Actual vs. Pythagorean wins

[Ellipse]

I know I've mentioned this before, but it seems worth bringing up again amid what seems to be a somewhat disappointing year so far:

Why do you think the difference between our actual wins and Pythagorean wins (expected wins based on point differential combined with schedule) is so stark under Quin Snyder? This year, our Pythagorean record (according to Basketball-reference.com) is 23-17, so we've "left 3 wins on the table," so to speak. Last year we had 5 more Pythagorean wins than actual wins.

You might figure that this difference between Pythagorean and actual wins is somewhat random and evens out over the years. But over the 4 1/2 years of Quin's coaching, we've had by far-and-away the largest difference between Pythagorean wins and actual wins of any team in the league. Only 4 teams are over 10, in either direction. We're at -19 (actual compared to Pythagorean). We've never had a season under Quin with more actual (or even the same number of) wins than Pythagorean wins.

Short answer: The model needs to be tweaked.

Also, it seems a bit silly to try replacing the actual record with a hypothetical record. If the algorithm isn't going to be used as a probability estimator of future wins, then it serves little purpose.

 

[framer]

I agree. But I don't see how this applies to Pythagorean vs actual record (if you meant it to).

Part of the issue is that good defense can only go so far. When you build your team's identity on it, you are going to have inexplicable failures where the other team shoots well despite your playing good D. Good offense usually beats good defense. Generally, however, good defense will destroy mediocre teams which is why our point differential gets so high.

If you make playing good Defense more dependent upon how the refs feel that day, you will get all kinds of variation in your statistics.

 

[idiot]

Seems that the "running-the-score-up-in-garbage-time" possible explanation doesn't fly this year

​

 

[NAOS]

Seems that the "running-the-score-up-in-garbage-time" possible explanation doesn't fly this year

​

I wonder how much that loss in Dallas is throwing off any/all statistics of this type. I know this stat is trying to adjust for that game, but I wonder what it would look like if that single game was subtracted from the sample.

 

[idiot]

I wonder how much that loss in Dallas is throwing off any/all statistics of this type. I know this stat is trying to adjust for that game, but I wonder what it would look like if that single game was subtracted from the sample.

Which direction are you thinking it throws things off?

(I tend to agree that a 50-point loss can throw things off. But I think that at least the original differential this post started with [actual vs. Pythagorean wins] would be even wider without it.)

 

[fishonjazz]

Jazz are 9-4 in their last 13 games even with a slew of injuries

Sent from my ONEPLUS A6013 using JazzFanz mobile app