Road Team Winning Percentage, based on number of previous visits
Posted by Jason Lisk on Wednesday, November 4, 2009
I've spent alot of time talking about familiarity and home field advantage. Most recently, I wrote about the New York teams sharing a stadium. Today, I'm going to do mostly a data dump of other research I had done on the familiarity issue.
From 1997-2003, a flurry of new stadiums opened in the NFL, with the following franchises playing in a newly built stadium: Washington, Baltimore, Tampa Bay, Cleveland, Tennessee, Cincinnati, Denver, Pittsburgh, New England, Detroit, Seattle, Houston and Philadelphia. I pulled the results for those thirteen stadiums, in terms of how many times the visitor had previously visited. I didn't use exhibition data, but did use all regular season and playoff visits. These numbers are through the 2008 season and playoffs, and do not include results this year. The win percentage is the wins and losses from the perspective of the visitor.
Visit W L win pct 1 134.5 242.5 0.357 2 60 113 0.347 3 50 54 0.481 4 29 34 0.460 5 20 30 0.400 6 19 25 0.432 7 21 18 0.538 8 6 10 0.375 9 7 7 0.500 10 4 5 0.444 11 3 2 0.600 12 2 1 0.667
These are the overall results. I divided the games out into divisional and non-divisional games. We know that divisional opponents get to play at a venue every year, whereas non-divisional opponents typically do not. There is a bit of a grey area when it comes to teams that were division opponents until the realignment in 2002. I treated games as divisional games if the road team was in the same division that season. So, Tennessee is a divisional road opponent of Baltimore for games before 2002, and a non-divisional opponent after 2002.
Here are the division opponent only results, sorted by number of prior visits to the same stadium:
Visit W L win pct 1 22.5 31.5 0.417 2 17 37 0.315 3 28 24 0.538 4 24 23 0.511 5 18 22 0.450 6 15 24 0.385 7 19 18 0.514 8 9 9 0.500 9 6 9 0.400 10 7 5 0.583 11 3 2 0.600 12 2 1 0.667
There are a small number of games once we get beyond five visits, but the combined win percentage of division opponents when visiting these stadiums for the fifth time (or more) is 79-90 (0.467). Also, with a couple of exceptions, every one of those first visits by a division opponent was in the first year that the stadium opened. We see that the division opponents actually perform worse in the second season.
Non-divisional opponents, on the other hand, will tend to have their first visit to a stadium after the home team has already been playing there for at least a season. Here are the non-divisional results, based on number of visits.
Visit W L win pct 1 120 211 0.363 2 47 78 0.376 3 22 34 0.393 4 9 12 0.429 5 4 10 0.286 6 4 4 0.500 7 2 3 0.400 8 0 1 0.000
If we exclude the first time visits that occur during the first year of a new stadium (when the non-divisional visitors went 24-29), then we get a 96-182 record (0.345 win percentage) for all non-divisional teams in their first visit to a new stadium more than a year after it opened.
This entry was posted on Wednesday, November 4th, 2009 at 11:32 AM and filed under Home Field Advantage. Follow comments here with the RSS 2.0 feed. Skip to the end and leave a response. Trackbacks are closed.
I think data is too noisy and sample sizes too small to draw any actual stastical meaning. But the idea makes sense. Though I wonder how much the non-divisional part is caused by things like more travel relative to divisional opponents.
Posted on 04-Nov-09 at 12:21 pm | PermalinkI don't disagree with any thing you said. Consider this a file drawer issue. I'm trying to think of ways to look at this. This is one of them.
The problem is that, unlike a drug study, where we can get thousands of patients across trials and get statistically significant results, I just can't get these half of these pesky owners to build a stadium every five years, while having a control group with half of them using the same stadium forever (Though I thank you, state of California, for trying).
It's the same issue with running back workloads. These coaches simply won't give enough running backs 400 carries so we can get the sample size up.
Here are some of the things I've considered:
looking at number of visits in X period of time:
But then, is a visit one year ago different than two visits three and four years ago? The sample sizes on the combinations would be too small. I'm guessing there are few occasions where a team has played a non-divisional opponent 3 years in a row in the same place, at least not enough to be worth anything.
Posted on 04-Nov-09 at 5:04 pm | PermalinkWhat struck me is that for both the total and the divisional opponents, the road team win percentage was higher with just one prior visit than with 2. I guess this reflects the home team gaining familiarity with their new stadium as well.
Posted on 05-Nov-09 at 10:11 am | PermalinkI hate doing this, but you left out a team that moved into a new stadium; in 2002, the Bears moved from old Soldier Field to Memorial Stadium in Champaign. Then, in 2003, they moved back into NEW Soldier Field. Does this not count for your study?
Posted on 05-Nov-09 at 11:10 am | PermalinkI left it out because I think there is a debate as to whether it is really a new stadium. I'd consider it a renovated stadium. It's at the same location. If I went back a few years, I could pick up the Raiders return to Oakland, for example. New stadium or not? Well, team's have played there before, it's just been a while. I decided to just go with stadiums where no games had been played before.
Posted on 05-Nov-09 at 11:26 am | PermalinkThanks for the answer, Jason. I don't know if I agree with you, but at least there's logic behind your decision.
Posted on 05-Nov-09 at 7:44 pm | PermalinkGreat stuff as always, Jason! A couple of ideas: To account for the home team gaining familiarity in their first few years in a new stadium, calculate the difference between road team's number of visits and the home team's number of games (perhaps capped at some point).
What if you used the player rosters on p-f-r to compute average familiarity of each team's starters with the stadium for each game?
Posted on 06-Nov-09 at 12:11 am | PermalinkAnother thing to look at is how team's do in their 2nd game at Giants Stadium. The Falcons play @NYG and then @NYJ, while the Panthers play @NYJ and then @NYG this season.
Posted on 07-Nov-09 at 3:28 pm | PermalinkDon't you need to account for the strength of these 12 teams, and the visiting teams? These twelve teams, in their 2nd, 3rd, and 4th seasons in the new stadiums (when many of the first-time visits must have occured), were a combined .564 overall. If you assume a HFA of 70 points, we'd expect them to be about .634 at home -- not that much lower than you found. If the first-time visitors happened to be slightly below average (which they likely were, since these 12 home teams are above average), that could account for the remaining gap.
I can believe HFA advantage is weaker in year 1 of a new stadium, because the home team doesn't yet feel at home. But an ongoing first-time-visiting-new-stadium effect on visiting franchises seems unlikely. A lot of out-of-conference players must be visiting a stadium for the first time in every game (because they are younger players). Conversely, even in a first-time visit by a franchise, some visiting players will have been in the stadium before (when playing for a different team). So how big an effect is plausible?
Posted on 08-Nov-09 at 6:32 pm | PermalinkThis is really great work...
Posted on 11-Nov-09 at 2:49 pm | Permalink