Do Preseason Predictions Affect a Team’s Success Against the Spread? A Deep Dive

As I determined in two earlier posts (here and here), the amount by which an NBA team either overperforms or underperforms  against their preseason win total predictions has a strong affect on their success against the spread during the season. While this effect is more pronounced earlier in the year, at all points during the season teams that are outpacing their predicted win totals perform significantly better against the spread, and vice versa. As it applies to betting advice, the main conclusion I was able to draw from this was that gamblers should bet on teams that are on pace to beat their O/U wins total, and avoid teams that are expected to undershoot it.

Unfortunately, the above strategy has a lot of holes and also leaves out a lot of potential value. For example, what should you do if two teams that are blowing past their respective preseason win totals are playing each other? Or, should you bet on a team that is roughly on pace with their expected number of wins if they are playing a team that is way behind in this regard? In order to address these questions I had, I decided to look at my above hypothesis on a game-by-game basis.

Hypothesis: My hypothesis was that in looking at each game on such a granular level, I would discover that the difference between two teams’ performance against their win total expectations at the time of playing one another would help to predict which of them would beat the spread.

In order to determine this, I took every match up from the 2014-2015 NBA season and determined the amount both teams were either above or their below their expected wins pace. For example, say the Knicks and Bulls are slated to play each other. Prior to this meeting, the Knicks were predicted to have won 18.5 games at this point in the season, but have only won 14. Similarly, the Bulls were predicted to have won 28 games at this point in the season, but have managed to win 30. So, heading into this showdown, the Bulls have an expected wins pace of +2.0, and the Knicks have an expected wins pace of -4.5.

Next, I looked at the average difference in expected wins paces between ATS winners and ATS losers. In the above example, say the Bulls are the ATS winner in that game. This would mean that they are placed in the ATS winners group with an expected wins pace difference of +6.5 (the gap between their total of +2.0 and the Knicks total of -4.5), and the Knicks would be placed in the ATS losers group with an expected wins pace difference of -6.5.

Lastly, I looked to see if there was a correlation between a teams’ win pace difference and the amount that they either beat or fell short of the spread for a given game. Using the above example again, say the spread for that game was Knicks +6, and that they lost by a final score was 109-100. In that scenario, the Knicks were -3 against the spread and had an expected wins pace difference of -6.5.

Results: For ATS winners and losers, I found that the mean difference in expected wins paces was .578 for winners and -.578 for losers. This difference was found to be significant at the .001 level. As for the correlation, I found an r value of .128 when correlating a team’s expected wins pace value with the amount they beat/fall short of the spread for a given game. While an r value of .128 might seem weak, it was found to be significant at the .001 level. However, one has to take this finding with a grain of salt, since my sample size being so large (2460) certainly had an effect on the determined significance of this result.

Analysis: Although there was clearly a trend towards winning ATS teams have a better expected wins pace than their opponents, I wouldn’t say that this result is enough to go off of on its own. Like I’ve stated for other analyses, I think that this is a valuable piece of information nonetheless, and is definitely something that could be used as part of a more comprehensive model. In the future, it might be interesting to get even more granular and update this model with mid-season predictions.

Leave a Reply

Your email address will not be published. Required fields are marked *