Looking Forward to the Playoffs

ESPN has put up its Playoff Predictor, powered by Accuscore, which is fun to play around with, and John Hollinger has a post (Insider required) that mentions some of the points we can take away from it.  As it turns out, I also have a playoff prediction model (also linked in the banner, along with part 2 of the model).  How do our predictions line up?

First, a quick description of my model in case you didn’t check out the link (which you should! there are graphs and everything).  It uses home court advantage and the difference in regular season point differential to predict who wins a 7-game series.  So if the Celtics played the Pacers right now, which is the current 1-8 match-up in the East, we would plug in home court and a 6.5 – (-1.3) = 7.8 point advantage for the Celtics and give them a 96.6% chance of victory.  Accuscore is a little more conservative, giving the Celtics a 90% chance of winning the series; I don’t know what all they take into account, but I’m sure it’s more than I am.  But my model works pretty well: the multinomial version (used for predicting how many games the series will go) is the all-time leader of the TrueHoop stats smackdown.

Here are my conclusions about some of the points in John’s article.  In the East, the team to avoid in the first round is Philly.  They have the best point differential of the 5-8 seeds.  But, given the quality of the top 4, it isn’t a big worry; any team should still be about a 90% favorite to move on.  John mentions the Knicks, but they don’t seem any better with Carmelo than they were without, and so are basically an average team.  It’s possible that their willingness to bomb away from 3 gives them the variability that an underdog wants in the playoffs, but I don’t see them as a big threat to steal a series.

Similarly, the team to avoid in the West is Denver.  Despite being the 5 seed, they have a better differential than both Oklahoma City and Dallas.  If they have to go on the road their chances drop, but they would still be a tough out.  OKC would be a slight favorite in a coin flip if they maintain home court, but would drop to a 33% underdog if Denver passes them in the standings.

The Accuscore ratings apparently put home court advantage at about 5%; even teams on a neutral site would be 50-50 to win but if you give one home court it moves to 55-45.  My model puts it closer to 9%; 50-50 moves to 58.9% – 41.1%.  But, that advantage depends on the quality of the teams that are playing.  Because of the nature of probabilities, it’s harder to get closer to 100% if you start closer.  In playoff speak, a team that’s already a lot better than their opponent simply doesn’t gain much from home court.  Boston, as already mentioned, is already 7.8 points better than the Pacers.  If I take home court away, they still have a 93% chance of winning.  Home court in that match-up is only worth about 3 percent.  It’s important to note that this isn’t an interaction (my data show no interaction between team quality and home court), but simply an outcome of the qualities of probabilities.

The first round of games in the West is more likely to have upsets, but because the teams there are closer in quality.  The 5-8 seeds are better than the East’s 5-8, and Dallas and OKC aren’t as good as the Lakers or Spurs.  OKC in particular is vulnerable, and the Mavs may have a tough time in the second round.  They would not do well against the Lakers.

As is usually the case, the second round of the playoffs will be much closer.  However, this is really only true in the East.  If the top seeds all win, the Lakers are much better than the Mavs (even if they have to go on the road) and the Spurs are similarly better than either OKC or Denver.  In the East, however, Boston and Chicago are slightly better than the Magic and the Heat are pretty close to both the Bulls and Celtics.  The East will be close to a toss-up and home court will be much more important there.

The Spurs should end up with home court in the Finals if they get there, and will be a favorite against any team except for the Heat (which would be as close to a coin flip as you can get).  Otherwise the Celtics, Bulls, Heat, Lakers, and Mavs are all pretty close to having home court, which all of them will need to differentiate themselves from their opponent.  The Mavs in particular will need it if they get to the Finals, since their record is better than their point differential suggests and thus may be a paper tiger.  The Heat are hurting themselves with the other problem, which is having a worse record than differential suggests.  By point differential they’re the best team in the league, but may have to play two or three series on the road to win the championship.

About these ads
This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.

2 Responses to Looking Forward to the Playoffs

  1. reservoirgod says:

    How does OKC look since the trade?

    • Alex says:

      OKC is 8-1 (one win in overtime) in the 9 games Perkins has played for them with an average point differential around 9, I think, which is pretty impressive. In the games after the trade without Perkins (if I picked out the right set of games) they were 6-3 (two wins in overtime) with a point differential around 4. So it seems like losing Green and Krstic didn’t hurt much, and gaining Perkins helped a lot. But they’re small samples, hard to be sure. Either way, the model doesn’t account for trades or injuries or the like, so it won’t know if the Thunder play better with Perkins or Denver without Carmelo. Could be a missing component of the model, but it seems to have done pretty well anyway.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s