That's like saying Miami's performance is determined by Spoelstra. The team goes as Harden goes, McHale just has to not screw up the line ups which he does on a consistent basis playing Beverly and Robinson in the 4th quarter instead of Asik and Lin. Also giving heavy minutes to Delfino in situations when DMo should've been in the game guarding bigger players. With as much turn over as we have on our roster, there is no point in making players earn playing time, and paying their dues. What is the point, they'll just be traded in 6 months aways. Play the best players you have, stop trying to teach a lesson. <br><br>
We tend to win a lot and lose a little. We are better than the records shows, but I'm hesitant to say we are underachieving because the expectation in preseason is low.
Nice thread by the way. Thanks. Knicks, Sixers. . . . That makes me think: I'd like to see this graph for Western Conference only. The East is a different beast. Sorry to be a pain
You probably don't realize this but you just helped me make my point. Disclaimer, I don't advocate adjusting stats to fit the narrative. It's something that is done a lot in mainstream media and it's something that I just abhor. Anyway, what I said: Blowouts are insignificant. The close games are what really skews Pythag win analysis. Let's remove 190 points from the games where the Rockets won by 10+, so that their margin of victory is just 10 points. Pythag Win/Loss is now 32.5. The Pythag expectancy dropped by 6.5 games and now the Rockets are over achieving. Now, for my test. Let's add 10 points to all the games where the Rockets loss by 8 or less. From my earlier post: the Rockets are 12-18 in games by 8 points or less. So we're only adding 180 points in close games vs 190 points in durvusa's blowout case. Guess what? The Rockets just won 18 more games. See what I did there? See how I compared removing points from blow outs vs adding points to close games? -6 wins vs +18 wins. Swing of 24 games from one scenario to the other. Relatively to the impact of close games, blowouts are insignificant. That's why you should look at the close games. They are the ones that make or break a team's playoff chances.
Yay a post that doesnt include how horrible our PF's are or how much you think we ****ed up by traded the second coming of jesus christ (PP).
Your comparing apples to oranges here: -6 wins according to pythagorean W-L versus +18 wins according to actual W-L. If you subtract 10 points from our score in games we win by 10+ points, then our pythagorean W-L drops from 38.7 to 32.5 (-6.2 expected wins), in accordance to what you calculated. But if you add 10 points to our score in games we lose by less than 8 points, then our pythagorean W-L increases from 38.7 to 42.9 (only +4.2 expected wins, not +18 wins).
I wonder where HOU ranks among all 30 teams in terms if the standard deviation of game margin. This measure should tell us something about the theory that this team is more inconsistent than usual. My casual impression is that, for a team near its neighborhood in terms of either avg margin or w-l record, HOU's team has had an unusual # of very high margin wins.
This. is. how. stats. are. used. But seriously, you'd have to get a STDDEV of all the teams in that graph, find the number of samples per team (I bet 82!), go to your student t distribution chart, find your delta S, and then use that to calculate your confidence interval. This not only would tell you the deviation from the norm of the data set, but it would also tell you where 95% of deviations would fall within assuming there is normal distribution. Until then, all the back and forth on these stats really doesn't matter much. Confidence intervals on some graphs are huge - I have had to make graphs bigger because the CIs were larger than the graph themselves. Large CIs usually mean there isn't a correlation to the data presented with the variables suggested, or that the test methods/protocols/procedures need to be improved to increase precision (not the case here obviously).
Or you can just select the data range in your spreadsheet and select the option "add error bars with standard deviation"
Except a few games with OKC and Spurs, we pretty much have a chance in every game. When we are hot, we blow out teams. The biggest reason maybe our three headed monster in Harden-Lin-Parsons. All of them can shoot(Lin catches up lately), attack inside(Harden and Lin more so than Parsons) and pass. Of all those qualities, the ability to attack the basket and draw fouls is the most important one. Not many teams in the league has their 1,2,3 so versatile and able to attack the rim as we do. Imagine a plane with 3 engines, it's as safe and powerful as it gets. Our second tier guys are good too. Delfino/Beverly/Gacia/Anderson can't attack the basket as effectively, but they all can shoot and pass and play some defense.
