Tuesday, April 24, 2007

NBA Playoffs

I'm still busy putting the book together ... Weighing the Odds in Sports Betting.

Here is a segment that will be in the book regarding NBA Playoffs. I figured it was topical, and I haven't posted in the blog in a long time.

Zigzag Theory
The simplest form of the zigzag theory in the NBA playoffs is: The team that lost a game is likely to cover the spread in the next game. For example, if the Lakers lost to the Suns in game 1 of a playoff series, the zigzag theory suggests betting on the Lakers to cover in game 2.

From a database of NBA playoff games from 2000 to 2006 (other results in this chapter uses the same set of data), the zigzag theory’s record was 231-211 for a winning percentage of 52.3%, basically break-even for -110 lines.

A refinement to the zigzag theory is to add in the qualification that the losing team did not cover the point spread. With that refinement, the zigzag theory’s record was 190-163 for a winning percentage of 53.8%. Data miners can get better winning percentages with more refinements, but testing them runs into the problem of small sample sizes.

I do not recommend you bet the zigzag theory based on this discussion alone. Line makers learn from history and sportsbooks adjust to action. Because the zigzag theory is well known, line makers probably have a good handle on its historical performance. Lines in the NBA playoffs may already have the zigzag factor embedded in them. Thus blindly betting the zigzag theory may be negative EV.

The next two sections present historical against-the-spread (ATS) records and line changes from game to game, which are more important. Bigger line changes than in the past suggest that the zigzag theory is embedded in the line.

Must-Win Games
Must-win games are games where a loss results in the end of the series or falling behind so much that it is nearly impossible to come back. The lines in these games often lean towards the team that needs the win.

Game 3, home team down 0-2
In 54 games, the home team has a straight-up record of 34-20 and an ATS record of 34-19-1. The average line in these game 3s was the home team laying 0.65 points.

In game 4 of those same series after the home team won game 3, the average line shifted to the home team getting 0.21 points. That game-to-game line shift of 0.86 points indicates that the betting market is aware of the must-win situation in game 3 for home teams down 0-2.

Game 4, home team down 0-3
This is a must-win situation. If the home team loses, it is out of the playoffs. But instead of playing harder, as generally happens in must-win situations, the 0-3 home team seems to play with less intensity. Perhaps the players think that being down 0-3 is too large of a hole to dig out of. No NBA playoff team has ever come back from a 0-3 deficit to win the series.

The database shows 15 games where the home team was down 0-3. In those 15 games, the average line was home team +3.47, up from an average line of +0.67 in the game 3. (The average over all game 3s with a 0-2 home team was home team -0.65; +0.67 applies to only the subset of those game in which the home team lost game 3.) That is a whopping game-to-game line shift of 2.8 points! The market expects a team down 0-3 to give less than full effort. Even with the extra points, the home team has had a tough time covering; its game-4 ATS record was 5-9-1. That sample size is too small for us to draw firm conclusions.

Game 6, home team down 2-3
When the home team is down 2 games to 3 going into game 6, the average line has the home team laying 1.8 points. In game 4 when the same team is at home (without counting the NBA Finals when game 4 and 6 have different home teams), the average spread is home team by 1.2 points. Thus the home team is favored by 0.6 extra points in game 6 when down 2-3 in the series. The must-win situation is probably the reason for this difference.

Game 7
Game 7 is the deciding game of the series; it is a must-win game for both teams. Both games 5 and 7 are played in the same venue, so it is interesting to compare the difference in the point spreads in those two games. Only 19 game 7s were played in the NBA from 2000 to 2006 (without counting the NBA Finals when game 5 and 7 have different home teams), but even with this small sample size, the line movement is interesting.

Averaging the lines in all game 7s, the home team was a 4.5 point favorite. In the average game 5 in every series that had a game 7, the home team was a 5.3 point favorite. The home team was a smaller favorite in game 7 by 0.8 points. In those 19 instances of game 7, the home team had an ATS record of 11-7-1. Perhaps the market erred in making the home team less of a favorite in game 7, but the sample size is too small to say for sure. The average line shift from game 5 to game 7 may be useful in calculating exact series lines.

No comments: