Do bookmakers bias the NBA Total Points markets?

As we all know, the market is continuously being verified by bettors. Top online bookmakers see this, and usually adjust the odds accordingly to the market flows, in order to be on the safe side and limit their losses. We are interested in something else, though. Do bookies actually bias the lines in either way?

According to some historical data taken from the NBA markets, we can see that on average, 50% of teams which were handicap favourites did cover the spread. Going a little forward we can also say, that looking at handicap favourites and underdogs there has been almost no difference. One research, on the other hand, found that punters tend to over-bet the favourites on the point spreads.

So the question is still the same if bookmakers can see that bettors are leaning towards one team or another, why wouldn’t they try to increase their returns by biasing the market.

The not equal distribution between Over and Under bets can be observed throughout many markets. It is unclear if that corresponds to the volume of bets, but it might suggest that bookmakers have a profit incentive to opt-in for biasing.

To really understand this betting phenomenon, we have to go beyond betting. One of the reasons for punters to like Overs more is loss aversion, the tendency of people strongly avoiding losses in comparison to acquiring gains.

Let’s take a look at a bigger picture. By itself Under bets are starting off as winning until the score goes over a certain line. Completely opposite could be said about the Over bets, by default, they start off as losing, until there would be enough points. 

For bettors who are sensitive to the loss aversion, this might be an issue. They would unintentionally tend to pick a loser that transforms into a winner rather than a winner, which transforms into a loser. 

Another interesting aspect that could be taken into consideration is psychology. There is an ongoing argument among psychologists that losing hurts twice as much as winning satisfies. Bouncing of this, we can assume that if that is really a case, then there should be expected twice as many bets on Overs than on Unders.

According to the data from the above-mentioned research, a decent-size sample of NBA matches showed a 64%/36% shares of Overs and Unders, in favour of the first. Let’s assume that if that numbers would’ve corresponded to the amount of money wagered, then the market biasing of just 1% (51%/49%) would increase bookmakers’ returns from 2.5% to a 3.0% profit margin, increasing the biasing to just 2% will result in a 3.0% to 3.6% increase.

Let’s go back to the historical NBA data, but this time focus on the Total Points market. Out of the more than 15.5 thousand NBA games, around 7.5 thousand (48.7%) finished with a final score being over the offered Total Points line, around 7.7 thousand (50.0%) finished below the line, and just 1.3% of games hit the exactly offered Total Points amount.

Assuming that there were no games that hit the exact score, we would end up with a 49.4%/50.6% split.

Putting this data through the chi-square test, we would get us a p-value of 0.11 or 11%. Such a result tells us that 11% of the time, we would get the already mentioned results by chance. It is a little too big of an assumption to make since we are looking for a p-value of at least 0.05, and preferably of 0.01 or even 0.001. That would allow us to completely rule out the possibility of results being affected by anything besides the pure chance.

But don’t rush yet. The above calculation cannot guarantee that bookies are not biasing their markets. To be 100% sure of the statement we have to get our hands on a bigger sample, and if the same percentage persists over it, we are likely to get a statistically significant enough p-value to confirm the biasing phenomenon.

Looking at the percentage of winning Overs and Unders we can see that winning Overs with the above-average lines (203) are less frequent compared to those games with below-average ones. The 7,530 NBA games sample with closing lines over 203 had a 48.8% to 51.2% win/loss ration. Take out the 205 matches that exactly hit the points line, and you will receive a roughly 50/50 deviation, which will give us a statistically weak significance (0.04 p-value). Just for comparison reasons, the games with below 203 Total Points line had a win/loss ratio of 49.9%/50.1%.

Again, it is hard to make a statement, that higher Total Points lines suggest a greater share of biasing from bookmakers against Overs. However, the highest lines above 225 showed win/loss ration in favour of Overs, which was above 50%. This, though, might not be enough to be used to support any of the theories due to small sample size.

One of the flaws of the research is that the initial data came from multiple sources. This severely affects the results as it might vary how bookmakers approach the lines. If that is the case, then analysing such a data sample will give us a misleading idea about the situation.

Generally speaking, there are lots of suspicions about the NBA Total Points markets being biased by the bookies. Lots of things point out on the books’ manipulations of some of their betting lines that affect inexperienced bettors. Unfortunately, this research cannot confirm or rule out such a scenario. Anyways, even if those biases exist, it is hardly enough for a regular bettor to cover the bookmakers’ margin. Despite being rumoured to exist, they are almost irrelevant for the overall practice, proving that NBA Total Points markets are fairly efficient.

NBA betting guide: advanced types of bets