## Trivia: Home/Road Wins in Division Games

Assume Team A and Team B are in the same division. In the first regular season matchup, Team A plays Team B at home, and wins. If this is all you know, how likely would you say Team A is to win in the rematch?

On one hand, we now know that Team A has to travel on the road for the rematch, and road teams win about 43% of the time. But we also have some evidence that Team A is better than Team B, so how does that impact things? And what about the idea that it’s hard to sweep a team — does that play into things?

I looked at all division matchups from 1970 to 2015. There were 1,297 times when the home team won the first matchup among division opponents — let’s call that team, Team A. What was Team A’s record in the rematch on the road at Team B?

Take a second to think about it.

….

If you’re so inclined, post your thoughts/guesses in the comments before clicking. But now if you are dying to know the answer…

Now, what if Team A loses at home? Well, that changes things, obviously. Now we have evidence that Team A is worse than Team B *and* Team A has to play on the road in the next game. About the only thing in Team A’s favor would be the “tough to sweep” idea. So what do you think the record of all Team As have been in this case?

There have been 1,018 such cases. What is the winning percentage of Team As in the rematch?

Again, if you are interested, post your thoughts/reasoning in the comments before clicking.

Trivia hint 1 Show

Finally, there have been 40 instances where Team A tied Team B at home. What is Team A’s record in the rematch?

Trivia hint 1 Show

Anyway, I don’t know why, but these sorts of things interest me. Let me know if there are queries in this area of the football universe that interest you, too, and I’ll see if I can run them.

{ 9 comments }

I find this type of thing interesting as well, so don’t stop!

I this case, I have a hunch Chase is being a trickster and trying to make us overthink this. Before clicking the answer, I predict that Team A wins the road rematch 43% of the time and wins the home rematch 57% of the time. In other words, the result of the first game has no bearing on the second.

Edit: Boy, was I wrong

• Richie

That was my line of thought – that the first game didn’t really matter.

• Guesses (typing as I go): 50%, 36%, 17-23.

Rationale 1: A fair number of those teams won the first game because they were better rather than just because they were at home. I’m just taking a wild stab that this factor essentially cancels out home field.

Rationale 2: A fair number of those teams won the first game because they were better, so I’m going to guess that it’s roughly the same effect as I guessed above because I might as well be internally logical while I’m likely being wrong. If road teams would otherwise win 43% of the time and I was giving a 7% bump above for many of the teams being better, that adds up to 36%.

Rationale 3: The only real information we have here is that B is at home, so I’m going to guess it matches the 43% road winning percentage, which is closest to 17-23.

After reading through the answers, my totally pulled-out-of-thin-air guess at the effect of some of the wins being by better teams was pretty close, though it is interesting to me that the sort of “effect size” is different. The first is .077 while the second is .107, almost half again as big. So why is it apparently more predictive to know that Team A lost game one? Is it because a number of those second games are late in the season when the team that we already know has a loss is more likely to be giving up on the season? That situation doesn’t seem common enough to make up that entire difference. Home teams will get more fluke wins than road teams, but that should be part of what we’re accounting for in that initial home field advantage. So . . . I really have no clue.

I’m going to just say “sample size” at the ties. I bet over time we would find that it’s closer to 43% than 50% for road teams. I’m somewhat tempted to go run a Fisher’s LSD to see our statistical power in that sample size, but that seems a bit much. 😉 (And I only vaguely remember the test. My statistics and research methods professors were very anti the Fisher’s LSD.)

I like these kinds of thought experiments, like when Tom Tango asked fans to choose a pitcher and it gave a sense of what our intuitive idea of replacement level was.

Chase – what is the difference in HFA for division games compared to non-division games? Maybe I have selective memory, but it seems like the road team wins more often within the division, which would make sense considering the level of familiarity between the teams. Especially Bengals / Steelers where the home team almost never wins!

• Richie

Yeah, I think I saw that someplace as well. One thought is that the divisional foes are more familiar with the stadiums than noon division.

But that doesn’t mesh with the conclusion from “Scorecasting” that the majority of HFA comes from influence on referees.

I would think that shorter travel and stadium familiarity should help the road team in division games. I’ve read the HFA analysis from Scorecasting, and I can’t say I entirely agree with their conclusions. Referee bias certainly plays a part, maybe the most important part, but I think they’re underselling the effects of cross country travel and the emotional boost the home team gets from a raucous crowd.

• Richie

Is there any validity to “tough to sweep”?

• JeremyDeShetler

I’ve wondered about that, and also about the 3x-sweep. I remember hearing the “it’s difficult (impossible?) to beat a team 3x during a season,” but thinking that it rarely seemed to happen so why did anyone bother to repeat the expression. There’s a Doug article on the PFR blog (http://www.pro-football-reference.com/blog/?p=458) about the 3-sweep. At that point (Jan 2008), it had happened 11 (of 17) possible times.

Not sure oft-hand how you would test the ‘tough to sweep’ (2 or 3 times) theory. Only thing I can think of is to take all home-and-home series since the merger (including the 5th place team home-and-homes from 1978), figure out the odds of either team sweeping, then look to see how many sweeps there actually were. Not sure how instructive that would be though.

• Richie

I guess you’d need to do something like look at each team’s “natural” ability. See how often teams of similar ability win games (home and road) against non-division opponents. Then compare that to the winning percentage of all the teams that won the first game of a divisional matchup in their rematch game.

And maybe you could simplify “natural” ability as “W-L record in the 14 games not against the potential sweep team”.