NBA Disparity and Changing the Playoffs

The 2014-15 NBA Season saw the predictable continuation of a longstanding problem: the Eastern Conference is laughably weak compared to its cross-country counterpart. It seems year after year the Western Conference has an incredibly competitive playoff race which sees a good team miss the postseason whereas the East routinely sees teams .500 or worse advance due to the conference’s lack of overall quality and depth with the exception of a couple title contenders. Due to the seemingly permanent disparity between the two conferences, in recent years many have called for the NBA to eliminate its conference system altogether, instead opting for a system where the league’s top 16 teams make the playoffs, providing a fairer environment where teams don’t have an unfair advantage in making the postseason just because of their conference affiliation. With the proposition rapidly gaining support, even with NBA Commissioner Adam Silver claiming the league is considering doing away with conferences, a question must be answered: are basketball fans everywhere jumping the gun with this radical proposal, or is there substantial evidence supporting a change?

Since the main issue people seem to have with the current structure is how at the end of the season, the West’s 8-seed regularly holds a much greater record than the East’s, this investigation will begin by measuring just how significant this record difference is.

For starters, the last season where the Eastern Conference’s 8-seed won more games than the Western Conference’s was in the 1998-99 season. This means that since 1999-2000, there have been 16 consecutive seasons in which the West has had the superior 8-seed. And it isn’t just a one or two-game difference, either. Over this span, the West’s 8-seed has had on average 5.875 more wins than the East’s. There is also reason to believe the gap is expanding since over the last eight years the gap has been 8.25 wins per season. Since each team has played 640 regular season games during these eight seasons (seven 82-game schedules plus the 66-game lockout-shortened 2011-12 campaign), and the 8-seed has won 66 more games combined (8.25×8), it can be concluded that the final playoff team in the West has won over 10% more of their games than their Eastern complement, a number which clearly shows the Western Conference’s greater depth during this time span.

If the shortened 2011-12 season is discounted, during the other 15 seasons since 98-99, the East’s 8-seed has never finished with a better record than 42-40, whereas the West’s 8-seed never won less than 42 games, even hitting the 50-win mark twice. Since each conference has 15 teams, the 8-seed is the team in the middle of the conference, record-wise. Ideally, these teams should be about 41-41, sometimes a couple games higher or lower, but eventually averaging out to an equal number of wins and losses. However, the average win total for the West’s 8-seed over the 15-year period was 45.73, and the East’s only 39.53. But seriously, 50 wins for an 8-seed? That would’ve tied for third place in the East this past season! In nine of the 15 seasons, there was at least one East team which made the playoffs despite having a losing record, including the last three years, which all saw the east’s 8-seed finish with a 38-44 record. All in all, during the last 15 full seasons, 12 sub-.500 teams made the playoffs in the East, along with 13 more .500 teams, for a total of 25 teams without a winning record, or 1.67 per season, compared to zero in the West. Here’s another way to look at this, which really shows the startling difference in talent: during the last 15 82-game seasons, the East has averaged 6.33 teams with winning records, compared to 8.8 in the West.

The last two paragraphs provide a lot of interesting statistics, which all seem to suggest that the problem of conference disparity is no joke. However, when teams compete for playoff positions, they only compete against teams from their own conference. It really makes no difference how the conference’s 8-seeds stack up against each other, since they’re still good enough to make it in the playoffs. With the proposed alteration to a league-wide 16-team playoff, however, teams would be battling it out with opponents they didn’t have to worry about before, making teams who would’ve finished ninth, tenth, or even farther down in the West still eligible for the playoffs. So are Western Conference teams missing the playoffs when they have records good enough to make the postseason in the East? Take a guess.

Still using our last 15 full seasons, there have been 23 Western Conference teams which finished ninth or worse in their conference with a better record than the East’s 8-seed, or just over one and a half per year. There have been two each of the past three years, and at least one every year since the 2006-07 season. 12 of the 15 years saw at least one team miss the postseason that would’ve advanced in a 16-team format. The worst year for this was in 2003-04, when unbelievably teams seeded 9-12 were all better than the East’s #8. To view it differently, that year only two West teams weren’t playoff-worthy in the East (the following season Charlotte would become the league’s 30th franchise, meaning at this point the West had only 14 teams). When good teams are consistently losing spots they’ve earned to inferior squads lucky enough to be facing less talented competition, a change needs to be made.

It’s understandable for the NBA to be taking their time doing research analyzing all the pros and cons before making a final decision on a potential switch, and Adam Silver’s acknowledgment of the problem and commitment to finding a solution is to be applauded. However, why are so many fans opposed to something that seems to make a lot of sense? They must all be fans of Eastern Conference teams. But in all seriousness, the main reason cited in the majority of articles against the elimination of conferences is this: conference disparity is cyclical. If everyone would just wait a few seasons, the Eastern Conference will improve, and then eventually become better than the West. Then after a few years the pendulum will swing back towards Western Conference dominance, and so forth. The West hasn’t even been better for that long, these people claim. Are these points valid?

Well, so far the focus of this article has been primarily on the mediocre teams, the 8-seeds in each conference and the West’s first couple teams outside of the top 8. Significant as these teams may be, they tell far from the whole story. To really see which conference as a whole has been better, all that needs to be done is to compare the conference’s records against each other. Since each team in the league plays every team in the other conference twice, the conference that has won more of these games overall should be the better one.

Just as the 1998-99 season happened to be the last time the Eastern Conference had the superior 8-seed, it was also the last year the East had a better overall conference record, winning nearly 55% of inter-conference matchups. The scatterplot below shows the Western Conference’s winning percentage against the East every year since.

Here the x-axis shows the season and the y-axis shows the percentage of games between a West team and an East team where the West’s team came out victorious.

The line going through the middle of the graph is the line of best fit. Basically, this line goes through the middle of the data as best as possible, showing how the data trends over time. Notice how the line is getting slightly higher across the graph. If the system truly was cyclical, there would be signs of the West getting closer to the East over time, which would result in the line of best fit lowering. However, over time the West has won on average slightly more games per year against the East.

Now to look at the numbers themselves. As this scatterplot shows, from the 1999-2000 season to the present, the Western Conference won more games against the East in every year but one. Actually, in every season besides 2008-09, the West won at least 54% of inter-conference games. In all the seasons combined, the West won an average of 57.5% of their games against the East, which is an incredibly large difference.

If it doesn’t seem significant, just think of it this way. Every year since 2004-05 when the league moved up to 30 teams, there have been 450 games played each year between a West team and an East team. During these years specifically, the West has won just under 57% of their games against the East. If this percentage is multiplied by 450, it can be found that the West has won an average of 256.4 games against the East per year, compared to just 193.6 games won by the East against the West. This means that in an average year since 2004-05, the West has had nearly 63 more wins than the East. Divide that by 15, and a random team in the West in any of these years would be expected to have about four more wins than a random East team. This is where the problem lies. How can the NBA continue to have one conference so much stronger than the other on an annual basis? It unfairly forces the teams in the Western Conference to acquire several more wins than they would need to in order to make the playoffs in the Eastern Conference.

Now sure, out of the 450 games between the conferences, it would be unreasonable to expect a 50/50 season to occur, where each side takes 225 victories. However, if the conferences were even, the outcomes should be akin to flipping a coin. If you flip a coin 450 times, the odds that it comes out an even 225/225 are actually only about 3.8%. However, most of the time, the outcome is close to even. The odds of either conference winning at least the roughly 57% of games the West has on average since 2004-05 is just .29%, meaning it will happen once randomly per every 340 seasons or so. However, if the odds are taken of a conference winning that often over a course of the eleven seasons starting with 2004-05 (4950 games total), the odds are nearly literally zero. In fact, taken to ten decimal places, the odds are .0000000000%. This would probably have to be taken to many more decimal places before any non-zero number presented itself. That’s how unlikely this would be if the conferences were even. Safe to say, they aren’t.

By this point, it should seem pretty clear that there’s a problem, and that having just one conference would be a good solution. Now if there were only a good way to see what a season like this would look like, to give evidence of the positive effects of this change. Look no further. Since no such metric exists, I took it upon myself to create one: the Opponent Proportionality Record, or OPR for short.

What the OPR does is show what a team’s record would look like if they played each conference a proportional amount of times to the amount of opposing teams in each conference. By neutralizing any advantages or disadvantages of being aligned in a weak or strong conference, OPR gives a better indication of how good a team really is. Here is the formula for a team’s OPR, which is represented in terms of wins in a season.

Here’s how it works. A team’s win percentage against teams within its existing conference is found by dividing the amount of wins a team had against its own conference and dividing by 52, the number of games a team plays against its conference. Then this number is multiplied by (14/29), the proportion of the other teams in the league that come from a team’s own conference, which is also the proportion of games a team would play against these teams in a one-conference system. This makes up half of the formula. The other half is very similar, and involves taking a team’s non-conference winning percentage and dividing it by 30, the games played against teams outside their conference. This is multiplied by (15/29) the proportion of teams in the other conference, making the second half of this equation. Once parts one and two are added together, a team’s OPR winning percentage is reached, which can be multiplied by 82 to see how many wins the team would have in the season if they played every team equally.

Here’s an example. Let’s say a team went 50-32, with 30 wins coming in conference and 20 coming out of conference. If 30 and 20 are plugged into the equation, it looks like this: [(30/52) x (14/29) + (20/30) x (15/29)] x 82, which turns out to be about 51.1 wins. Since this team performed better in non-conference games, which there are fewer of than conference games, they are rewarded with an OPR slightly better than their actual record. A team’s OPR will very rarely change by more than a couple games, so a bottom-feeder isn’t going to magically become a playoff team. Additionally, if a team wins half their games in and out of conference, their OPR won’t be any different than their actual record. Again, what OPR does is eliminate the effect of conference on record.

So let’s get to the interesting part: how this effects actual NBA teams. To do this, an OPR will be found for each team from this past season. First, each team’s OPR will be compared to their actual record.

 

Upon first glance at this chart, it is pretty evident that while there are a couple exceptions on both sides, the Western Conference generally benefits from OPR while most Eastern Conference teams are damaged by it. While as I stated earlier there are no huge win total changes (the Lakers improve the most, at +2.8 wins), there are certainly some interesting twists upon closer analysis of the table. Last season, the Pelicans and Thunder finished tied for the West’s 8-seed, and the tiebreaker was won by the Pelicans. However, the Thunder receive the third-biggest OPR gain while the Pelicans drop slightly. The Thunder have a higher OPR by 2.6 wins, meaning they would’ve made the playoffs in a one-conference system. The Suns go from a sub-.500 team to a winning team, while the Bucks drop 2.6 games from their actual .500 mark. The Spurs improve to third in the West compared to their actual #6 seed. Lastly, at the bottom of the league, the Timberwolves go from worst to third-worst in the league, and the Knicks would’ve had the best odds in the lottery.

Note: The OPR formula may be further refined in the future by finding a team’s projected win total if they played every team in the league an equal amount of times. This is slightly different from the current formula because while a team will play all non-conference opponents twice within a season, they play 10 of the 14 remaining teams in their conference four times in a year, and the other four teams only three times. However, the current formula finds a team’s record if they played the conferences equally, slightly simplifying the same-conference half of the equation.

Since now we have the OPR for every team in the league, let’s look at how these records would influence the playoff picture in a one-conference system where the league’s top 16 teams made the playoffs. For this chart, teams in orange are from the Western Conference and teams in blue represent the East. A number following a team’s names shows what seed they were in last year’s playoffs.

The most obvious takeaway from this table is how much orange is featured on the left side, which would be the playoff seeding in a 16-team playoff seeded by OPR. This means that in a one-conference system, a staggering 11 West teams would have been expected to make the playoffs last season, compared to just five for the East. Additionally, six of the eight teams with home-court advantage would be from the West.

Perhaps the craziest result of this reordering is that the opening round’s 7-10 matchup would be between the Cavaliers and Bulls, the East’s 2 and 3 seeds in real life. The Thunder, previously not in the playoffs, are safely in the postseason with the #12 seed, and would be a part of a great series against the Clippers.

So let’s say the NBA decides to go with the one-conference system. How should they go about creating the schedule? Well, the season has 82 games, so first of all, a team should play every other team twice, one home and one away. This leaves 24 remaining games. Then a list should be randomly generated pairing each of the 30 teams in the league with another team, with this other team being labeled team 1. Then this will be repeated with the condition that two teams already paired with each other cannot be paired together again, and the next matchup for a team will be labeled team 2. This will continue until each team has every other team as an opponent labeled 1-29.

For the first season, a team will play a third game against each team numbered 1-24, with home games coming against odd numbered teams, and road games coming against even numbered teams. Then for the next season, each team’s number in the order will be increased by five, with teams 25-29 becoming 1-5 in the order. This creates a rolling list, so that teams get a third game against a team at home, away, or sometimes not at all. Over a 29-year span, a team will play a total of 12 extra home and away games against any team in addition to the standard one home and away matchup per year.

Now, this might sound like a pretty confusing way to create a new schedule. However, this method makes schedules as fair as possible, eliminating the random chance of getting unlucky and facing a championship-contender an extra time on the road for several consecutive seasons. Since this method may be hard to understand, here is a visual representation of what it would look like for two seasons. The Warriors will be used as our sample team. If an opponent is in red, the Warriors would have an additional home game against that team. If blue, there will be an additional road game, and if the cell isn’t colored in, the Warriors will only play that team twice during the season.

Clearly, under this formula, no team will ever play an extra home or away game against an opponent in consecutive seasons. If, for example, the Warriors play the Pacers an extra time at home one season, the next season they will play the Pacers an extra game on the road. If they go from an extra road game to not playing an extra game versus a team, the following season they will play an extra home game. It always alternates.

There is significant data to show that the NBA would be fairer if conferences were dropped altogether. However, this change would’ve been made already if there weren’t some issues associated with it. These issues- including travel, playoff scheduling, and the elimination of rivalries, will be discussed in part two of this series.