These show the game-by-game Game Score ratings for some of our players. The red line is the Game Score rating per 36 minutes played over the last 5 games. Keep in mind this is just based on the box score, so its not a perfect player rating by any means (defense isn't covered particularly well, for instance). Game Score tracks PER very closely, which rates Lin, Parsons, Patterson, and Asik all very similarly (close to 15 which is league average).
Interesting. Sorry that I need a reminder on how GmSc is calculated..... Not sure whether I read the graph in the right way, is that means that Jeremy, Parsons, and Asik are relatively stable in GmSc36/5gm, and Harden and Patterson have big swings? Not sure whether Harden's swing correlate with the winning and losing streak or not, looks like it. would that suggest Harden's performance would have big impact on whether we win it or not? Good thing is that all 5 players have an upward trend towards the end of the graph, so hope they keep it that way.
Thank you for promoting GmSc! Would rep you if I could. I was thinking of doing the same thing at some point, but thought it would be more meaningful to do it at the end of the season. (Plus I can't start topics) While I wouldn't say it's a perfect stat, I do think we need a stat to show single game performances that scales according to performance. Right now, GmSc is really the best available stat for that purpose. PER and GmSc are basically derived from box score stats and it doesn't surprise me that they trend the same way. I'm curious as to why you did GmSc36/5gm instead of just GmSc extrapolated to a 36 min basis? The former has the effect of rounding out the peaks and valleys no? Quick things that I'm reading from the graphs: -It's really interesting to see that Lin, labeled as "Linconsistent" is actually the most consistent person among the starting five. -Harden had a tremendous peak in Dec/early Jan. Even in his "lulls" he's still an above average player, which is just incredible. -GmSc doesn't do Asik justice because of his defensive contributions which aren't really recorded in the box scores. (One of my critiques of GmSc is that I don't think the coefficients used is properly balanced among box score stats. Specifically, rebounds should weigh more since they lead to more possessions and possessions is what you need for points.) -2Pat is by far the most inconsistent. I really think 2Pat is the key to the Rockets success this year more than any other player. If we plays well, he's at least an average PF. If he doesn't, the Rockets are playing 4v5 essentially speaking. I would guess that if you overlay the Rocket's W/L record over 2Pat's GmSc's you'll see a very strong correlation.
When people say "consistent" they mean "consistently good". Lin is not "consistently good" he is "consistently not good", thus inconsistently good. That is what they mean.
I would say Lin is inconsistent on a game by game basis, but it is eerie how consistent his five game averages are.
I'm not a fan of Game Score because of the flaws related to defense as you noted. I think it was TD's game against OKC where he scored 17pts and had 5 steals with a game score of 15, but his +/- was -30 or something like that. The Rockets were behind by 23 @ the start of the 4th and TD scored 6pts in the final 3 minutes of the game and still ended up losing by 30.
Here is the formula for Game Score: "Points + (FGM x 0.4) + (FGA x -0.7) + ((FTA-FTM) x -0.4) + (OREB x 0.7) + (DREB x 0.3) + STL + (AST x 0.7) + (BLK x 0.7) + (PF x -0.4) - TO" I did notice the big swings for Harden and Patterson as well. I expected something similar for Lin, but except for November when he really struggled out of the gate, his overall performance has been fairly even.
You do realize how the terms "consistently good" and "consistently not good" is subjective. Depending on how you define that, you can find stats to defend your argument. Since the premise of the argument begins with someone's opinion and people clearly don't have the same opinions about Lin, people continue to have disagreements about Lin's perceived value and consistency. Every stat quantifies a narrow range of events and thus are limited in their representation. That does not make them more or less valid or invalid. That makes them tools to serve in advocating or countering an argument. The key, as you'll probably agree, is that you use these tools in situations where it is logical to do so. As a means to quantify offensive performance and some defensive aspects of a player in a singular game, GmSc is the best stat at achieving the goal. Until the NBA starts keeping more detailed records of games or until Synergy makes their data public, it is unlikely the situation will change.
My point is that I can look a regular box score to see if a player did well or not. Add in the play by play log and I can check details about the game to see if there was any garbage time stat sheet stuffing or see if player x was on the court when player y scored 20pts in a quarter. I don't need a "game score" to tell me that. It's just one more stat that we dont really need. It's a "magic formula" box score stat like Hollingers Power Rankings and PER, they have some use but in the end you'll find many flaws. Morey said in a radio interview that all publicly available(ie free) defensive stats are useless. I think he was talking about all the advanced hybrid box score stats.
He's consistently average (this season). That's not a terrible thing, it means he is only slightly overpaid.
No, look at the scale to the left. His goes from -5 to 40. The biggest range of any of the players. He has more negative games than Parsons, but his top 2 are better than Parson's top 2.
Yeah his individual games are actually the most inconsistent, but he seems to always even out over 5 games. If he has a really good game (Spurs 38 pts) it's surrounded by duds. He doesn't string a ton of really good or really bad games in a row.
And I would disagree with that. While my opinion is not based on the above chart, the above chart does support that conclusion. They defined average as "15". He eclipsed average a mere 14 games this season. He was near or below "10" ~30 games this season. The remainder of his games lie right around 15. So to summarize, volume and performance: ~28% of the time he plays above average ~12% of the time he plays average ~60% of the time he plays below average In my book, that is a below average basketball player.
So your argument is what? There's too many stats? Can a mechanic have too many tools? Not if the tool makes some aspect of his work easier. As a person who gets paid to work with stats, I love looking at new stats. What's your other arguments? That there is no "magic stat" that puts players in a nice, neat ordinal rank? That all stats are narrow in their scope? That no one stat can cover everything or explain everything? That the present defensive stats suck? Because I've already said all of that explain why it is the way it is and precisely why we need more stats. You know what, I like GmSc. It serves a purpose for me. People have thanked me for posting these stats at the end of Rockets game, so it's logical to assume that other people find value in them as well. Why don't we leave it at that. Feel free to NOT use GmSc all you want.