Archive for the 'Data Dump' Category

With a 2-0 lead in their conference semifinal series, this year's Suns team is currently looking like a good bet to defeat their hated archrivals, the San Antonio Spurs, and make the Western Conference Finals. The Suns have never advanced past the Conference Finals in the Nash era, and this year's likely WCF opponents have been a tough nut for Phoenix to crack, but it looks like the 2010 season is shaping up to be as good an opportunity as any in recent memory for Phoenix to make the Finals.

If you had told me a few years ago that we'd be saying this in May of 2010, after Mike D'Antoni departed and Steve Kerr shifted the franchise's direction with the acquisition of Shaquille O'Neal for Shawn Marion, I would have said you were crazy. Even now, this Suns team doesn't "feel" as great as the team that went deep in the 2005/06/07 playoffs, but here they are, with a chance to do something that the earlier incarnations of the Nash Suns never did. So how do these Suns stack up to Phoenix's past playoff teams?

Read the rest of this entry »

As a quick follow-up to my post this morning about the best elimination-game performances by total points and John Hollinger's "Game Score", I took commenter Mike G's advice and calculated single-game Win Shares for each player in an elimination game as well. Here are the Top 100 performances by WS since 1991 in a playoff game where the player's team was one loss away from going home for the summer:

Read the rest of this entry »

Playoffs Home ▪ 2010 Playoff Previews

On the eve of the 2010 Playoffs, I thought it'd be cool to run down the current list of winningest NBA "people" -- players and coaches. We know which coaches coached which games, so we can give them accurate credit for W-L records, but if you recall for players, we have to estimate the team's W-L record for games in which they appeared. Fortunately, this is pretty easy and a pretty decent kludge: take the team's winning percentage in all games (in this case, all playoff games in a season) and multiply by the player's games played for wins, then subtract that from his games for losses. This isn't perfect, but in the absence of pre-1991 playoff gamelogs, it's the best we can do (and it's pretty darned accurate for such a simple solution). With the explanation out of the way, here are the winningest playoff "people" of all time:

Read the rest of this entry »

Today is the last day of the 2010 regular season -- and in honor of that, here's a list of each franchise's all-time W-L records on the official final day of the NBA regular season:

Read the rest of this entry »

This is basically a big data dump, but I thought I'd share the thought process & methodology involved. The premise is to find players whose teams won more postseason games than they were "supposed to" based on their regular season winning %'s and those of their playoff opponents (if this sounds familiar, you may have seen this Doug Drinen post, which Google found for me about halfway through the creation of this post)...

Read the rest of this entry »

This is another stab at something I've worked on for years -- the idea is to take the performance for each team in the Four Factors, and find teams with similar profiles in the past to determine how teams of that ilk eventually do in the playoffs. Here's the methodology:

1. Calculate the Four Factors for each team on offense and defense (technically, I guess that would make it Eight Factors, but whatever).
2. Calculate the Z-score for each team's factors by subtracting the league average and dividing by the league's standard deviation.
3. Compute the difference between the two teams' z-scores and square it... Do this for all 8 factors.
4. Add the squared differences together, weighted by the following: Offensive & defensive eFG% --> 0.2 each; offensive & defensive TOV% --> 0.125 each; offensive & defensive ORB% --> 0.1 each; offensive & defensive FT rate --> 0.075 each.

Read the rest of this entry »

Many NBA fans like to measure players by team accomplishments like wins and championships, perhaps because that is every team's ultimate goal before each season, and to a certain extent all players on teams that didn't win the championship have failed. With this in mind, I decided to compile a list of the greatest winners in NBA history (since 1952, at least) by taking the minute-weighted average of winning % and SRS for the teams they were on. Since we don't have game logs for players prior to 1986-87, this will serve as a reasonable proxy for what Dean Oliver calls "game by game winning percentages", or the team's W-L% when a player played for them. We'll break things up into 2 divisions -- small samples (those with at least 2,000 career MP but fewer than 10,000) and large samples (those with at least 10,000 career MP) -- and look at the best and worst players by career WPct and SRS. First, the top winning percentages:

Read the rest of this entry »

Regular-season play resumes tonight after 5 days off for the All-Star break, so I thought it would be interesting to look at every NBA season and see how the team with the best record at the break fared the rest of the year:

Read the rest of this entry »

Continuing our series on huge playoff upsets, I've taken the methodology I laid out on Friday and applied it to every playoff series from 1991-2009 to determine the probability of each team winning, given the distribution of minutes for their players in the series. This is accomplished by finding weighted averages of the team's players' and opponents' seasonal SRS-SPM scores (see Part I for an explanation), and plugging them into the following equation to produce a single-game expected winning %:

xWP = 1 / (1 + exp(0.622 - 0.168(tm_srs) + 0.168(opp_srs) - 1.244(homecourt)))

Again, where homecourt = 1 if the team is at home and 0 if they're on the road. Armed with these single-game probabilities, all that's left is to use them to calculate the odds of winning a series of a given length with a given # of home games. This of course means you need to calculate not only the probability of each series outcome, but also the probability that the series ends in each specific # of games (for instance, winning a series in four games would require 4 consecutive wins -- WWWW -- while you could win in five games four different ways -- WWWLW, WWLWW, WLWWW, or LWWWW).

Read the rest of this entry »

Back in the spring, I used Statistical Plus/Minus to predict individual playoff series outcomes (and to pretty decent effect, too -- I mean, yes, it did miss on Cleveland-Orlando, but then again so did just about everyone, including the bigwigs at Nike). Anyway, today I'm going to use a similar method to look back on every playoff game from 1991-2009 and see which games ended in the most unlikely outcomes.

Read the rest of this entry »