Does the NBA preseason matter?

(revised to include 2006-07 season data)

Sure the preseason games matter to the players trying to make the team, the veterans trying to test out their bodies coming from injuries, and for the league itself perhaps to have some build up for the regular season tipoff. The question though is do the preseason records matter?

Is a team's won-lost mark in the exhibition games any indication or tip-off as to how the regular season will unfold? You hear announcers hand down their opinions, but I've yet to see any study on the matter...so let's examine the issue 82games style!

Tracking down preseason stats is no easy feat, but I've compiled a five year database for the 2001-02 season through 2006-07. That's not a huge sample, but a decent look for our purposes.

Correlations:
.35 Preseason win% --> Regular Season win%
.57 Last Year win% --> This year Regular Season win%

Now the confidence intervals on these are large due to an n=177, but still it's fair to say that a .35 correlation is reasonably healthy and suggests that yes the preseason records do have some significance in setting a tone for how things may go when the real games begin. The NBA does have more consistency from year to year than many sports, the NFL correlation from one year's record to the next for example was just .25 in one recent five year span.

BEST preseason records ('01-02 to '06-07)

Year	Team	Preseason		Regular Season
2002-03	Detroit	8-0	1.000	50-32	0.610
2004-05	Phoenix	7-1	0.875	62-20	0.756
2005-06	Denver	7-1	0.875	44-38	0.537
2003-04	Utah	7-1	0.875	42-40	0.512
2003-04	Memphis	6-1	0.857	50-32	0.610
2004-05	Denver	6-1	0.857	49-33	0.598
2002-03	New Jersey	6-1	0.857	49-33	0.598
2006-07	Toronto	6-1	0.857	47-35	0.573
2001-02	Minnesota	5-1	0.833	50-32	0.610
2001-02	Toronto	5-1	0.833	42-40	0.512
2006-07	Golden State	5-1	0.833	42-40	0.512
2005-06	Dallas	6-2	0.750	60-22	0.732
2002-03	Sacramento	6-2	0.750	59-23	0.720
2005-06	Phoenix	6-2	0.750	54-28	0.659
2006-07	Chicago	6-2	0.750	49-33	0.598
2005-06	Memphis	6-2	0.750	49-33	0.598
2005-06	LA Clippers	6-2	0.750	47-35	0.573
2004-05	Memphis	6-2	0.750	45-37	0.549
2005-06	LA Lakers	6-2	0.750	45-37	0.549
2004-05	Philadelphia	6-2	0.750	43-39	0.524
2004-05	Cleveland	6-2	0.750	42-40	0.512
2006-07	Orlando	6-2	0.750	40-42	0.488
2003-04	Seattle	6-2	0.750	37-45	0.451
2005-06	Houston	6-2	0.750	34-48	0.415
Average				47-35	0.573

Average gain of 4 wins from coming season compared to last

WORST preseason records

Year	Team	Preseason		Regular Season
2004-05	Seattle	2-6	0.250	52-30	0.634
2004-05	Sacramento	2-6	0.250	50-32	0.610
2002-03	NO/Oklahoma City	2-6	0.250	47-35	0.573
2004-05	Boston	2-6	0.250	45-37	0.549
2003-04	New York	2-6	0.250	39-43	0.476
2001-02	Washington	2-6	0.250	37-45	0.451
2006-07	Charlotte	2-6	0.250	33-49	0.402
2003-04	Philadelphia	2-6	0.250	33-49	0.402
2006-07	Portland	2-6	0.250	32-50	0.390
2003-04	Phoenix	2-6	0.250	29-53	0.354
2002-03	LA Clippers	2-6	0.250	27-55	0.329
2001-02	Denver	2-6	0.250	27-55	0.329
2003-04	Washington	2-6	0.250	25-57	0.305
2005-06	Portland	2-6	0.250	21-61	0.256
2005-06	San Antonio	2-7	0.222	63-19	0.768
2001-02	Cleveland	1-5	0.167	29-53	0.354
2006-07	Cleveland	1-6	0.143	50-32	0.610
2001-02	Philadelphia	1-6	0.143	43-39	0.524
2004-05	Atlanta	1-6	0.143	13-69	0.159
2003-04	Orlando	1-7	0.125	21-61	0.256
2002-03	Denver	1-7	0.125	17-65	0.207
Average				35-47	0.427

Average decline of 3 wins from coming season compared to last

So that's some pretty nice visual confirmation for preseason believers. Of course it might cross your mind that preseason records could vary in significance based on the prior year success of a team. After all a great team the previous year probably lets the mainstay guys ease through the exhibition games with minimal minutes while the team looking to improve might be running the regular season starters out there for big minutes.

Correlation of Preseason Record to Regular Season Record:
.66 Teams coming off a less than 30 win season
.42 Teams coming off a 30 to 39 win season
.42 Teams coming off a 40 to 49 win season
-.06 Teams coming off a 50+ win season

Aha! You say. That proves it then, great teams can ignore the preseason. Well, not entirely, since the sample sizes again get too small to really have conclusive value. Still it is suggestive of that theory, and more importantly perhaps it seems that for a poor prior year team, the preseason results really do have some bearing on whether improvement can be expected.

Teams with winning preseason records off a <30 win season

Year	Team	Preseason		Regular Season		Prior Yr
2004-05	Phoenix	7-1	0.875	62-20	0.756	29-53
2003-04	Memphis	6-1	0.857	50-32	0.610	28-54
2006-07	Toronto	6-1	0.857	47-35	0.573	27-55
2003-04	Miami	5-2	0.714	42-40	0.512	25-57
2006-07	New York	4-2	0.667	33-49	0.402	23-59
2003-04	Toronto	4-2	0.667	33-49	0.402	24-58
2003-04	Denver	5-3	0.625	43-39	0.524	17-65
2006-07	Atlanta	5-3	0.625	30-52	0.366	26-56
2002-03	Houston	4-3	0.571	43-39	0.524	28-54
2004-05	Orlando	4-3	0.571	36-46	0.439	21-61
2003-04	Cleveland	4-3	0.571	35-47	0.427	17-65

To put it in perspective, the rare poor prior year team that strings together some wins in the preseason has seen an average gain of 17 wins in the next year!

All right, so which lousy 06-07 teams have shown signs of life in 07-08 "meaningless" games?

7-1 Atlanta (30-52 last yr)
4-2 Memphis (22-60)
4-3 Boston (24-58)

(Indiana went 6-2 in preseason, after a 35-47 regular season)

Are we destined to see some revivals this coming season? Trying to project performance from the preseason alone is likely a foolish endeavor, but on the other hand if you're a fan of a team that's anything less than a rock solid 50+ win type, you might want to root for some exhibition wins after all.

And if you're an Atlanta fan? Believe that the Hawks are the second coming of Phoenix circa 2004-05!