clock menu more-arrow no yes mobile

Filed under:

Statsketball: The Early Western Conference Playoff Picture

With most team's relative playoff seeding starting to work itself out, now seemed like a good time to look at how the playoff picture in the Western Conference looks so far.

Jerome Miron-USA TODAY Sports

This has been a weird year when it comes to reconciling how our preseason expectations have jived with reality. It turns out, we were really wrong about a lot of things.

We expected that this was the year that the East would finally be good; or, at least, pretty tolerable, with 5 teams (Heat, Pacers, Bulls, Knicks, Nets) in solid deep-playoff contention and 4 more teams (Hawks, Wizards, Cavs, Pistons) that would be respectable.

As it turned out, this is the year that the East has finally lit itself on fire and leapt off of the very top of Minas Tirith to its doom, as Rose and Horford got hurt -- dooming the Bulls and Hawks to mediocrity or worse -- and the Knicks, Nets, Wiz, Cavs, and Pistons are all just really, really godawful.

The Blazers, Wolves, Mavericks, Nuggets, and Pelicans were all supposed to be fighting for the last 2 playoff spots, while Memphis was supposed to be comfortably situated with the Rockets, Warriors, Spurs, Thunder, and Clippers as an elite deep-playoff team. For God's sake, we thought that Phoenix would be even worse than Philly (who must be almost cursing how good Michael Carter-Williams is at this point).

In reality, Memphis was bad even before everyone got hurt, and then everyone got hurt. The Blazers are possible title contenders, Denver and the Pelicans are both deep disappointments, Phoenix is lighting the world on fire, and really only the Wolves and Mavs are competing for that last playoff spot.

The point of all this is that now we finally have a good look at what the year's teams really look like. We no longer have to rely on speculation of what these teams should or will look like at this point, but we can say with more certainty what these teams do look like. This is especially important this year, since we were apparently so very wrong about how a lot of these teams were going to work out.

As such, I thought that now might be a good time to talk about early playoff odds in the Western Conference, given that we now have a better, more legitimate, look at what the future might bring.

Playoff Picture So Far

I've seen some really wacky stuff from John Hollinger's playoff odds calculator in the last few days: recently, Dallas and Minnesota were both in the playoffs, and Houston was projected to be out. As of right now, Dallas is projected to have the 7th seed, over Houston in the 8th. This seemed a little unbelievable to me.

Here's Hollinger's projections, as they stood as of the writing of this article:


For what it's worth, this seems, in general, pretty reasonable to me -- with the exception of Houston. The Rockets have the same record as Golden State, and they've been missing a lot of really important players (Harden, Beverly, Lin) for long stretches of time. They've been really good this season for a lot longer than they've been even "meh." I have no idea how anyone can justify them being below Phoenix, LAC, or Dallas. Similarly, LAC after CP3's injury seems a little high, but I could be wrong about that.

It's also worth pointing out that I did have another agenda in breaking down the Playoff Odds so far. For one, after a 9 hour road trip in which I flew through about 7.5 hours of basketball podcasts, I got incredibly frustrated at the degree to which people were simply assuming that the Minnesota Timberwolves are going to make the playoffs over the Mavericks, and in 90% of these playoff discussions people didn't even acknowledge that the Mavs existed. People were simply saying things like, "when the Wolves get to the playoffs..."

It's not that I don't think that the Timberwolves have a legit shot at the playoffs either: they've beaten the Mavs 2 out of 3 times, and their January is quite easy. Nonetheless, the Wolves remain one win under .500, and the Mavs are 5 wins over, so it's hard seeing why, exactly, it's being taken for granted the Wolves are a lock for that spot. They could easily snag it, but I'm not sure why people are taking them as locks.

The frustrating part, to me, was that I couldn't think of a good way to prove that the Wolves and Mavs are simply in as close contention for the playoffs as anyone else, no more and no less. How could I, mathematically, calculate playoff odds while still accounting for schedule?

What follows is my methods and results in setting up my own Western Conference Playoff Projection.

My Playoff Picture

To casually project how many wins a team will have at the end of the season, I (and others) like to extrapolate a team's current win percentage out for 82 total games. Meaning: right now, the Mavericks are 19-14, so they've won 19 out of their 33 games, or 57.58%. To see how many games the Mavs will win in the simplest possible way, a lot of people would say, "well, 57.58% of 82 total games is 47, so the Mavs are likely to win 47 games."

There's a problem with this, though: a team's win% (the percentage of their total games that they've won) changes after every game. And, generally, a graph of a team's Win% over time looks like this:


The problem is, when you extrapolate from a team's current Win%, you're not taking into account that a team is probably in the middle of a streak, putting them either above or below their average number of Wins per Games Played; or in one of the hills or troughs above or below the red line on the graph above.

A lot of playoff projections and power rankings, by definition, suffer from a form of recency bias, where they overvalue current trends without accounting for the fact that those trends will, in all likelihood, level themselves out in the long run.

So, instead of going off of current Win %, for the Clippers, Suns, Rockets, Warriors, Mavericks, Timberwolves, and Pelicans, I went through and averaged all of their Win Percentages from each game until now.

In order to not bias early games (because until the 15th game, there are still small sample sizes in the percentages, and changes occur often and wildly in a way that would seriously mess with the data) I averaged the first 15 games, and used that average only 3 times in the main calculation (I'm happy to answer questions about this on email or on twitter at @HalBrownNBA).

I then added each Win % from each game after the first 15 games, and then averaged all of those, to get an average Win Percentage for each team.

The thing is, though, that I can't just use the Average Win Percentage over 82 games, either, because schedule matters, and -- as I pointed out before -- some teams have easier schedules going forward than others.

So I used a model in which the rest of the NBA schedule is simulated 5,000 times, based on these Win %'s. This means that, on the whole, after the simulation has been repeated many, many, many times, teams with the better Win%'s were going to beat teams with worse Win%'s. Teams that were really, really close in Win % often split games with each other. In general, the West beat the East with the exception of Indiana and Miami; and Portland, OKC, and San Antonio beat any team that I had calculated average Win% for.

Also, for LAC, I assumed that the Clippers would be about New Orleans-level for the next 4 weeks without Chris Paul. So, for the next 4 weeks on the Clippers' schedule, I used Nawlin's Win% for the model, and then switched back to the Clippers' normal Win% after the 4 week mark. This is an intensely suspect and almost arbitrary way to project the Clipper's wins, so they're the big wild-card here, depending on how good they'll be without Chris Paul.**

That said, as you'll see farther on, as long as CP3 rehabs on schedule, the injury isn't quite bad enough to affect the playoff picture in any way more substantive than seeding in the 4-7 range. The odds that the team and injury is bad enough to pit them with the Mavericks and Timberwolves fighting for the last seed is really low. Same goes for the Bledsoe injury, which is (barring more information) short enough to not warrant a change in methodology.

Here's the results of those simulations; the "best record" and "worst record" were calculated using the standard error.


As well, here's a graph of each team's Win% as it related to the number of Wins that the model projected. It's a good way to visualize each team's strength of schedule going forward, as well as how the teams all relate to each other.


Notes Regarding the Playoff Projections (and there are many)

  • These projections are far from perfect. The NBA is a very strange place, and every single one of those rankings could be horribly, horribly wrong at year's end. The problem with predictive statistics like these is that behavior in all respects is, generally, not calculable the way that past trends are. Even if, over 5,000 repetitions of the season from this point on, my chart will be pretty accurate, anything from injuries to trades to teams figuring themselves out and just getting better or worse can throw everything off. After all, my math doesn't account for actual improvement or regression as a team. Maybe everyone figures out the Mavs and they slide. Who knows? That said, I would also argue that this method is going to be more accurate in the long run than most other projections. "More Accurate" still doesn't mean "right," though.
  • Note the very tiny 2 game difference between the Mavs' best and worst potential record and their primary projection. Just given how wacky the NBA is, and how subject to change the league is, the odds that the Mavs only finish within two games of that projection are probably low, but it tells you how little variance there's been in the Mavs' Win% this year. They've hung around the 59% mark with very little variation all season. So, say what you will, at least they're consistent. Or consistently inconsistent. Or inconsistently consistent. Or whatever.
  • On the completely other end of the spectrum, notice the massive difference in the projected wins for Minnesota -- 9 games better or worse than their main projected total. They have had a very, very large variance in their Win %'s all season, which tells you how incredibly streaky they've been. They could easily end up sub-500, but they could also be a 5th or 6th seed. This January, then, will be huge for them in determining how their playoff odds look.
  • As I mentioned earlier, the Clippers' record is going to be more variable than their "best" and "worst" records indicate. It's impossible to say how good they'll be without CP3, but you have to imagine that they can, at least, tread water for a few weeks.
  • Notice the massive dropoff for New Orleans' win total after Minnesota's. NOLA projects to be a .500 team, and that was without taking the newest Ryan Anderson injury into consideration. Prior, I might have been inclined to think that they'd finish on the optimistic side of their projection, but they were a .350 team without Anderson. They might still be able to make the playoffs, but they'd need to play their best basketball for the rest of the season while Dallas and Minnesota would both have to underperform. Frankly, I don't see this Nawlins team playing their best basketball without one of their most important players. Poor New Orleans, we had such high hopes for you.
  • How bad is the Eastern Conference that every single playoff team in the West projects to be a 50+ win team? A lot of these teams will probably play to the under, but holy crap, that's bad.
  • Not taking strength of schedule into account (a big caveat), and using average Win %, there is a slightly below 2/3 chance that the Mavs end the season with a better record than the Wolves. If the final playoff spot comes down to the two teams, then, it appears that the Mavs probably have a slight advantage in terms of making the playoffs -- even when accounting for schedule -- but it'll be really, really close.
  • To that last point, there is a roughly 50% chance that the Mavs end the season with 48 or more wins (not taking schedule into account). Similarly, there is a roughly 25% chance that the Wolves end with 48 wins or more (also not taking schedule into account).
** Post Script Note: since the writing of this column, it's been announced that Chris Paul will be out for closer to 6 weeks. This will affect the standings possibly in a way that affects the Mavericks, and makes the whole thing far more interesting. My sincerest condolences to Chris Paul, and I hope to all that is good in the world that the injuries stop.

Statsketball is a registered trademark