Despite what some disgruntled sports fans may have you believe, referees are human beings. Being as such, refs are susceptible to the same biases and predilections as everyone else. Although it is part of their job description and training to act as impartial judges, I personally do not feel it is ever possible for a person to abandon all of their natural tendencies.
Not only are refs (in my opinion) incapable of being truly unbiased, but there have been many cases in which they have been accused of purposely affecting the outcome of games for the benefit of themselves or others. The most notorious of these examples involves a ref by the name of Tim Donaghy, who was sentenced to 15 months in prison in 2007 for passing along insider information and betting on games which he officiated. According to sports gambling expert R.J Bell, games that Donaghy worked from 2003 to 2007 hit the “over” 57% of the time (which would seem to imply that Donaghy would consistently bet the “over” on games he was working). According to Bell, the odds against this discrepancy occurring naturally were roughly 1,000 to 1. For his part, Donaghy has come forward with a book alleging a number of improprieties on the part of his fellow NBA refs, including rigging games to keep them close, playing favorites, and having competitions to see who could wait longest before calling a foul.
Although I do not have the knowledge or access to uncover the next Tim Donaghy scandal, what I can do is look to see if there are any useful trends among today’s referees. What I would hope to find is refs that consistently call either significantly more or significantly less fouls than average, thus leading to a higher or lower final score, which would in turn help a bettor looking to make a wager on the over/under for a particular game. Here is what I found:
Results: The first thing I wanted to check in this analysis was whether the average number of fouls called by a particular ref had a significant impact on the average number of points scored in games that they were involved in. What I found was that during the 2014-2015 season, there was indeed a positive correlation between average number of fouls called and average points scored in a game (Pearson Correlation of .28), significant at the .03 level. This was rather unsurprising, as it is generally known that the more times you send teams to the foul line, the more points they will score.
Next, I wanted to see two things: If there were referees that called a particularly high or low number of fouls per game, and if there were referees that had a particularly high or low total number of points per game in contests they officiated. After performing this analysis, I was in fact able to find some outliers, which can be seen in the graphs below. For total points per game, I found a Mean of 100.07, a Standard Deviation of 1.21, and a range of 96.89 – 103.51. For total fouls per game, I found a Mean of 20.30, a Standard Deviation of .66, and a range of 18.91-21.63. As you can tell from these results, there were refs that were as much as two standard deviations either above or below the mean in both analyses.
Refs who called a significantly higher than average number of fouls:
- Zach Zarba (21.63)
- Bennett Salvatore (21.57)
Refs who called a significantly lower than average number of fouls:
- Eric Dalen (18.91)
- Haywoode Workman (19.13)
Refs who participated in games with a significantly higher than average scoring total:
- Scott Wall (103.51)
- James Capers (102.53)
Refs who participated in games with a significantly lower than average scoring total:
- Eric Dalen (96.885)
- Haywoode Workman (97.885)
Analysis: It seems that one could gain a slight edge in predicting Over/Under outcomes by taking referee biases into account. However, would this edge be enough to create a profitable strategy? Unfortunately, wading through the Over/Under outcomes of each game the above referees were involved in would be an extremely tedious process, but my guess would be no. Part of the problem here is that each game has 3 refs working the floor, and in many cases a lenient ref may be paired with a very strict ref in terms of foul-calling. I would think that knowledge about referee trends could be used as sort of an additional piece of info, rather than something to build a strategy upon. For example, if you already have a hunch that a game might hit the under, and then you happen to see that both Eric Dalen and Haywoode Workman are working that game, you could reasonably be more secure in your hunch. In the future, I would like to drill down this analysis further to see if refs have any glaring biases when things such as home-court advantage are taken into account.