How to Win (or Lose) at Playoff Basketball

facebooktwitterreddit
April 20, 2015; Oakland, CA, USA; Golden State Warriors guard Stephen Curry (30, right) dribbles the basketball against New Orleans Pelicans guard Norris Cole (30, left) during the fourth quarter in game two of the first round of the NBA Playoffs at Oracle Arena. The Warriors defeated the Pelicans 97-87. Mandatory Credit: Kyle Terada-USA TODAY Sports
April 20, 2015; Oakland, CA, USA; Golden State Warriors guard Stephen Curry (30, right) dribbles the basketball against New Orleans Pelicans guard Norris Cole (30, left) during the fourth quarter in game two of the first round of the NBA Playoffs at Oracle Arena. The Warriors defeated the Pelicans 97-87. Mandatory Credit: Kyle Terada-USA TODAY Sports /

*Note: I’ve previously written a similar article analyzing playoff performance. This piece is meant to be a renewal of that, with deeper analysis and mathematical methodologies performed on data set. Dropping three season worth of data and adding in the 2014-15 playoffs is an interesting exercise, and the reach of Nylon Calculus has grown significantly in a year, making this a worthwhile study again. 

The discussion surrounding playoff basketball– what tendencies, traits or factors increase/decrease expected performance compared to the regular season– is filled with unsubstantiated nonsense. Every year heading into April we hear the same discussions about whether three-point shooting teams can win, that defense and rebounding wins championships and that you need superstars to win, along with other similar conventional wisdom-based claims.

Now some of these cliches about playoff basketball are true, but some quite likely are not. I’ve looked over the past five seasons of regular season and playoff data, and in this article I’ve tried to answer the question; what statistical tendencies and markers indicate a team is likely to perform better in the playoffs than during the regular season? I chose five seasons to keep the teams and play modern so that it’s possible to think my analysis isn’t only a description but a possible indicator of future behaviour, while getting a reliable sample size of 80 post-season runs. In five seasons of data if something isn’t found or hinted it’s not likely to be consequential.

First I built a regression model to predict ‘expected playoff wins’. The model only needs to take into account two factors. First, how good a team was in the regular season in an absolute sense. And second, the level of competition in the conference around the team. From those ‘expected playoff wins’, young teams, as well as those based on defense or 3-point shooting can be separated and compared to expected results.

Schedule Adjusted Net Rating vs. Playoff Wins. Expected Playoff Wins Linear and Quadratic Regression. By: Mika Honkasalo
Schedule Adjusted Net Rating vs. Playoff Wins. Expected Playoff Wins Linear and Quadratic Regression. By: Mika Honkasalo /

On each of the charts shown below the X-axis represents wins added, increasing from left to right. The Y-axis shows a team’s ranking in a particular category, going up from the best to the worst.  Meaning an upwards slope would point to success in a statistical category being detrimental to playoff performance, and a downwards slope being beneficial.

Defense beats offense

Offensive Rating and Defensive Rating vs. Deviation from Expected Playoff Wins. By: Mika Honkasalo
Offensive Rating and Defensive Rating vs. Deviation from Expected Playoff Wins. By: Mika Honkasalo /

On average the better a team is on offense, the poorer they perform against expected results. And on defense, the result is the opposite. Note that this doesn’t mean that a team shouldn’t strive to be great on offense (you should clearly try to be as good as possible). If a team outscores opponents by an average of 10 points per 100 possessions, but 8 of those are credited to offense and only 2 points to the defense, that team hasn’t performed as well as their counterparts.

To make this clearer: Teams that were in the top 7 on defense and outside the top 15 in offense, won about 2 games (compared to expectations) more than did teams with the reverse profile of top 7 on offense and with a below average defense. Two playoff wins is a sizable amount. Teams which were better on offense than defense averaged 1.1 wins fewer than expected. Additionally, top-5 teams in defense won an extra 1.6 games compared to teams in the top-5 on offense.

On average, offensive rates and efficiency tends to drop a few percent in the playoffs compared to the regular season.

Rebounding is a push

Rebounding Percentage vs. Performance Against Expected Playoff Wins
Rebounding Percentage vs. Performance Against Expected Playoff Wins /

As with the offense, team rebounding percentage actually has a negative correlation with beating expectations. Good defensive rebounding has been a slight plus, while offensive rebounding a small minus. Teams in the top 5 in offensive rebounding percentage have performed 0.8 games below their expected win totals, and the 10 best offensive rebounding teams in the past five seasons have have done particularly poorly; 2.3 games below their expected win totals. This is the biggest departure from conventional basketball wisdom. Being a good rebounding team hasn’t turned out to be an advantage in the playoffs at all.

Execution is key

Effective FG% vs. Performance Against Expected Playoff Wins. By: Mika Honkasalo
Effective FG% vs. Performance Against Expected Playoff Wins. By: Mika Honkasalo /

This was by far my favorite finding. While offensive efficiency was a negative, effective field goal percentage turned out to be an indicator of overperformance.

There are a few potential explanations at work here. Even though free throw rates go up by nearly ten percent in the playoffs, free throw rate doesn’t indicate anything about playoff performance. It is slightly odd that overall offensive efficiency is a minus, but isolated shooting efficiency is a plus. I think this is perhaps capturing something about the ability to execute with precision. This might carry over to crunch time as well. 6 out of the 7 teams that most outperformed expectations were in the top 7 in effective field goal percentage.

Teamwork beats superstars

Assist% vs. Performance Against Expected Playoff Wins. By: Mika Honkasalo
Assist% vs. Performance Against Expected Playoff Wins. By: Mika Honkasalo /

Here’s the one place where this analysis significantly differs from my previous article and it’s something that can easily happen since there are so few “superstars” in the NBA, making this a highly volatile category. Teams that are very good in Assist% (the percentage of baskets that are assisted), have done fantastically well against expectations, and being in the top-5 was a plus of around 1.7 games. Having superstars– defined as the top players in win shares per 48 minutes with significant minutes played– seems to make no difference whatsoever. It doesn’t matter if you choose 7, 10 or 12 players.

Note that this doesn’t necessarily have to be the case when it comes to playing for and winning the championship, just that overall in the playoffs the added value of superstars has been negligible over the past 5 seasons. Overall, changes of approximately two games one way or another may not sound like much, but they do represent massive increases in probability to go a round further.

The role of pace is changing

Before last season, pace in the playoffs has been on average around 4 percent slower than in the regular season, and over 60 percent of playoff teams from 2006 to 2014 played at a below average pace. This changed last season and the average pace was 94.4, compared to 93.9 during the regular season[1. Ed: Worth noting that the apparent increase in pace could be attributed to some degree to the increase in “Hack-A” strategies which have the effect of increasing the number of possessions, e.g. “Pace” while slowing the game to a crawl in the colloquial sense.]. Among the 13 teams that have ranked in the top 5 in pace during the regular season, only three have outperformed expectations, the Warriors in 2012-13, and last years’ Rockets and Warriors.