NBA UNDERS just miss going 9-0!

Choosing this size fraction of 2 random seasons, the following table shows the std dev of shooting % in each category. That is, we should expect one year to differ from the other by about this amount: FT% 0.89% FG% 0.45% 3FG% 0.89% Before evaluating these, notice that since this thread started, 2009-10 shooting % numbers have dropped meaningfully. The current (thru 11/4) numbers are: FT% 75.6% FG% 45.2% 3FG% 35.9% Each category has dropped 0.4% since the ESPN article. We are considering 11,000 FGA, 3,600 FTA, and 2,400 3FGA. So at first blush the shooting % numbers appear to be off last years' by 1, 2, and 2.6 std dev. Actually, the numbers are even less, because we know that last year was a bit of an anomaly. I don't expect anyone has multi year averages through a certain fraction of the season? Still, these numbers are not really small. There is obviously some correlation between them, but I believe it is very small for FT% compared to the other 2. So while it is wrong to consider these as independent support for the hypothesis of increased scoring, they are certainly much more meaningful than any 1 (or even 2) individually. My tentative conclusion is that this years' scoring increase is not just a small sample artifact, although I don't claim a high level of confidence in that conclusion. Even less certain (but more valuable) is the relationship of that to totals' betting trends. We've done a lot of analysis on scoring; what about the lines? How do this season's totals lines look compared to the past?

[QUOTE=rhinoceros;8448] Still, these numbers are not really small. There is obviously some correlation between them, but I believe it is very small for FT% compared to the other 2. [/QUOTE] Could it be because FTs are less likely to need the softness of the rims to go in, therefore benefiting less?

I was talking about the size of the correlation, not the size of the effect. 3FG% is a part of FG% so there is a meaningful correlation right there. The correlation between FG% and FT% is much more delicate: Higher FT% increases the cost of fouling, motivating looser defense and increasing FG%. Wow. That's gotta be a really small effect. Intuitively, I would have expected FT% to be the category most affected by different rims. And, so far, I would have been way wrong. Data is usually better than intuition, and almost always better than my intuition. That's why I do numbers instead of watching the damn games.

I think it's more randomness than anything... So now free throw percentage is down from last year... Last year was .771, which throws everything out the window... Teams change, Memphis first 5 games last year avg. 174.4 points per game, this year 221.8... There's a bit more pace this year, etc, etc

[QUOTE=rhinoceros;8448]Choosing this size fraction of 2 random seasons, the following table shows the std dev of shooting % in each category. That is, we should expect one year to differ from the other by about this amount: FT% 0.89% FG% 0.45% 3FG% 0.89% ... So at first blush the shooting % numbers appear to be off last years' by 1, 2, and 2.6 std dev. [/QUOTE] Hang on a sec. Those are SDs for an entire year. We are only 5 games into this year. SDs for 5 games should be only 5*5/82/82 = 1/269th the size of the whole year's SD, no?

[QUOTE=PerpetualCzech;8460]Hang on a sec. Those are SDs for an entire year. We are only 5 games into this year. SDs for 5 games should be only 5*5/82/82 = 1/269th the size of the whole year's SD, no?[/QUOTE] I couldn't find the original ESPN article, but I thought the numbers were for last season through a similar number of games. Anyway, I calculate SD by treating each shot as an independent event. I accumulate them as a binomial distribution. Since different players shoot at different rates, this is a simplification, but the difference is TINY. SD of overall scoring is higher than that of shooting % because the number of shots is variable--think of the crazy 2OT games, etc. Also, I think you erred in squaring 5/82 when you should take a square root. But be careful to distinguish between SD of shots made and SD of shot% rate.

I know this is a little out of place, but we're making this the "Thread of the Week" on LVASports.com and we wanted to include this good post from RealWorldSports that appeared in an associated thread. That post begins below. New rims in NBA and increase in 3-point shooting but not 2-point shooting. Overall scoring up pretty good, but 56% unders so far. If new rims truly are a factor, could be a big deal to know what kind of rims each college venue uses before touching a college hoops total. Wrote this in another venue. Feel free to shoot holes, add thoughts....... NBA teams are scoring an average of 99.96 points per game, vs. 95.17 points per game last season, which ESPN’s Marc Stein attributes to the new, more forgiving rims in use by the Association. The rims which instituted as a safety measure, as they collapse from both the back and the front. Shooting is up just a smidge over 1%, and three-point shooting has improved dramatically, from 33.4% to 36.3%. Surprisingly, NBA bettors have not been scoring with their over bets. In fact, over the season’s first 50 games, unders actually have the advantage, with 28 games going under the marketplace projection, and only 22 games going over. So scoring is up, yet unders are cashing at a 56% rate. Reason for the disparity? Teams can shoot lights out in a game, send it way over the total, yet it only counts as a single over. For instance, in Sunday’s 133-123 Denver win over Memphis, the teams obliterated the total of 207 by 49 points. Each team shot over 55% from the field and the Grizzlies and Nuggets combined to shoot 17-35 from outside the arc. The 168 shots in the game is a healthy number, as well. So while a game like that can have quite an impact on the overall shooting percentages, it is only a single total. We’re also looking at a small sample size, only a week of action consisting of 50 games. In the first four nights unders cashed at a 20-11 rate, but scores have been higher from Saturday through Monday, with overs going 11-8. The new rims and increased scoring is something to be aware of, but there clearly hasn’t been a lot of profit to be made at this juncture and the sports betting marketplace is sure to adjust quickly. Chances are very good that by the time the last shot is made next spring, overs and unders will be very close to 50/50. An edge might be available in early season college basketball for totals players. Assuming the new rims are not standard in college basketball, knowing what kind of rim is being used could provide a meaningful edge for college hoops bettors.

Let's revisit this subject. Who's got relevant, current data?

[QUOTE=rhinoceros;11100]Let's revisit this subject. Who's got relevant, current data?[/QUOTE] I've got plenty. What precisely would you want? The median/mean dichotomy is essentially gone. Medians came up to the mean: Mean YTD: 199.67 Median YTD: 199 At the end of November last year we had: Mean: 195.99 Median: 198 Average closing total for games as recorded by Don Best so far this year: Mean: 199.01 Median book total: 196.5 Over - under - push YTD: 128 - 128 - 3

[QUOTE=npc;11102]Over - under - push YTD: 128 - 128 - 3[/QUOTE] I have only 258 games played this year so far, not 259. I also get slightly different numbers for means and medians. Number of wins in the standings also adds up to 258. Not a big deal, just an FYI if you are interested in some data maintenance.