The Playoff Odds

April-May 1996

Up Front Results and Comments

Based on matchup probabilities, I calculate the following odds for the playoff series.
NBA Finals

Chicago over Seattle with 85-90% certainty
	After 6 G's:  Chicago wins 4-2, which was the most likely result.

Third Round

Chicago over Orlando with 91% certainty
	After 4 G's:  Chi beats Orl 4-0
Seattle over Utah with 85-87% certainty
	After 7 G's:  Sea beats Uta 4-3


Second Round

Chicago over New York with 99% certainty
	After 5 G's:  Chicago beats NY 4-1
Orlando over Atlanta with 92% certainty
	After 5 G:  Orlando beats Atlanta 4-1
Seattle over Houston with 93% certainty
	After 4 G's:  Seattle beats Hou 4-0
San Antonio over Utah with 66% certainty
	After 4 G's:  Utah beats SA 4-2


First Round

Chicago over Miami with 99% certainty
	After 3 G's:  Chicago wins 3-0
Cleveland over New York with 50.4% certainty
	After 3 G's:  New York wins 3-0
Indiana over Atlanta with 70% certainty
	After 5 G's:  Atlanta wins 3-2
Orlando over Detroit with 90% certainty
	After 3 G's:  Orlando wins 3-0

Seattle over Sacramento with 97% certainty
	After 4 G's:  Seattle wins 3-1
LA Lakers over Houston with 64% certainty
	After 4 G's:  Houston wins 3-1
Utah over Portland with 82% certainty
	After 5 G's:  Utah wins 3-2
San Antonio over Phoenix with 89% certainty
	After 4 G's:  San Antonio wins 3-1

Third Round Comments

Chicago played it right, not letting the three point shooting attack of the Magic get going. Shaquille is a good player, but the high variance players for the Magic were what was necessary to beat the Bulls and they just never got it going..... Fortunately, the Seattle-Utah series is competitive because its winner gets to be the next sacrificial lamb for the Bulls in what will likely be another lopsided series.

Second Round Comments

The San Antonio-Utah series really was closer than I'd expected. But the fact that Utah won the series justifies these numbers and reminds me why I never trust just one source of information.

I didn't believe the prior probability of Seattle beating Houston, but it did pretty well, if one can say these things about probabilistic events. Houston threw everyone's odds out the window last season, so I'm personally glad their gone this year.

The most likely result of the New York-Chicago matchup was a four game sweep (52%). The next most likely result was Chicago in five games (34%). Not bad. Ho-hum.

New York's sweep of Cleveland in the first round was not a major surprise. Prior to the series, such a result was to be expected 12% of the time.

The Article: Introduction

This will be an evolving work through the course of the playoffs. Its purpose is to use the matchup probabilities to estimate the probability that certain teams will win their playoff series. The matchup probabilities work by estimating the chance that team A will beat team B on team A's home court. The method uses team A's home court record, team B's road record, and the league home court advantage to estimate this according to

P(A beats B at A's Home) =

                (A's Hm Rec)(1-B's Rd Rec)(Lg Rd Rec)
-----------------------------------------------------------------------------
[(A's Hm Rec)(1-B's Rd Rec)(Lg Rd Rec)+(1-A's Hm Rec)(B's Rd Rec)(Lg Hm Rec)

A's Hm Rec	Team A's Record at Home
B's Rd Rec	Team B's Record on the Road
Lg Rd Rec	The League Average Winning % on the Road (0.396)
Lg Hm Rec	The League Average Winning % at Home (0.604)
This formula comes from Bill James' Baseball Abstract '87 and its theoretical development has apparently been published somewhere academic, though I honestly do not know where.

With the above formula and the schedule for each of the playoff series, you can evaluate each team's chances of winning their first round series. There is a bit of combinatorics involved, but it's not too bad. (For those who know statistics: note that I did not do any Bayesian updating of the game probabilities, neither for season records nor for playoff outcomes. There is a straightforward way to do it for the playoffs, but I'm not convinced that the straightforward way is actually right.)

The results of this method are shown below:

Second Round

Chicago over New York with 99% certainty
Seattle over Houston with 93% certainty


First Round

Chicago over Miami with 99% certainty
Cleveland over New York with 50.4% certainty
Indiana over Atlanta with 70% certainty
Orlando over Detroit with 90% certainty

Seattle over Sacramento with 97% certainty
LA Lakers over Houston with 64% certainty
Utah over Portland with 82% certainty
San Antonio over Phoenix with 89% certainty

It is important to remark that these estimates are really just base approximations based upon season records. They do not account for Reggie Miller missing from Indiana, Seattle's tendency to choke (Denver had a 1% chance two years ago against them), or Houston's high variance style increasing their odds as underdogs. Accounting for these missing pieces involves a considerable amount of work.

Finally, you can carry out these calculations throughout the playoffs and you find that the Chicago Bulls have at least a 75% chance of winning it all -- they are 1-3 favorites. That is pretty close to what the oddsmakers have been saying, so the method appears to be reasonable, despite its limiting assumptions. For those with more spare time than I have, they can figure out all the combinatorics involved and calculate the probabilities of winning the Championship for all the teams.