The Long Arc of Home-Court Advantage

In his debut column for The Fieldhouse this month, Ken Pomeroy wrote:

Teams are protecting their home court less and less these days. In conference games last season, home teams won 59.0 percent of the time. That continues a trend that has existed for at least the last 15 seasons. And I can only say ‘at least’ because I am unaware of a detailed study that goes back further. However, it’s highly likely that last season was the worst year for home teams in the history of the game.

So it’s time for a detailed study that goes back further! Using game-by-game data from Sports-Reference.com, I’ve calculated home-court advantage in conference play going back to 1950. The results back up Pomeroy’s claim: 2017 was indeed the worst year for home teams in college basketball history.

The previous single-season low was .597, set in 1962. But that percentage was based on only ~700 conference games, and in context it looks like a fluke. (Two years later, home teams surged to an all-time high of .668.) Last year’s nadir of .590, in contrast, came in more than 3000 games — and it was the continuation of a consistent downward trend.

When measuring home-court advantage in points per game* (which is more granular than binary wins and losses, and thus a bit more stable), last year is an even clearer outlier. Home teams outscored conference opponents by just 2.8 ppg in 2017, an all-time low by 0.3ppg. Across both metrics, the trend is roughly flat (maybe a slight decline) throughout the late 20th century, followed by a structural decline that started around 2000 and continues today.

*Ideally we would use points per possession, but not enough data is available historically. Instead, I’ve adjusted for era here (and in all following charts) by indexing each season’s total points per game to 2017 levels. The era adjustment has very little impact outside extreme scoring environments of the 1950s and early 1990s.

Why is home-court advantage in decline? Here’s what the data says about a few hypotheses:

Refereeing seems to be the primary channel. But why? In another analysis this summer, Pomeroy found that (a) home foul margin is a key predictor of home-court advantage, and (b) home foul margin is also declining. Prior research across many sports has pointed to referee behavior as the most important factor in home teams’ edge, so it is not surprising that whistles are behind college basketball’s recent trend.

But chalking the decline up to referees just begs the question. Why, then, are referees favoring home teams less?

Television does not seem to be a major factor. In an SI.com story on declining home-court advantage last season, then-Louisville coach Rick Pitino gave this explanation: “I think the refereeing has gotten much stronger because of television. Now, if a ref acts like a homer or makes a mistake, the TV guys will jump on them. That has made them conscious of not being homers.”

If that was true, home-court advantage should have declined earlier and more severely in the major conferences, whose games have been televised longer and on more prestigious channels. Instead, the decline has been steady throughout D-I, and power conferences actually have a larger home-court advantage than mid-majors. (Note: Major conferences include the Power 5 plus Big East, including predecessors; other multi-bid leagues based on three-year rolling average, recently including the American, Atlantic 10, Mountain West and WCC)

Declining crowd size is a plausible driver. Per-game attendance has been declining in college basketball since 2008 (from a peak of 5,300 to 4,600 last year). If crowds are less noisy — especially given the concurrent trend toward larger arenas — it could reduce unconscious bias in refereeing.

There are a few pieces of counter-evidence. First, the majority of the attendance decline (~60% of the post-2000 difference) is driven by new D-I teams — which are generally small mid-majors, skewing down the national average — rather than falling attendance for existing teams. This is inconsistent with the pattern shown above, where home-court advantage has declined just as much within power and multi-bid conferences as in the mid-majors. Second, home-court advantage is also declining in the NBA (albeit more gradually), despite record-setting attendance.

This is a ripe area for further study, such as: Is home-court advantage declining more among teams with falling attendance? Do new D-I teams have a smaller home-court advantage?

Technology enables better selection and oversight of referees. This idea was also advanced by Pitino, among others. I can see either side: On one hand, the whole theory of home-foul bias is driven by unconscious factors, which are not as easily corrected on review as, say, the technical details of a blown charge call. On the other hand, better knowledge of the issue — or selection of better referees in general — might reduce home-court favoritism. Perhaps the new officiating alliances will have a noticeable effect (especially in smaller leagues that are likely to benefit most); until then, I’m not sure how to test this claim.

Better travel is not helping much. Travel absolutely affects home-court advantage in general: Last year, the average home scoring margin between teams separated by less than 200 miles was just 2.0 ppg, compared to 3.1 ppg in all other games. (The average road trip has become significantly longer as a result of conference realignment, making the recent decline in home-court advantage even more remarkable.)

Technology has made travel more comfortable over time, and the influx of money in college athletics has upgraded some long bus trips to short flights, or commercial flights to charters. But those changes aren’t driving the decline in home-court advantage: The average margin for short road trips has fallen by just as much as the margin for long road trips. There was a significant drop in the impact of distance in the 1950s (with a small-sample caveat), and a much more gradual decline over following decades. But since the 90s, the gap between long road trips and short road trips has been quite stable.

Code behind the analysis and graphics is available here.