losses; being 6-3 right now following two blowouts would not make everyone a happy camper, although it would take some of the edge off. The problems with the team are greater than just turnovers.
November 2nd, 2009 at 3:01 PM ^
So, you think there is absolutely no correlation with the fact that Michigan has gone 1-4 when its turnover margin has gone in the tank?
November 2nd, 2009 at 3:04 PM ^
Inconclusive will be his response.
November 2nd, 2009 at 3:33 PM ^
I didn't say that; read more closely. Straw men are people too. My point, which is in the post, is that turnovers alone don't explain the massive beatdowns in the last two games, and we were somehow positive in turnover margin in the loss to MSU.
Correlation =/= causality. That's a basic concept.
November 2nd, 2009 at 3:11 PM ^
is the only unexpected result in your list. With 4 additional turnovers, you'd expect a loss by more than two points. M's strong running that night, good red-zone performance, combined with Warren's pick six on Iowa's lone turnover made that a closer game than you'd expect.
November 2nd, 2009 at 4:08 PM ^
When the turnover margin is -1 or better, Michigan is 5-1.
When the turnover margin is worse than -1, Michigan is 0-3.
Causation? Possibly not. Correlation? Absolutely.
November 2nd, 2009 at 4:10 PM ^
The Illinois margin is a little skewed because of that last TD by Illinois when everyone just failed to tackle the guy at the line. That was just a random play that should never have happened. If you exclude that meaningless score, the margin was 18 points, which could tied to the -3 turnover margin.
To a greater point, I think the turnovers are killing this team because they (a) are not good enough to give up/leave points on the field against even bad teams, and (b) they come at the worst possible situations. Tate's first fumble completely deflated the team; for all the Patrick waxing about momentum shifts, the team still had a chance down 28-13 to start the 4th quarter. Score a TD there, suddenly the game is quite winnable and maybe Illinois gacks it up on the next series. But instead, the team fumbled the ball away and with it the game.
November 2nd, 2009 at 4:43 PM ^
Does this include turnovers on downs? It's not a true turnover, but it might as well be counted as one. I can think of at least one in the MSU game that would make that margin 0.
November 2nd, 2009 at 6:17 PM ^
I'm not sure; I just stole the turnover margin from the ESPN recaps of games, and I don't know how they track those.
November 2nd, 2009 at 6:08 PM ^
I did some research before the season started regarding turnovers. Unfortunately, I cannot find the spreadsheet at the moment. However, I do remember one piece of information that supports the OP. When I compared turnovers between 2007 and 2008 there was one huge outlier, Baylor. Baylor moved from a
-1.5 turnover ratio to +1.3, almost a 3 turnover swing per game. The interesting bit was all of the extra turnovers they gained resulted in only 1 more win. My current hypothesis is that good teams can overcome their mistakes but I will have to find time to dig deeper to test this.
November 2nd, 2009 at 7:11 PM ^
at somewhere between 3 - 4.5 points. So on average for the season, it's been worth about the same as home field advantage is worth per game. Against Iowa, PSU and Illinois, though, it's been worth 11 points per game. And based on a rough calculation I made of points to wins, the TOM problem will be worth -1 win on the season (presumably Iowa). Instead of a shot at 8 wins with some luck, we're trying to get to a bowl game.