Big Ten Home Field Advantages
Apparently the Big Ten Network's website is running a poll to determine which Big Ten team has the "best home-field advantage". Popularity contests do not good data sets make, so I figured I'd apply a lot of counting and a little math and see what I came up with.
The rules:
- For each Big Ten team, I tallied up their total wins over the last 11* years, and seperately tallied how many of those wins came at home.
- I ignored nonconference games. Those will naturally boost home winning percentages as you invite the baby seals to get clubbed at your house, and play home-and-homes against teams that might actually beat you.
- I wanted to compare how well a team did at home compared to how well it did on average, rather than just totalling home wins and saying "golly, Ohio State must have the best home field advantage because they won at home a lot". Well, unfortunately, they won on the road a lot too, so it doesn't tell you much.
- Of course, the inverse of saying a team has a "Strong home field advantage" would be to say that same team "Sucks on the road". I'm looking at you, Indiana.
*I had planned to look at the last 10 years, but made my spreadsheet a big too large and went on my merry way entering in data. I was all done by the time I realised my mistake and I saw no reason to discard the 1999 season just because it was one more than I had planned to look at.
The Results:
First, and just for the record, here's your overall Big Ten winning percentages for the last 11 years:
| Rank | TEAM | WINNING % | Home Wins | Home Wins Rank |
| 1. | Ohio State | 76.14 | 35 | 1. |
| 2. | Michigan | 67.05 | 32 | 2. |
| 3. | Wisconsin | 57.95 | 28 | T4. |
| 4. | Penn State | 56.82 | 28 | T4. |
| 5. | Iowa | 55.68 | 30 | 3. |
| 6. | Purdue | 51.14 | 27 | 6. |
| 7. | Northwestern | 44.32 | 22 | 8. |
| 8. | Michigan State | 43.18 | 23 | 7. |
| 9. | Minnesota | 39.77 | 20 | 9. |
| 10. | Illinois | 34.09 | 16 | T10. |
| 11. | Indiana | 23.86 | 16 | T10. |
Yeah, I know. I don't like it any more than you. Anyhow, as you can see, there's not a lot of difference between a team's overall rank and its rank in terms of raw number of home wins. A bad team is a bad team at home or on the road, and ditto for a good team.
Surely there must be something to the fearsome reputations to such locations as Beaver Stadium and the Horseshoe though, right?
At first, I tried expressing home field advantage as the percentage increase of home winning percentage over total winning percentage. However, I found that this simply weighted the home success of bad teams much higher. Instead, I totaled the number of wins each team had at home, subtracted the number of wins each team had on the road, and averaged over 11 years to yield a number I'm calling the Expected Increase in Wins at Home (EIWH). In other words, every year each team plays 4 Big Ten home games and 4 Big Ten road games. How many more wins, on average, does a given team expect to claim at home than it will on the road? The results are as follows:
| TEAM | EIWH |
| Indiana | 1.00 |
| Iowa | 1.00 |
| Purdue | 0.82 |
| Michigan State | 0.73 |
| Penn State | 0.55 |
| Michigan | 0.45 |
| Minnesota | 0.45 |
| Northwestern | 0.45 |
| Wisconsin | 0.45 |
| Ohio State | 0.27 |
| Illinois | 0.18 |
The results have some suprises. Iowa, a slightly-above-average team overall, earns an average of one more win at home than it does on the road, as does celler-dwelling Indiana. Indiana has only won five Big Ten road games in the past 11 years. Iowa has a reputation as a tough place to play, especially at night, but the Indiana results are inexplicable.
On the other end of the spectrum, Illinois has only earned 16 of its 30 victories at home, which makes for an interesting contrast with Indiana in spite of the two school's proximity at the bottom of the overall standings. Strangest of all, the feared Horseshoe in Columbus grants a very modest advantage to the hated Buckeyes. They have less of a home field advantage than such teams as Northwestern (a school which, from my personal experience, barely fills half its stadium with home fans) and Minnesota (who played in the sterile Metrodome for all of the period of this study).
What's the message here? It seems that the level of hype attached to particular stadiums has little relation to the advantage those stadiums grant to the team playing there.
Buckeye fans tore down one of our (I'm an IU alum) goalposts the year they clinched Cooper's Rose Bowl berth in Bloomington. The team had to return to the field to defend the other post. One of the more ignoble things (there are many) that have happened to the Hoosiers in the past generation.
While IU is bad to awful at home, their lack of success there is dwarfed by what they've failed to accomplish on the road. It's not that there's much of a homefield advantage for IU - the stadium almost never sells out, and Hoosier fans have been outnumbered by the opposition fans.... it makes me wonder if the last three coaches effectively conceded the away games and spent all their prep time on games in Memorial Stadium.
A ten year comparison of Big Ten stats should have eleven years. Pretty soon we'll need to include twelve years in every ten year study.
what you did there, now thats funny I don't care who you are
Your system punishes teams that win on the road (good teams) and rewards teams that lose on the road (bad teams). If Indiana has a 1 win season, do you think that is going to come at home or on the road?
I dont think he is punishing anyone. I think what he is trying to say is that "Home Field Advantage" is more hype than actual "advantage". A good team is a good team at home or away and a bad team is a bad team at home or away.
I'm interested to know the strength of schedule related to each home game. I could see there being some inbalances based on who is rotated off certain schedules in certain years.
And to that same effect, how many times did each team win as an underdog at home? That could be tougher to find. The road compared to home winning percentage does a good job of somewhat explaining what I'm looking for, but it's not a perfect metric either.
That shouldn't matter a great deal, since in 2 years, you play the same in-conference teams home and away once each. If you get a cupcake, it's likely to add one win to each column; if you get a hard team, you're going to be more likely to lose both home and away. Over time, the effect should basically cancel out. The only reason that's not perfect is that there are occasionally huge swings in how good a team is from one year to the next and, since there were an odd number of years in the study, there's one year where the home/away games aren't balanced by another year, but overall it should be pretty negligable.
the effect of cupcakes, paid to come get slaughtered at home, will make it seem that teams like Michigan have a better home field advantage. What you should have done is take the win percentage of all teams at home in Big 10 conference games, and compare that to their win percentage in B10 conference games on the road. The difference will more or less give you the home field advantage each team enjoys. You current method is definitely flawed though as it A) rewards you for losing on the road/hurts you for winning on the road and B) is effected by the number of home vs. away games a team has each year, which varies across the league.
If you'll read what I wrote, you'll see that I did, in fact, only consider big ten conference games for the exact reasons you list.
I think the OPs topic is interesting. I'm sure the Mathlete could lend more advice in this area, but there's two ideas for a statistical analysis that the OPs diary made me think of:
1. A regression analysis where the dependent variable is points scored and one of the variables of interest is whether or not the game was home or away. The coefficient on the home/away variable would show the difference in points scored whether or not the game was home or away.
Essentially, using this we would be able to say playing a home game is associated with scoring x more (assumed it would be more) points than an away game.
2. A regression analysis where the dependent variable is whether or not a team won the game. Again, an independent variable or interest would be whether or not it was a home or away game.
Using this setup, we would be able to say, playing a home game is associated with an x percentage point increase (assumed it would be an increase) in the probability of winning.
With both, we could also find the effect of say, number of starting seniors, etc, on points scored per game, or probability of winning.
Additionally, there'd have to be a lot of other variables that are included to control for stuff, like opposing teams rank, etc. If you guys think its interesting, I could definitely start working on it.
I don't think regression is even necessary (or wise - you'd have to assume a fair amount about the populations to get anything meaningful). I tend to believe that scores don't mean jack in CFB because differential scores reflect a tendency to run up the score in garbage time. Without having access to a play-by-play database (from which one could excise garbage time scores and get a better measure of score differential), I'd actually prefer to simply measure wins and losses and do some non-parametric tests (too many indeterminate nuisance factors in CFB; assume as little as possible!) to compare home and away records to acquire a p-value for each team. Then you could compare the p-values to determine the relative confidences (which is pretty dodgy and has no statistical rationale).
In fact, I may do this myself.
I agree with you, a regression analysis is probably not wise, and may not even be helpful.
I always thought that simpler is better when doing these sort of things. The idea that comes into my head is simply plotting Home Winning % vs. Total Winning %. Then draw a line with slope = 1 , and intercept = 0 on that graph. The further that you are above the line, the better your home field advantage. If you are below that line, you hate your fans. Another benefit to this idea, is this data has to be readily available somewhere out there on the interwebs, just waiting for someone to put into graphical form. One more benefit to doing this is that it removes the "ceiling problem" that someone else mentioned (almost...it is totally ineffective in years that you are undefeated...which isn't a big problem for most teams).
Thoughts?
...when I came across this post. (The players have to think one game at a time, not me.) And I was hoping that RR could survive '10 to get to '11, when UM should be really competitive BLAH BLAH. And smiling that OSU will be coming to the Big House, et cetera.
Then I thought, gulp, what if Michigan's first game against Nebraska is in Lincoln? Not saying it wouldn't be a W. And maybe expansion will set up an even more advantageous schedule, if either (at) Wisconsin or (at) Iowa fall off temporarily. But there is uncertainty.
But, f-it. Like RR says, "They'll tell us what it is and then we'll go."
Still, I'd rather welcome Nebraska to the Big Ten and the Big House simultaneously.
I don't want to see Michigan play Nebraska in a conference game in 2011. Hopefully, this sets up an inaugural Big Ten Championship game matching undefeated Michigan against undefeated Nebraska with a MNC berth on the line.
That would be a great way to start the expansion era of the Big Ten.
The study is set up very well and is very interesting. The only caveat I would add is that there is a ceiling problem--maybe the Horseshoe is the single hardest place to play in the world, but Ohio State does so well in their away games that there's really not much possibility to improve at home. Ditto for Michigan if you don't count the last couple of years--the better you are, the harder it is to improve.
I would expect that the better a team is, a bigger home field advantage would be necessary to keep the same EIWH. It would be hard to really show that without a very large study, so there's not much to do in order to address it, but I think it's still important to note the limitation.
I think you're dead on. I appreciate the OP's work, but I think that ceiling is hard to get around. Indiana having the best advantage in this system isn't at all surprising to me. They're terrible, so their expected road wins are essentially zero. Consequently, any games they manage to win are almost definitely going to be at home. I guess that is a home field advantage to some degree, but that degree is hard to quantify I think.
I don't necessarily have a better idea, but I think you have to go away from wins and losses. Some ideas off of the top of my mind would be to compare results against the spread home and away and see what that shows (obviously that introduces the variable of what Vegas things), comparing points scored at home or away, or (my favorite) have Mathlete use his expected points per play to compare how teams do on the road compared to at home.
Anyway, this is a tough one to quantify.
The horshoe should actually have more credit than it's given, because of the fact that the people that inhabit it are used to being loud..
always have an advantage when it comes to weather, seeing as they're used to playing in such frigid conditions, those SEC babies always falter when the come up here, and also our time, effort, and money into large stadiums (cough* Bighouse... Happy Valley.... Horshoe)
also keeps the fan base large
When was the last time a SEC team played in a Big Ten stadium, other than Kentucky playing Indiana, which no one cares about? I can't think of a single one in the past 20 years that has played in the Big House, and we haven't even hosted a southern BCS team that I can remember since Florida State in 1991.
Vandy - if you want to count them as an SEC team, they are, well Vandy...
The SEC will win the high-profile matchup between the Big Ten and SEC this year. Alabama is losing a lot on defense, but I still wouldn't want to be Penn State's new QB facing that D.
to me this should be a measure of how difficult it is to play at a particular away stadium for the road team, not how good each team is at home. To put this in mathmatical terms: how much more does team A lose at stadium B versus stadiums C-J compared to their average away record. what you did was look at how much more Team A wins at home, a slightly different measure
i think that interpretation will correlate more with our feelings of how it is difficult to go into the shoe and win or travel all the way to iowa.
somewhere in that equation you have to control for the relative strengths of each team (maybe divide by the teams' overall winning %) so that it doesnt just put the best teams at the top of the list, but i think there is a way to do it
Comments