Pavlovic hasn't registered a Reb or an Ast in 63 Min; nor any FT.
Lebron is about 11 'units' (T rate) off his game; guarded by Prince, who is down about the same. _________________ 40% of all statistics are wrong.
Lebron's foul rate is unconscionably low. About half that of any other major player. His TO are also quite decent.
Cavs have the two best scorers? _________________ 40% of all statistics are wrong.
Historically the Piston's 2 greatest stepper-uppers, Billups and Prince disappeared offensively.
Gibson was 12-24 from the arc, 33-37 on FT. _________________ 40% of all statistics are wrong.
Seems like this is similiar to the 2006 NBA Finals. Lebron and good defense -- Dwayne Wade and good team defense beats the Mavs. Could this happen against the Spurs as well? Will Lebron dominate?
Joined: 14 Jan 2005 Posts: 1521 Location: Delphi, Indiana
Posted: Tue Jun 05, 2007 2:10 pm Post subject:
Wade scored (standardized) about 5 pts/36 better. Lebron Assisted about 5 better, and he rebounded another 2.2 . All else is about equal, and LeBron played an extra 2 mpg.
This year's RPR (replacement player rate) was about 14.5 in the 3rd round. In last year's Finals, it was 13.5. The lower RPR gives more eWins to lesser players; leaving a bit less for the stars.
Last year's rates are adjusted to essentially be equivalent to previous years. The difference is that players' unassisted-FG % makes some look better and others look worse. _________________ 40% of all statistics are wrong.
Joined: 31 Dec 2004 Posts: 881 Location: Durham, NC
Posted: Tue Jun 05, 2007 8:29 pm Post subject:
Mike G wrote:
This year's RPR (replacement player rate) was about 14.5 in the 3rd round. In last year's Finals, it was 13.5. The lower RPR gives more eWins to lesser players; leaving a bit less for the stars.
When you determine RPR for a specific series, what is the basis for your calculations? Is it just the 2 teams involved, all playoff teams, or the entire league?
Joined: 14 Jan 2005 Posts: 1521 Location: Delphi, Indiana
Posted: Tue Jun 05, 2007 9:37 pm Post subject:
RPRate just falls out of the eW calculation. When I started experimenting with eW, I assumed a guy with T=15 is contributing half as much as a T=30 guy. But I subtracted an arbitrary number from T*Min, thinking it would be close to zero. But it's generally around 12.
It's counterintuitive, but RPR seems to rise when competition is severe. Players rates drop AND less of their stat total seems to be correlated to wins. Wins seem to be dictated even more by the best players. Maybe that does make sense.
I've just come to realize the way to go about it is to have a separate sheet for every playoff series. So far, I've only done this for the last 2 Finals. (A sheet generates unique parameters: weights for stats, starter/sub factor, tm/opp reb exponent, RPR, etc.)
I just have 5 sheets for the playoffs; one for each level, and one for the total field. This last one is weird because teams have played anywhere from 4 to 16 games. Player eWins added series by series don't equal the summary sheet's results.
It's all good, though. Each year builds on the previous one. Maybe it'll settle into a 'formula', and maybe not. _________________ 40% of all statistics are wrong.
I guess what stinks is with all these methodology/formula changes you have, their is no base to compare past players with todays players. I guess you could from about 2003 to NOW due to additional stats and likelihood you tried different approaches and included more stats with those particular years.
Do you have standardized Ewin/Rate stats from about 2003 to now?
It's all good, though. Each year builds on the previous one. Maybe it'll settle into a 'formula', and maybe not.
Have I mentioned how much admiration I have for your work on eWins based on your posts here? If not, allow me to.
I think there is no substitue for experimentation in method. I do hope that you continue to go back the other direction and think about what theoretical results are implied by the formula.
For example, I'm really curious about this:
Mike G wrote:
Players rates drop AND less of their stat total seems to be correlated to wins. Wins seem to be dictated even more by the best players.
What do you mean when you say that "less of their stat total seems to be correlated to wins." Are individual components of their total less correlated, or is there a general reduction in correlation?
I have been trying to figure out a statistical explanation for the puzzle of why (1) playoff games tend to be closer than regular season games AND (2) there are less upsets in the playoffs than in the regular season. Why is that?
Joined: 14 Jan 2005 Posts: 1521 Location: Delphi, Indiana
Posted: Wed Jun 06, 2007 1:33 pm Post subject:
Nick, I worded that most awkwardly. RPR is subtracted from T-rate, and this is multiplied by a player's minutes, and by a number generated by the 'goal seek' function in Excel. The 'goal' here is for the total of all players' eWins to add up to the number of games played.
So in my example, if RPR = 12, the T=30 player is left with 18 'positive credits' per 36 minutes, and the T=15 player is left with just 3. So I'd say the 'star' is producing wins 6 times as effectively as the 'sub', per minute.
Are there fewer upsets in playoffs? Well, series are a different dynamic than single games vs an opponent. In the season, the Spurs don't win as often as they should, but they rest their players more than anyone else. They get upset during the season, and in playoffs they prevail.
In playoffs, we seldom see a major player miss a game due to small injuries or personal issues. LeBron will play, and let his daughter deliver herself. _________________ 40% of all statistics are wrong.
Are there fewer upsets in playoffs? Well, series are a different dynamic than single games vs an opponent. In the season, the Spurs don't win as often as they should, but they rest their players more than anyone else. They get upset during the season, and in playoffs they prevail.
I believe that if you modeled a playoff series as a set of unrelated events with a probability based on some estimated win% that you would predict more upsets than actually occur. I also believe that this would be true if you used teams after-the-fact point differential based on that series.
I have not run the math on this, but I feel like playoff series are less random than you would expect.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum