Sunday, 6 September 2020

NBA: Evolving Play-Offs and Game 7 Totals

After the two NBA Game 7s this week, here are some more details about how the Totals in a long series tend to play out.

First of all, some history. Since 2014, NBA play-off series are all Best-of-Seven, played in a 2-2-1-1-1 format, with the highest seed (better team based on regular season) having Home advantage in Games 1, 2, 5 and 7, although sometimes Game 5 and often Game 7 are not required,

Prior to 2014, the series were played in a 2-3-2 format, and the format has been tinkered with on several occasions, which means we have to be careful when comparing data across seasons, especially in Games 5 and 6.

The format was changed in 2003, when the First Round became a Best-of Seven series, reducing the chances of a higher seed being eliminated, (i
n 2007, the Golden State Warriors became the first team to defeat a number one seen in a best-of-7 First Round series), but also increasing the probability of a team sitting around for a few days waiting for their opponent who may be playing three games more.

In 2005 the two conferences (Eastern and Western) were realigned into three divisions each, with each division winner qualifying for a top-three seed regardless of their record.

This proved unpopular, and in 2007 the rules were changed again so that the division winners are now only guaranteed a top-four seed. If the top two teams in a conference are in the same division, this change means they won't meet until the conference Final, whereas before they could meet in the second round (the conference semi-finals).

Eight teams from each Conference make the play-offs, with the winner of the Eastern Conference playing the winner of the Western Conference for the NBA Championship.


So each season there are 15 play-off series, 7 between Eastern Conference teams, 7 between Western Conference teams plus the Finals Series. 

Teams need to win 16 games to be champions, and 71 of the 73 winners to date have been seeded either number 1 (49 wins), 2 (16), or 3 (6).

My suspicion is that there is more interest from the general public in play-off games, and that interest increases within each series peaking with Game 7s and also as the play-off rounds progress peaking with the Championship Series, and the public like to see points which may offer value on the Unders. 

Does the data support this idea?

Since 2002, when we first have data, Unders has been the result 51.5% of the time which is not a huge edge, but at the standard (Pinnacle) 1.952 (US -105) odds, you only need to win more than 51.23% of the time to be profitable.

In the 'If needed' games, i.e. Games 5, 6 and 7, the Unders record all-time is impressive, and notably more impressive if the game was preceded by one or two Unders also.

Unfortunately since the latest change in format in 2014, the trend is not so strong. One reason may be that we have fewer matches to look at, but also it seems reasonable to assume that markets are more efficient these days than they were in the 2000s.
There's also a noticeable divergence in results between Conferences since 2014. 

In the NHL play-offs, where 58.6% of Game 7s go Under, there was another one last night between the Philadelphia Flyers (nominally the Home team) and the New York Islanders. With four goals in total, this was another win for Unders.

With baseball not playing out this season anything like it typically does, systems that have performed well for many years are struggling. With different rules in place, it seems likely I should exclude these matches from analysis in future, but Signora Cassini is happy with her hometown San Diego Padres looking likely to end their thirteen year play-off drought with the third best record in the National League West, albeit playing under American League rules. Her father was a minor league player with the Angels (then California Angels) organisation, so he's not quite so happy. 

No comments: