One and Done: A Thing of the Past? An Analysis of the NBA First-Round Playoff Format

By Will McMillan

In the middle of the 2002-2003 NBA season, the league made a decision to switch the first round of playoff series from a best-of-5 to best-of-7 format. This was a puzzling move at the time, with the most cited motive of David Stern and Co. being a decrease in volatility of these match-ups of high seeds versus low seeds. That is to say, it was expected that this change would protect the league’s best regular season teams from being bounced due to a few poor performances, or at least that it would lower the likelihood of these teams being punished for a bad shooting night or two after a long season of solid games.

Conspiracy theorists pointed to the early-season struggles of the 2002-2003 Lakers (they of the marquee players and gigantic marketing base) as key in the league’s pulling the trigger on this matter. Others saw it as just another way to pull in money and help the bottom line. In any case, the switch was made, and what has followed in terms of first-round results, and also further implications in the playoffs, is quite interesting.

This was a pretty basic study — because of the other rule changes for seeding, something like a matched-pair analysis wouldn’t have been worth it. I simply found the winning percentage of the higher-seeded teams in the 64 first-round match-ups since the change 8 playoffs ago, and compared it to the 64 most recent series from the best-of-5 era. I chose this method so that I would have equal sample sizes for each format, and to maximize the similarities of players, rosters, rules, etc. The results are:

Best-of-5 Format (1995-2002):

Higher-Seeded Team = 49-15                        Winning Percentage:            76.56%

Best-of-7 Format (2003-2010)

Higher-Seeded Team = 52-12                        Winning Percentage: 81.25%

It’s plain to see that there really isn’t much of a difference between the percentages of higher-seeded teams that advance out of the first round, with only 3 out of 64 more favorites moving on. A statistical hypothesis test showed that the results were nowhere near significantly different. It seems that the NBA’s change in series length has yet to have a measurable effect.

Absence of evidence is not evidence of absence — it could very well be that there has been an effect, but we just haven’t seen it. Also, none of this is to say that the impact of the switch has not been felt later on in the playoffs. Since 2003, the 12 “bracket-busters” have combined to win 16 games after the first round, for an average of 1.33 wins per team, and no team seeded 5 or lower has advanced past the second round.

However, from 1995 to 2002, the 15 teams seeded 5 or lower that won their first round match-up won 35 games after the first round, for an average of 2.33 wins. This group of underdogs contained the 1995 NBA champion Houston Rockets (a 6 seed) and 1999 Finals runners-up New York Knicks (an 8 seed).

This may not seem like a huge difference, especially because it only takes a couple longer playoff runs to significantly alter the balance, but there may be something to this trend that’s worth a later look. Perhaps teams that play longer series, as the underdogs likely did, are less likely to win in later rounds. There’s also the issue of realignment — because of the original rule that division champions had to be the first three seeds, some very good teams ended up seeded lower than they might otherwise have been. This might have counteracted the effect in the first round, but not have mattered in the second and beyond.

I will acknowledge that this is a small sample size, and doesn’t fully account for all rule changes (which would have made the samples even smaller). This is definitely something that should be revisited in 5 or 10 years to better gauge whether either of the things I have discussed is a real measurable change (or in the first case, a non-change). It may be that we have just seen an 8-year anomaly; maybe favorites will start winning their first round series with even greater frequency, and maybe some of the underdogs that do get through will make serious runs at titles.

But for now, it looks like the NBA has (perhaps tangentially) done a good job of making sure that the most successful regular season teams are the ones vying for the championship in June. Only time will tell whether or not Cinderella can stage a comeback.

About these ads
This entry was posted in NBA Basketball and tagged , . Bookmark the permalink.

4 Responses to One and Done: A Thing of the Past? An Analysis of the NBA First-Round Playoff Format

  1. David Pinto says:

    If the NBA wants the higher winning percentage teams playing, why not remove the first round of playoffs? In basketball, with a small sample of regular season games, there’s much less certainty as to which teams really are the best teams. A team with an intrinsic talent level of 50 wins could easily win 58 games, and vice versa. I’m actually surprised we don’t see more upsets in the first round.

  2. Enrique says:

    I think it makes more sense to analize how #1 seeds do or #1 and #2. Usually the #4 and #5 seeds are equally good, but that’s not usually the case with #1 seeds.

  3. Will McMillan says:

    David, I agree with you that it is strange that the first round seems so predictable, especially when the NBA espouses parity and contract fairness right down to the trades that are allowed to be made. Still, I’m not sure that this would justify getting rid of Round 1. I think a good way to look at this is by examining two extreme cases, college basketball and Major League Baseball. The former has a relatively short season, and so the NCAA uses a tournament with 6 rounds and 65 teams, since each has had less of a chance to distinguish themselves from their competition during the regular season. Meanwhile, MLB teams have 162 games to prove themselves, so only 8 out of 30 teams get a shot at the title. Going by this, I think that 16 out of 30 teams with 4 rounds of playoffs is the right amount for the NBA (and the NHL), even if the current trends are showing that half these teams have a pretty low chance of getting through the first round. This difference in percentage of playoff teams and the difficulty of getting to the championship once a team makes the playoffs, based on the length and “usefulness” of the regular season, seems to be pretty standard across the four major sports, so I guess this is just how the leagues feel they can maximize competition (and make a lot of money). Maybe this is another reason why the BCS seems so innately wrong.

    Enrique, you’re right that there isn’t really much of an appreciable difference between 4s and 5s, but I think it does make sense to still include them in the discussion because the 4 seeds still have the possibility of an extra home game, and get the chance to presumably go up 2-0 at home in the first two games. I think that home-court advantage is pretty important in the playoffs, if only because it’s one of the only things that can really separate a lot of these similar teams.

  4. dadler3 says:

    Will, enjoyed the post. I’m with Enrique; I’d love to see if there’s any difference for the top 3 seeds (i.e. remove the 4-5 matchup from the data). If you have that handy in your spreadsheet, I’d love to see it in another post. If the 4-5 matchups are driving most of the upsets in both samples, it won’t make a difference.

    One other not very well fleshed out thought…Home court advantage may have been a bigger deal under the five game format (higher seed had 50% more home games than the lower seed if it went five games vs. 33% more under the current system)…although I do like your “up 2-0″ hypothesis.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s