Last week, I shared projections for player performance in the 2017-18 season. These projections weren’t perfect, but in general, they were useful. They expected young players like Joel Embiid, Andrew Wiggins, and Kristaps Porzingis to take steps forward while older players like Vince Carter and Jason Terry would take steps back. To get a more specific look at those projections, feel free to peruse the article I shared last week.
Now, I pivoted from player projections to team projections. To do this, I took team historical records from the past 4 seasons and combined them with standardized plus minus (the metric I introduced to perform my projections, which is a blend of RPM and VORP) in a linear regression to approximate the amount of wins a team would earn for their amount of team SPM. Using this regression in combination with my projections, I was able to arrive at an estimate of wins and losses for teams in the 2017-18 season. I made a couple of assumptions along the way, which I will go into further detail below.
The first assumption I had to made concerned rookies. My projections did not include rookies, as they did not play in the NBA last season, so I had to account for their performance in another way. Since I created projections and rankings for the draft and published a little over a month ago, I had the firepower to solve the problem. To do this, I took VORP from players in their first seasons and standardized it, and combined it with my standardized rankings. They are not calculated the exact same way as other NBA players, but they are on the same scale and reasonable in comparison to the rest of the NBA. For instance, most rookies are ineffective players in their first season, if not more, and most rookies had a negative rating for their Projected SPM. Only Lonzo Ball, Markelle Fultz, Dennis Smith Jr., and Josh Jackson had positive values, with Lonzo leading the pack at 0.32 PSPM (Projected Standardized Plus Minus).
My next assumption had to do with international draftees and signees. To do this, I calculated the value of a replacement player, assuming an NBA team composed of only replacement level players would go 10-72, and assumed that all of these players were at replacement level. It is not the best way to handle the issue, as I could create a model predicting the success of international players, but in general it is a good way to approximate performance without adding too much additional work. It leads to a few issues down the road, such as when predicting the success of the Boston Celtics. The Celtics have a host of international players coming over, of which a few are expected to make an impact greater than replacement level, but in general the assumption is one that needed to be made for the final projection calculation.
Lastly, not all teams had 15 players (the maximum amount), as some had more and others less. To account for this, I added replacement-level players to teams with less than 15 players, and took away replacement-level players from teams with more than 15 players.
Without further ado, here are my projections for the 2017-18 NBA season:
Much like what is expected, the East looks far weaker than the West.
A few teams draw attention, let’s start with the East. I mentioned the Celtics above. SPM does not think the Celtics are very deep. Although they have top players in Al Horford, Gordon Hayward, Jae Crowder, and Isaiah Thomas, there is a bit of a drop off from there. With that said, I think the main reason they may be a few wins less than expected has to do with the reason I depicted earlier, as Ante Zizic and Guerschon Yabusele are expected to be plus players, rather than replacement level.
Next I’ll call the Cavs into question. This ranking is with Kyrie Irving as a Cavalier, so admittedly it looks a little low. Diving deeper, SPM likes LeBron, Love, and Kyrie (3.395, 1.637, and 1.467 SPM, respectively), but beyond them, the Cavs look weak. Tristan Thompson, at 0.531 SPM, is there next best player projects to be Channing Frye, as my projections expect JR Smith to take a drop off due to injury history and age. Their free agency signings did not help, either, as their best acquisition was Derrick Rose (-0.381 projected SPM in ’17-18) and they became the next team to overrate the perennially overrated Jeff Green (-0.982 SPM in ’17-18). Additionally, my projections expect older players to regress (perhaps a bit more than they actually do), so their veteran core of Korver, Jefferson, and now Jose Calderon do not look too impressive (-0.803, -0.677, and -0.988 projected SPM, respectively), as if they ever did. Note that the Cavs’ problem is depth, which becomes significantly less important when the regular season ends, should they turn it on in the playoffs again.
Moving on to the West, the first two teams I will highlight briefly are the Clippers and Timberwolves. The Clippers dealt Chris Paul, but were able to acquire enough pieces to partially make up for his departure, in combination with signing Danilo Gallinari. The Wolves acquired Jimmy Butler, possibly the biggest acquisition in franchise history, as they acquired an All-NBA player for a solid price of Zach LaVine, Kris Dunn, and the pick that became Lauri Markkanen (Minnesota also got the Bulls’ 16th pick, turning it into Justin Patton). Although the rest of the Wolves’ off-season was questionable, this deal moved the needle so much to account for those mistakes and put the Wolves in serious playoff contention.
Now I’ll mention the Thunder. Despite getting Paul George and Patrick Patterson, the projected SPM expects them to miss the playoffs. I can point to a few possible reasons. First, PSPM expects Patterson to take a big step back next season, saying his past season was a bit of an outlier. I am not sure I agree with this, and it may be adversely affecting the model. Outliers are a given when projecting out this much, so that one might be on the model. The other reason, like the Cavs, is a lack of depth. Their projected starting 5 of Westbrook, Andre Roberson, George, Patterson, and Steven Adams is a great lineup (with a projected SPM sum of 7.294), but behind them they don’t really have anyone else. Additionally, SPM projects Semaj Christon and Doug McDermott to be some of the worst players in the NBA next season, and by replacing their efforts with replacement-level talent would put them in the playoffs.
I don’t expect the projections to be perfect, but they are a good reference point to start with.
excel file –> s_pm