In a previous post we looked at the accuracy of spreads for college basketball games. But one simple critique would be that yes, spreads may not be as accurate as first thought but that is because for a large part of the season Vegas isn’t working with much data. Preseason and Postseason polls are often quite different and teams under and over perform every season.
So does having more data improve the accuracy of spreads? This is relatively easy to determine. We just need to look across seasons and months to determine if the spreads in February and March are more accurate than those in November and December.
| Season | Nov | Dec | Jan | Feb | Mar | Season Median |
|---|---|---|---|---|---|---|
| 12-13 | 8.0 | 6.5 | 7.0 | 7.0 | 7.0 | 7.0 |
| 13-14 | 7.0 | 7.0 | 7.0 | 6.5 | 7.0 | 7.0 |
| 14-15 | 7.0 | 7.5 | 7.0 | 7.5 | 6.0 | 7.0 |
| 15-16 | 7.0 | 7.0 | 7.0 | 7.0 | 7.0 | 7.0 |
| 16-17 | 7.25 | 7.0 | 7.0 | 7.0 | 6.5 | 7.0 |
| 17-18 | 8.0 | 7.5 | 7.0 | – | – | – |
| 5-yr Median | 7.5 | 7.0 | 7.0 | 7.0 | 7.0 | 7.0 |
So is there a pattern? It does seem that November has the least accurate spreads, but that is only by a small margin of half a point. And while in some seasons March is the most accurate, in others is falls in line with the season median.
So does more data make Vegas more accurate? A little.
Gamblers Analytics 