Behind the Schedule, Part II (07/29/16)
Last week, we took a closer look at the anatomy of the NBA schedule, dissecting the 2014-15 Atlanta Hawks' 60-win season. The Hawks took advantage of opportunities presented by the schedule, both structural and circumstantial, to take the top seed in the Eastern Conference. With the majority of the roster intact – promoting in-house project Kent Bazemore to fill DeMarre Carroll's starting spot, how did Atlanta drop 12 wins down to a 48-34 record and the number four seed in the East? This week, we cycle through various objective lenses with the Atlanta Hawks on the stage as we expand our exploration of the NBA schedule across seasons.
As we saw last week there are many ways to view an NBA schedule. The first lens we can use is inherent to the structure of the schedule – days of rest between games. If the NBA schedule was completely fair, each team would have the same number of total rest days and each matchup would feature teams coming off the same number of rest days. However, the realities of travel, broadcast obligations, and Taylor Swift concerts take fairness out of the scheduling vocabulary. Take the Hawks for example. Atlanta's total rest remained stable from 2015 to 2016, growing only by one day to 96 total days of rest. However, Hawks opponents rested for 101 total days in 2015, followed by only 93 days the following season. We need to zoom in to the game level in order to gain insights into how these rest day changes are reflected in the Hawks 12-win drop-off.
The table below uses the same methodology for last week to compare the Hawks record in the various scenarios of rest in each of the last two seasons. Starting with Atlanta's rest situations, the scenario with the largest decline in wins (two days' rest) reflects the dynamic nature of the schedule from year to year – they won seven fewer games, but there were seven fewer games available. Same goes for zero days of rest, where the two fewer wins may just be a casualty of two fewer games in that scenario. These fewer available games didn't just disappear, they were simply redistributed, primarily to one day of rest. Despite seven additional games, Atlanta won four fewer games (and lost 11 more) with one days' rest.
Decreasing available games in rest categories for the Hawks cost them proportional wins. The inverse of that pattern did not hold for increases in available games in opponents' rest day categories – zero added wins despite two more games with two days of opponent rest and four more games with zero days of opponent's rest. Untangling the effect of the schedule from these perspective is unclear. Thankfully we've got more lenses to explore.
Instead of separating the rest days between Atlanta and its opponents, we can combine them to see which team has the more days' rest coming into each game. The table below shows Atlanta's record in both years in each of the three advantage categories – equal rest, Atlanta has rest advantage, the Hawks' opponent has rest advantage. Here we can see the biggest disappointment from the Hawks perspective last season. In 2016, they were only able to convert their eight additional games with the rest advantage into one win. Analyzing the rest day breakdown for Atlanta, there is no glaring structural disadvantage in the schedule to explain 12 fewer wins – in fact, the schedule was arguably more in Atlanta's favor. We'll have to dig deeper to explain the Hawks demise.
Beyond simply days of rest, there are certain quirks in the schedule that are perhaps more impactful on performance. By quirks, I am referring to back-to-backs, three games in four nights, and four games in five nights. Since Commissioner Silver took over in 2014, he has made it a point of emphasis to reduce these burdens, with the 2016 schedule including major decreases in back-to-backs and four in fives. These developments play an important role in comparing schedules across seasons, as they amplify the regular variance.
Framing the schedule through these quirks, we can add another layer to how a schedule plays into wins and losses. Taking a look at the table below, we can see evidence of the Commissioner's agenda on the Hawks schedule – decreases in both back-to-backs and four in fives for Atlanta. Note that there is overlap between back-to-backs and the other two categories – a four in five is a tasty back-to-back sandwich (three in fours don't overlap with four in fives here). The largest changes to Atlanta's record came from the opponents' side. In the three games in four nights category, Atlanta faced only one fewer opponent, yet won seven fewer games, an alarming trend for a theoretically advantageous scenario. Overall though, these quirks echo the rest of the analysis – Atlanta faced no structural disadvantage and more exploration is necessary into the elusive 12-win fall.
Team by Team
Up to this point we've focused on the inherent structural impact of the schedule, which has been fairly inconclusive in explaining the change in wins for the Hawks. Perhaps the best way to uncover the mystery behind this 12-win deficit is to track the transfer of wins and losses to each of their opponents. Think of it this way: those 12 wins that didn't go to Atlanta last year had to go somewhere else. The table below shows the net gain/loss in wins for Atlanta versus each opponent from 2015 to 2016.
As we can see, Atlanta reclaimed nine wins from eight teams while giving back 21 wins to 14 teams for a net deficit of 12 wins. When reading this chart, there are several factors consider. Take the first line – BOS +1. In 2015, Atlanta went 2-1 versus Boston. In 2016, Atlanta went 3-1. The net gain of one win is in large part a structural result – the Hawks played an additional game against Boston, which they won. This points to an interesting part of the schedule worth explaining.
Being a member of the Eastern Conference, Atlanta plays each Western team twice (home and away) and each Eastern team three times. The Hawks play a fourth game against each member of its division (MIA, CHA, WAS, ORL). This is where the variable part of the schedule that changes each year begins. A fourth game is played against three of the teams in each of the other two Eastern Conference divisions – leaving four teams per year with only three games against Atlanta, or any team for that matter. In 2015, those teams were Boston, New York, Chicago, and Indiana. In 2016, they were Brooklyn, Philadelphia, Milwaukee, and Cleveland. In the opposite years, Atlanta gets a fourth game against these teams, which we can compare directly as a means of partially explaining the change in record. However, in this instance it is a wash – Atlanta went 3-1 in the fourth games in both years.
The concept of win transferring established above is fascinating. Building off last week's single season analysis and the structural effects explored in the previous sections in combination with the win transferring concept above is the key to understanding the NBA schedule beyond this specific Atlanta Hawks case. There are a number of exciting directions to go from here, which I want to establish as an appetizer for future work.
To get even deeper into the win transferring by opponent, we can compare each game in sequence versus each opponent. For example, in the above table, Atlanta is +1 versus Memphis. In the first meeting between the two teams in each season, Atlanta was victorious, a net zero on the +/- win total. Zooming in, in each season, both teams were coming off equal rest. Additionally, in 2016, both teams were playing a third game in four nights. The game in 2015 was in Atlanta, while the 2016 game was in Memphis. For Atlanta, 75% of the minutes played in the 2016 game came from players who played in the 2015 meeting, while only 59% of Grizzlies minutes returned from 2015 (this is a substitute measure for the “full strength” idea introduced last week).
In contrast, Memphis won the second meeting in 2015, but Atlanta won the second meeting in 2016 – reclaiming a victory for 2016, thus +1 versus Memphis. They were even in days' rest in 2015, but Atlanta had the advantage in 2016 – Memphis played on back to back days to cap off three games in four nights in the 2016 meeting. Similar to the first meeting, 70% of Atlanta's 2016 minutes came from those returning from 2015. For Memphis, they might as well have been a different team, with only 9% returning from 2015 – Marc Gasol, Mike Conley, and Zach Randolph were all out of action in the 2016 game.
There is a lot to unpack from this single example and even more to expand upon going forward. Stay tuned as we continue our journey through the intricacies of the NBA schedule next week.