Support MGoBlog: buy stuff at Amazon
Was Michigan lucky or unlucky last year? Who were the luckiest teams in the Big 10 last year? What teams were the unluckiest nationally?
To try and answer these questions, I took my team PPG values for the full 2009 season and then “re-played” the regular season schedule to see how the season would play out if the teams played at that consistent level and the fluky plays were eliminated. All first half plays and any in the second half with the game within 2 touchdowns were included. Interceptions are included, fumbles are not. Standard special teams plays are included, punt blocks, on-sides kicks etc. are not.
The results were based on the actual schedule (excluding conference championship games or Bowls) and home-field advantage was worth about 3 points per game, based on the actual results.
So what did I find…
Michigan 2009
Michigan was a fairly unlucky team last year. Based on how their play was over the course of the season, on an average year, they would have won 6.3 games (most likely 6 with an outside shot at 7). Michigan’s results were about 1.3 wins below the expectation based on their schedule and their average performance over the full year.
On average, they should have won about 5.1 of their 8 home games. In reality, they won 5 of 8. Notre Dame was a game with a 40% win likelihood based on the full season for both teams. Indiana was a 73% chance. The pickup from those two games was offset by failing to pick up victories over long shots Penn St (34%) or Ohio State (13%) but mostly due to failing to defeat Purdue which Michigan should have won about 58% of the time.
Michigan was only favored to win one (Illinois, 56%) of its four road games and on average, would have won 1.2. Michigan failed to pick up any of the wins, falling to Michigan St (29%), Iowa (13%) and Wisconsin (22%).
For the season as a whole, Michigan went 1-2 in relative toss-up games. They didn’t lose any games they should clearly won but didn’t win any they clearly shouldn’t have. Michigan was the 20th most unlucky team in the nation last year.
Michigan 2008
Very similar to 2009 in terms of luckiness of the results. Michigan finished the year a game and a half unlucky. This team was still not a good team by any stretch, but thanks largely to fumbled punts and 100 yard interception returns, the record indicated a season even worse than it should have been.
The biggest chunk of the unluckiness came against Toledo, where an 81% win probability turned into one ugly loss. Beyond Toledo, Michigan went 2-3 in toss-up (40-60% win odds) games (Wisconsin, Minnesota, Illinois, Michigan St, NW) and 0-2 in longer shot games against ND and Purdue.
Before two data points become an indictment on coaching, the scatter plot of 2008 vs 2009 in terms of “luckiness” does not show any correlation between the two. That doesn’t mean that Rich Rodriquez couldn’t be unlucky or that Pat Fitzgerald should be out buying lottery tickets, but it isn’t saying that with any certainty. Two points may make a line, but not much of a trend.
It is interesting, however, that the Big 10 as a whole exhibited a lot more consistency of luck in 2008-09 than the rest of the nation.
The Big Ten
The luckiest team in the nation resided in the Big 10 this year.
| Team | Luck Variance |
| Northwestern | 2.4 |
| Iowa | 0.9 |
| Minnesota | 0.8 |
| Wisconsin | 0.4 |
| Penn State | 0.4 |
| Ohio State | -0.7 |
| Purdue | -0.9 |
| Indiana | -1.0 |
| Illinois | -1.2 |
| Michigan | -1.3 |
| Michigan State | -1.5 |
| Notre Dame | -1.7 |
Northwestern did not perform like a team that would win 8 games against their schedule. In fact, it was a stretch for them to be bowl eligible. Iowa and Minnesota both came in with about an extra game with Ohio St, Purdue, Indiana and Illinois all managed a game or so of underachievement. Michigan St and Notre Dame proved to be the unluckiest teams in the greater Big 10.
The Nation
Top 10 Lucky Teams in 2009
| Team | Luck Variance |
| Northwestern | 2.4 |
| Bowling Green | 2.3 |
| Wyoming | 2.3 |
| Rutgers | 2.1 |
| California | 2.0 |
| UCF | 2.0 |
| Ohio | 1.8 |
| UNLV | 1.8 |
| Georgia Tech | 1.7 |
| Oklahoma State | 1.5 |
Bottom 10 Unlucky Teams in 2009
| Team | Luck Variance |
| Oklahoma | -3.0 |
| Western Kentucky | -2.5 |
| North Texas | -2.5 |
| Louisiana Tech | -2.4 |
| Ball State | -2.2 |
| Arizona State | -2.1 |
| Tulsa | -2.0 |
| Maryland | -2.0 |
| Colorado | -1.8 |
| Colorado State | -1.8 |
Oklahoma’s season turned out to be unlucky in more ways than one. Despite getting only a handful of snaps from returning Heisman Trophy winner Sam Bradford and losing one of the top tight ends in the country before the season, Oklahoma managed to have the breaks go against them in win column as well. Throughout the season, the Sooners played like a 10 win team, but only managed 7. This is what happens when you have four of your five losses by a total of 12 points and your average margin of victory in your six FBS wins is over 31 points.
Beyond
I think there is a lot of true “luck” that can come into play in these numbers. I do think that with a more substantive history, this will also be a good measure of one of the strengths of coaching. A team/coach that consistently shows overwins, or overlosses, throughout the course of several years would be a good testament to the classic “little things” that is often luck but possibly true for a select group. As I get more years in my database, I plan on returning to this topic and seeing how various coaches stack up on this metric over time.
this confirms what I've been telling friends and family wrt Michigan's '09. and if we'd gotten that extra win, nobody'd be complaining/fretting for this season.
could you run this for '08 too? i assume Michigan was somewhat unlucky that season as well, as expected for any team that performs below average.
Right, that rogue win actaully cost us a shot at a bonus win vs fairly light competition in a December bowl. There's a HUGE perception gap between 7-6 and 5-7. Iowa, Illinois, and Purdue are the games that stick in my craw the most from a "shoulda had that one" perspective.
I'd have to add MSU to that list. The late comeback despite the poor start was set up perfectly for an epic game in Michigan's vault of comebacks.
I'm a bit surprised that Michigan was the 20th-most unlucky team in the country, given that their win total was just one fewer than would have been predicted statistically. In contrast, I thought that the 2008 Michigan team was considerably more unlucky. (Did you do this analysis for 2008?)
Other data support your analysis. In games decided by 3 points or less, Michigan went 1-2. In games decided by 6 points or less, Michigan went 2-3. And that doesn't count the Illinois game, where Michigan was a yard away from going up by 2 touchdowns in the second half. One can quite easily see that, with just average luck, one of the three closer games, or the Illinois game, probably would have been a win.
Butler Butler Butler
great analysis. Always nice to see factors like this and I think the year in year out will most likely show something similar to the turnover factor in that it's extremely random but can have a major impact on the season. Here's to hoping that luck swings in our favor next year so we can hang on to RR for a few more years.
If this were true and "luck" were instead an unobserved quality that some teams have, then we would expect most teams to be either lucky in both years or unlucky in both years, not lucky in one and unlucky in the other. Most teams had the same coach in both seasons, and all had many or most of the same players.
Instead, we see as many teams in the northwest and southwest (lucky one year, unlucky the other) quadrants as in the northeast (lucky both years) and southwest (unlucky both years). This suggests that a lot of what the OP is calling "luck" is indeed luck.
statistic'd.
I agree with you, man
A onside kick recovery against and a miss PAT--two plays that if percentages fall as they should dont happen--in the Purdue game is what separated us from a bowl game and about 75 percent less angst this off season.
Its a fine line. Always.
Also: Great diary as always by the Mathlete
Northwestern's luck was aided by a senior QB. Its funny how teams with experienced and good QBs seem to make their won luck
that was too young/inexperienced/critically lacking at certain positions/unlucky.
it's hard to remember, but Rich really has had nothing but success in his tenure as a coach. it's difficult to think he would, for no good reason, lose the mantle of great coach.
Good post, their were so many things that didn't go our way in the Purdue game. This year though hopefully the little things go our way.
Iowa was pretty lucky last season. Lucky that MSU let them score a TD in like a minute and lucky that they blocked 2 FG in a row against Northern Iowa.
Whew, I saw that and instantly was terrified that Michigan was somehow "lucky" last year--regardless of how horrible their luck actually was.
It's interesting that MSU came in below Michigan in the Big 10. Notre Dame I'll buy, but MSU didn't seem like things went too badly for them (outside of the normalized "Sparty Error" that affects them at all times).
the difference between exceeding expectations and performing below them is a really thin line. Like 3 inches against Illinois. Or an on-side kick against Purdue. Or a clear-headed tate Forcier against Iowa on the last drive. Or a 1 inch longer arm on Brandon Graham against Ohio St and Terrel Pryor's screen pass for the TD.
Great job by The Mathlete as always.
analysis. to me, it makes all those cruel twists all the more meaningful. you get to find out exactly how much it hurt the quest for a bowl. it's like getting punched in the gut all over again...but with the soothing knowledge that it probably won't happen again. probably.
FEEL unlucky...
check the correlation between '08/'09 luck and quarterback experience, or just experience more generally?
I'm sure our luck looked even worse because we had huge ball security problems that year.
I have a hard time believing that Northwestern won 6 games more than it statistically should have over two years.
Also, what is the average luck over all teams you studied. Off the top of my head it seems like luck (as you have defined it) should be a zero sum game. If one team is lucky, his opponent is unlucky. Thus it seems everything should average to zero in the end.
I was thinking the same thing. Here's another thing though; did UM's ppg numbers include baby seal U? That would really skew the ppg out of whack in comparison to the level of competition in the rest of the games.
...the "luck" aspect of football can be boiled down to two or three big plays per game most often seem to dictate the outcome. We just didn't see enough of those last year.
On the plus side, the times they are a-changin.'
I'm not surprised that Northwestern is at the top. They have been infuriatingly lucky ever since Gary Barnett was coaching them.
have my babies? I want little mathletes.
great post. I wonder if you took this further back how would Northwestern, OSU, M appear. M for obvious reasons. Northwestern and OSU because it seems like Northwestern has been getting lucky like this for YEARS. The whole 2001 OSU national championship seemed ridiculously lucky and Tresselball in general seems to get really lucky.
I've found that net turnovers are fairly well correlated with the disparity of the matchup. How to measure the disparity? The closing spread on the game. For every 23 points a team is favored, they will average a +1 in net turnovers. Its a linear relationship.
So if a -23 point favorite suffers a -2 in the turnover category, they've been unlucky to the tune of about 3 turnovers, or about 10 points in that game.
If you run the numbers for all of Div I in 2009 Michigan does come up on the short end of the stick. But I can't recall that they made my Div I "most unlucky" list.
Is that a 23-point favorite (before the game is played) or a 23-point winner (after the game is played)? If it's the latter (after the game) then I think that makes sense because of the whole opportunity cost scenario plus giving the opponent a shot at getting another score. Put more simply: the more the turnover margin is in your favor, the more you should win by. I find it hard to believe that the pre-game spread has a near-linear correlation--or any correlation for that matter--to the turnover margin for that game. Still, it'd be interesting to see those data.
could you send me the data on this for the ACC? I'd love to write a post about this, crediting you, or perhaps you write one and I'll link to that. Great work.
