The Most Competitive Playoff Series Ever
Posted by Neil Paine on March 1, 2010
When the book closed on last year's Boston-Chicago first-round matchup, it was already hailed as one of the great playoff series of all time, as 4 of the 7 games went into overtime (including one game in double-OT and one in triple-OT!) and 5 were decided by 3 points or fewer. But while there's no doubt Celtics-Bulls 2009 was an instant classic and a great thrill ride, was it the most competitive playoff series ever? In fact, how would we measure "series competitiveness" anyway?
In the absence of play-by-play data (or even box scores) for the vast majority of basketball history, we're going to have to rely on final scores in this process. If we did have complete PBP records, we could actually calculate team Win Probability for each possession and identify series where neither team's WP was above a certain level before the final seconds in any game... But alas, basketball record-keeping hasn't been as diligent as that of its cousin, baseball, and we're left with formulating a metric from what the score was at the final buzzer, which any basketball fan knows can sometimes fail to tell the story of how close a game actually was throughout. But if we can accept this flaw, then we can begin to judge playoff series by the final margins of their games. My first stab at a "competitiveness index" would be to simply average the winner's MOV for each game of the series, and sort from smallest to largest:
| Year | TeamA | W | TeamB | W | avgMOV |
|---|---|---|---|---|---|
| 1989 | NYK | 3 | PHI | 0 | 2.7 |
| 1976 | DET | 1 | MIL | 2 | 3.0 |
| 1957 | MNL | 2 | FTW | 0 | 3.0 |
| 1985 | BOS | 3 | CLE | 1 | 3.5 |
| 1950 | FTW | 2 | ROC | 0 | 3.5 |
| 1981 | HOU | 2 | LAL | 1 | 4.0 |
| 2000 | NYK | 3 | TOR | 0 | 4.0 |
| 2000 | NYK | 3 | MIA | 4 | 4.0 |
| 1961 | SYR | 3 | PHW | 0 | 4.0 |
| 1975 | GSW | 4 | WSB | 0 | 4.0 |
| 1980 | SEA | 4 | MIL | 3 | 4.0 |
| 1984 | BOS | 3 | WSB | 1 | 4.3 |
| 1955 | SYR | 4 | FTW | 3 | 4.3 |
| 1956 | PHW | 4 | FTW | 1 | 4.4 |
| 2002 | SAC | 3 | UTA | 1 | 4.5 |
| 1988 | DET | 4 | BOS | 2 | 4.7 |
| 1965 | LAL | 4 | BAL | 2 | 4.7 |
| 1993 | SAS | 3 | POR | 1 | 4.8 |
| 2006 | CLE | 4 | WAS | 2 | 4.8 |
| 1954 | MNL | 2 | FTW | 0 | 5.0 |
| 1973 | NYK | 1 | LAL | 4 | 5.0 |
| 1980 | LAL | 1 | SEA | 4 | 5.0 |
| 1955 | MNL | 2 | ROC | 1 | 5.0 |
| 1981 | KCK | 2 | POR | 1 | 5.0 |
| 1976 | CLE | 3 | WSB | 4 | 5.1 |
| 1989 | CHI | 3 | CLE | 2 | 5.2 |
| 1987 | MIL | 3 | PHI | 2 | 5.2 |
| 1986 | DAL | 3 | UTA | 1 | 5.3 |
| 1999 | IND | 4 | PHI | 0 | 5.3 |
| 1997 | HOU | 4 | SEA | 3 | 5.3 |
Notice how the top of the list is dominated by short series, though, including sweep at #1! I don't know about you, but I don't think a sweep can be the most competitive series of all-time, so let's add a requirement that the series must have gone the full amount of games:
| Year | TeamA | W | TeamB | W | avgMOV |
|---|---|---|---|---|---|
| 1976 | DET | 1 | MIL | 2 | 3.0 |
| 1981 | HOU | 2 | LAL | 1 | 4.0 |
| 2000 | NYK | 3 | MIA | 4 | 4.0 |
| 1980 | SEA | 4 | MIL | 3 | 4.0 |
| 1955 | SYR | 4 | FTW | 3 | 4.3 |
| 1955 | MNL | 2 | ROC | 1 | 5.0 |
| 1981 | KCK | 2 | POR | 1 | 5.0 |
| 1976 | CLE | 3 | WSB | 4 | 5.1 |
| 1989 | CHI | 3 | CLE | 2 | 5.2 |
| 1987 | MIL | 3 | PHI | 2 | 5.2 |
| 1997 | HOU | 4 | SEA | 3 | 5.3 |
| 1981 | BOS | 3 | PHI | 4 | 5.3 |
| 1952 | NYK | 1 | BOS | 2 | 5.3 |
| 1984 | NYK | 3 | DET | 2 | 5.4 |
| 1984 | DAL | 3 | SEA | 2 | 5.4 |
| 2002 | LAL | 4 | SAC | 3 | 5.4 |
| 1961 | STL | 3 | LAL | 4 | 5.4 |
| 1969 | BOS | 3 | LAL | 4 | 5.6 |
| 1974 | CHI | 3 | DET | 4 | 5.7 |
| 1979 | PHO | 2 | POR | 1 | 6.0 |
| 1950 | INO | 2 | SHE | 1 | 6.0 |
| 1987 | BOS | 4 | MIL | 3 | 6.0 |
| 1979 | LAL | 1 | DEN | 2 | 6.0 |
| 2009 | BOS | 3 | CHI | 4 | 6.0 |
| 1994 | HOU | 4 | NYK | 3 | 6.1 |
| 1994 | UTA | 4 | DEN | 3 | 6.1 |
| 1998 | CHI | 4 | IND | 3 | 6.1 |
| 1984 | PHO | 3 | POR | 2 | 6.2 |
| 2006 | DAL | 3 | SAS | 4 | 6.3 |
| 1958 | PHW | 1 | SYR | 2 | 6.3 |
That's better, since now we're capturing series that came down to one do-or-die game... But we're still getting a lot of short (read: early-round) series in there, meaning the stakes -- while high -- weren't as pressure-packed as a tight series late in the playoffs. Also, by taking the average absolute MOV, it's possible we aren't punishing a series enough for having even one lopsided game; after all, we're looking for series that were ultra-competitive throughout. So to combat these factors, let's only consider 5- or 7-game series, and instead of looking at the average absolute MOV per game, let's look at the squared MOV per game.
Making these final adjustments, here are the 10 most competitive playoff series in NBA history by the MOV²/G metric:
10. Milwaukee 3, Philadelphia 2 (1987) - 44.8 MOV²/G
| Game 1: MIL 107 vs. PHI 104 (4/24/1987) |
| Game 2: PHI 125 @ MIL 122 (OT) (4/26/1987) |
| Game 3: MIL 121 @ PHI 120 (4/29/1987) |
| Game 4: PHI 124 vs. MIL 118 (5/1/1987) |
| Game 5: MIL 102 vs. PHI 89 (5/3/1987) |
9. Cleveland 4, Washington 3 (1976) - 44.6 MOV²/G
| Game 1: WSB 100 @ CLE 95 (4/13/1976) |
| Game 2: CLE 80 @ WSB 79 (4/15/1976) |
| Game 3: CLE 88 vs. WSB 76 (4/17/1976) |
| Game 4: WSB 109 vs. CLE 98 (4/21/1976) |
| Game 5: CLE 92 vs. WSB 91 (4/22/1976) |
| Game 6: WSB 102 vs. CLE 98 (OT) (4/26/1976) |
| Game 7: CLE 87 vs. WSB 85 (4/29/1976) |
8. L.A. Lakers 4, Sacramento 3 (2002) - 44.0 MOV²/G
| Game 1: LAL 106 @ SAC 99 (5/18/2002) |
| Game 2: SAC 96 vs. LAL 90 (5/20/2002) |
| Game 3: SAC 103 @ LAL 90 (5/24/2002) |
| Game 4: LAL 100 vs. SAC 99 (5/26/2002) |
| Game 5: SAC 92 vs. LAL 91 (5/28/2002) |
| Game 6: LAL 106 vs. SAC 102 (5/31/2002) |
| Game 7: LAL 112 @ SAC 106 (OT) (6/2/2002) |
7. Houston 4, New York 3 (1994) - 42.7 MOV²/G
| Game 1: HOU 85 vs. NYK 78 (6/8/1994) |
| Game 2: NYK 91 @ HOU 83 (6/10/1994) |
| Game 3: HOU 93 @ NYK 89 (6/12/1994) |
| Game 4: NYK 91 vs. HOU 82 (6/15/1994) |
| Game 5: NYK 91 vs. HOU 84 (6/17/1994) |
| Game 6: HOU 86 vs. NYK 84 (6/19/1994) |
| Game 7: HOU 90 vs. NYK 84 (6/22/1994) |
6. New York 3, Detroit 2 (1984) - 35.8 MOV²/G
| Game 1: NYK 94 @ DET 93 (4/17/1984) |
| Game 2: DET 113 vs. NYK 105 (4/19/1984) |
| Game 3: NYK 120 vs. DET 113 (4/22/1984) |
| Game 4: DET 119 @ NYK 112 (4/25/1984) |
| Game 5: NYK 127 @ DET 123 (OT) (4/27/1984) |
5. Chicago 3, Cleveland 2 (1989) - 34.4 MOV²/G
| Game 1: CHI 95 @ CLE 88 (4/28/1989) |
| Game 2: CLE 96 vs. CHI 88 (4/30/1989) |
| Game 3: CHI 101 vs. CLE 94 (5/3/1989) |
| Game 4: CLE 108 @ CHI 105 (OT) (5/5/1989) |
| Game 5: CHI 101 @ CLE 100 (5/7/1989) |
4. Houston 4, Seattle 3 (1997) - 32.4 MOV²/G
| Game 1: HOU 112 vs. SEA 102 (5/5/1997) |
| Game 2: SEA 106 @ HOU 101 (5/7/1997) |
| Game 3: HOU 97 @ SEA 93 (5/9/1997) |
| Game 4: HOU 110 @ SEA 106 (OT) (5/11/1997) |
| Game 5: SEA 100 @ HOU 94 (5/13/1997) |
| Game 6: SEA 99 vs. HOU 96 (5/15/1997) |
| Game 7: HOU 96 vs. SEA 91 (5/17/1997) |
3. Seattle 4, Milwaukee 3 (1980) - 26.3 MOV²/G
| Game 1: SEA 114 vs. MIL 113 (OT) (4/8/1980) |
| Game 2: MIL 114 @ SEA 112 (OT) (4/9/1980) |
| Game 3: MIL 95 vs. SEA 91 (4/11/1980) |
| Game 4: SEA 112 @ MIL 107 (4/13/1980) |
| Game 5: MIL 108 @ SEA 97 (4/15/1980) |
| Game 6: SEA 86 @ MIL 85 (4/18/1980) |
| Game 7: SEA 98 vs. MIL 94 (4/20/1980) |
2. Syracuse 4, Fort Wayne 3 (1955) - 22.6 MOV²/G
| Game 1: SYR 86 vs. FTW 82 (3/31/1955) |
| Game 2: SYR 87 vs. FTW 84 (4/2/1955) |
| Game 3: FTW 96 vs. SYR 89 (4/3/1955) |
| Game 4: FTW 109 vs. SYR 102 (4/5/1955) |
| Game 5: FTW 74 vs. SYR 71 (4/7/1955) |
| Game 6: SYR 109 vs. FTW 104 (4/9/1955) |
| Game 7: SYR 92 vs. FTW 91 (4/10/1955) |
1. New York 4, Miami 3 (2000) - 22.6 MOV²/G
| Game 1: MIA 87 vs. NYK 83 (5/7/2000) |
| Game 2: NYK 82 @ MIA 76 (5/9/2000) |
| Game 3: MIA 77 @ NYK 76 (OT) (5/12/2000) |
| Game 4: NYK 91 vs. MIA 83 (5/14/2000) |
| Game 5: MIA 87 vs. NYK 81 (5/17/2000) |
| Game 6: NYK 72 vs. MIA 70 (5/19/2000) |
| Game 7: NYK 83 @ MIA 82 (5/21/2000) |
March 1st, 2010 at 7:44 pm
Nice stuff. I wish there were good ways to really get the feel of some of these series—I wasn't exactly watching Syracuse vs. Fort Wayne several decades before I was born.
It would be really nice to be able to better reward overtime games—they've certainly got to be one of our best indicators of a competitive game, even stronger than MOV (since the score difference at the end of regulation was actually zero, making the MOV essentially zero before including the overtime). But in the absence of data providing that info, I guess we take what we can get.
March 1st, 2010 at 7:44 pm
This could be refined by using data for the scores at the end of each quarter, and not just the end of the game. This will do a better job of suggesting whether the game was close throughout, or simply as a result of a run by one of the teams later in the game.
Also, how is it that Boston-Chicago doesn't make your top ten? Is it really the case that the 21-point blowout in game 3 was enough that the series doesn't make the list?
March 1st, 2010 at 7:51 pm
Nice work.
I'm wondering, though, whether overtime games/periods should be given a bonus. Much of the thrill of Celtics-Bulls came from knowing, in two games, 48 minutes wasn't enough - and neither was 53.
March 1st, 2010 at 8:16 pm
One way I could do it would be to calculate MOV/MOV^2 per 48 minutes instead of per game -- this would lower the avg of a series that went into a lot of OT, particularly the Boston-Chicago series, which had one game go 63 minutes. Squaring the MOV does hurt that series, though, because a blowout is punished pretty hard. It's kind of a philosophical question there, whether a tight series should be allowed one game with a huge margin if the other games are extremely close...
March 1st, 2010 at 9:24 pm
That 94 Finals was so close, and so heart breaking. I never really recovered.
March 1st, 2010 at 9:33 pm
Neither did John Starks.
March 1st, 2010 at 10:11 pm
Here's the list if you go by avg. squared MOV per 48 minutes:
That 21-point blowout in Game 3 really get punished in BOS-CHI.
March 1st, 2010 at 11:02 pm
It's pretty rough, but what about just subtracting the number of OT periods from the final margin of each game? I'm no expert, but even the negative numbers that could result don't bother me. If they bother smarter people, you could perhaps subtract a point or two times the number of OTs in that game divided by the most OTs we are dealing with in this sample, to scale the bonus.
March 1st, 2010 at 11:25 pm
I'd just count an OT game as a 0 difference.
March 1st, 2010 at 11:53 pm
Perhaps you could increase the weight for each game as the series progresses, meaning that early blowouts aren't as costly as late blowouts.
March 2nd, 2010 at 12:12 am
As Downpuppy said... give an OT game a score of 0. I would argue that a 4 point overtime win is more competitive than a 3 point regulation win. Your current method (even the per minute version) would have the 3 point win in regulation as "more competitive".
March 2nd, 2010 at 5:11 am
Also whats more competitive a game ending by a point in regulation or a game ending by 5 points in 3 overtime??
I think overtimes have to be considerer in the equation somehow
March 3rd, 2010 at 12:22 am
"Perhaps you could increase the weight for each game as the series progresses, meaning that early blowouts aren't as costly as late blowouts."
FTW.
March 5th, 2010 at 11:55 am
I agree--Overtimes should be a 0
March 6th, 2010 at 10:16 pm
Well why not do the victory probability trick for the playoff series that have taken place since such data has been available? I'd be curious to see which are the most competitive playoff series of the last, say, 10 yrs using that system!