Hoops Analysis: Home Court Advantage

The 2020-21 season of college basketball like no other.  The impact of COVID-19 has already caused a shortened preseason, multiple postponed games, and multi-week team shutdowns. Another obvious impact is the lack of fans in Big Ten arenas and in arenas across the nation.

There has been a lot of discussion about how much impact this lack of fans is having on the results of games. So far, MSU has only played three home games in Big Ten play, and the Spartans have lost two of those three games (versus Purdue and Wisconsin). It seems that the advantage of playing the Breslin Center may be decreased, but with a sample size of only three games, it is hard to say for sure. Is there a way to try to quantify this effect?

One piece of data to look at is the overall win percentage of all Big Ten home teams. As of February 4th. Big Ten teams are 44-28 overall, which is a win percentage of 61.1 percent.  Historically, this turns out to low side of normal in Big Ten play. 

Figure 1 below compares the final home team win percentage in Big Ten conference play from 2002 to the present. As the Figure shows, there is a fair amount of variation over the years. The overall average is 64.3 percent over a total of over 2,100 games, but the standard deviation is 4.3 percent.  The home teams have won as many as 72 percent of conference games (in 2002 and 2003) and as few as 57 percent (in 2014).

Figure 1: Home team win percentage in Big Ten conference games back to 2002

As for the rest of the country, the story seems to be about the same. In conference play across all of Division 1, home teams are 745-565 (57 percent) as of the morning of January 30th, when I last pulled this data.  In addition, there are a total of six conferences where the home teams actually have losing records. Those conferences are America East, Big Sky, Atlantic Sun, Atlantic-10, the Big 12, and the Patriot League.

The data from 2020 has the overall home win percentage at 61 percent. Furthermore, only one conference, America East had a home record of .500 or worse over the full season. Based on this data, it does suggest that home teams are wining at a slightly lower rate, but only by a few percentage points. So far the home court advantage for the 2020-21 season seems to be, once again, just on the low side of normal.

However, there are two things that trouble me slightly about this simple analysis above. First, I am comparing the full season worth of data in 2019-20 to only a partial set of data in 2020-21. It is not an apples-to-apples comparison. Second, I am basically assuming that the if there was no home court advantage, that we would expect the home teams to win 50 percent of all games.

In the cases where it is possible to play a full round robin schedule, this is likely to be the case. But, most conferences have some form as an unbalanced schedule, and this imbalance may cause the home teams or road teams to have a slight advantage overall. 

Fortunately, there is a was to check this effect fairly precisely using my favorite source of data: Kenpom efficiency margins. As I have discussed many times, efficiency margin data can be used to estimate point spreads and victory probabilities. But, baked into the calculation of these spreads and probabilities is an estimation of home court advantage for each team.

According to the Kenpom website, he calculates a home court advantage for each team "based on last 60 home and road conference games. Values are per game differences between home and road margin."  For Big Ten teams, these correction factors vary by team from a value of three to four points.  The current home court advantage for Breslin center is 3.7 points.

I do not think that Kenpom himself is making any specific correction based on the lack of fans during COVID-19. It is possible that there is some measurable effect, but the fact that the calculation takes into account the last 60 conference games would make the impact of the 2020-21 conferences games is small.  So far, I have not observed any change in the values on Kenpom.com for any Big Ten team.

Regardless, I calculate the point spreads, win probabilities, and expected number of wins based on the preseason home court advantage values.  Therefore, it is simple to calculate the number of expected wins by the home Big Ten teams using the normal (i.e. non-COVID) home court correction factor. In addition, it is also possible to calculate the expected number of home wins if the home court correction factor was zero. The results of this calculation are shown below:

• Actual home team record: 44-28 (60.0 percent)
• Expected home team record, with normal home court correction: 41.6 - 30.4 (57.8 percent)
• Expected home team record, with no home court correction: 37.3 - 34.7 (51.8 percent)

These results tell us a few things. First, it is true that a lack of a home court advantage does not automatically imply an even 50-50 split between home teams and road teams. But it is pretty close. Based on the games completed so far, the home teams should have a very narrow advantage (1.8 percentage points or about one to two game) over road teams.

Second, if the preseason, non-COVID adjustment home court advantage factors are correct, the home teams would be winning between 57 and 58 percent of the time. In reality, the home teams are actually wining slightly more often than that, by about a two and a half games overall.

In other word, the current Big Ten home court advantage overall is actually slightly higher than expected relative to the past several season (i.e. the last 60 conference games). This strike me as a surprising result.

I don't have the data off hand to run the same analysis for the rest of Division 1, so I cannot comment on whether this effect is also observed in the other conferences, and especially the high-major ones. As stated above, the current numbers overall still look like the "low side of normal."  But the data from the Big Ten is clear.  

The impact of having no fans in the stands does not seem to be impacting the overall home court advantage for the conference, on average. That said, the experience of each individual team may vary. But, if this statement is generally true, it may offer some hope for the Spartans. 

MSU will still play up to seven more games in the Breslin Center this year if the postponed games are all made up. If the Green and White can right ship enough to be able to defend Breslin, the odds for a strong finish to the season and the possibility of a post-season will start to look much, much more likely.