NBA UNDERS just miss going 9-0!

Average score so far may be 197.8 but we bet into medians, not averages. Median score so far is 191.0. A few crazy 12x-12x OT games are throwing numbers out of whack. Way too few games played so far to conclude much of anything IMO

[QUOTE=PerpetualCzech;8389]Average score so far may be 197.8 but we bet into medians, not averages. Median score so far is 191.0. A few crazy 12x-12x OT games are throwing numbers out of whack.[/QUOTE] Agreed. But I seriously doubt that new rims will change the relationship between the mean and the median. So which is the artifact? [QUOTE]Way too few games played so far to conclude much of anything IMO[/QUOTE] You don't get the best bets by waiting until the data is statistically conclusive. I've got the significance at a bit over 1 std dev, which is meaningful for a simple trend like total scoring. Notice that shooting% for all categories is up.

[QUOTE=rhinoceros;8393]I've got the significance at a bit over 1 std dev, which is meaningful for a simple trend like total scoring.[/QUOTE] Hmm, still not sold over here. There's a 32% chance of randomly falling one SD or more away from the norm. I'm still firmly in the undecided camp.

[QUOTE=PerpetualCzech;8396]Hmm, still not sold over here. There's a 32% chance of randomly falling one SD or more away from the norm. I'm still firmly in the undecided camp.[/QUOTE] Half of that 32% is on the side of the curve where you've already won. And experience has taught me that 1 std dev events are where the $ is. By the time the data looks like 2 std dev, the game has already changed. OTOH, unders are really doing well. So I am also in the undecided camp, with strong feelings both ways. I'm looking for the eureka idea that will make me believe I have the answer. FWIW, I have today bet 3 unders, 1 over, and 1 1H over.

[QUOTE=rhinoceros;8397]Half of that 32% is on the side of the curve where you've already won.[/QUOTE] No no no no. You're resulting. You're claiming you have something unusual because it's 1 SD from the norm. There's a 32% of that happening, period. Does ComptrBob have access here? Help! [QUOTE=rhinoceros;8397]And experience has taught me that 1 std dev events are where the $ is. By the time the data looks like 2 std dev, the game has already changed.[/QUOTE] I'm really surprised you're saying this. Deviation from the mean doesn't have "momentum". Of all cumulative results you analyze, going forward you can actually expect the majority of them to travel back towards the mean rather than the other way.

[QUOTE=PerpetualCzech;8399]Of all cumulative results you analyze, going forward you can actually expect the majority of them to travel back towards the mean rather than the other way.[/QUOTE] Or, put another way, a typical 2 SD phenomenon will have cumulative results that are 2 SD's the whole way as more results come in over time. If they are 1 SD to begin with and then move to 2 SD then that's just variance; no more unusual than initially being 3 SDs off and then moving to 2 SDs.

Czech, Your observations about statistical truths are correct. But I think you are nonetheless being deliberately dense about the points I am making. I agree that anyone mining 1 std dev trends and betting them is a fool. But there is more going on here. The corroboration of FG%, 3FG% and FT% is meaningful. And the new rims are meaningful. It's a situation where statistics used as a tool for persuasion is weak, but used as a tool for understanding, it is useful. The changed relationship between mean and median is very, very interesting. I haven't posted much on this and I don't fully understand it. My hunch is that this will be key to having an opportunity with totals awhile longer than otherwise. My point about betting 1 std dev events is not that there is some sort of "momentum" involved. Rather, by the time you find real trends with stronger statistical support, the market has already reacted and the apparent betting opportunities are gone. It is easy to say "small sample size," but I see precious few examples of people actually calculating how small a sample is. I believe the best opportunities arise when the sample is small enough that the statistical conclusions are weak, but large enough that the conclusions are not irrelevant.

I may be dense, but it's not deliberate. All I was responding to was the evidence you were giving and what I thought were some holes in logic. We need evidence. All we have so far is the 1 SD result and all I'm trying to say it's not near enough to conclude we are in the midst of a permanent uptick in scoring. The increase in FG%, 3FG% and FT% is indeed helpful, especially the FT%. That's something different than seeing that the increase is due to increased number of possessions. How many SDs are those to the right by? (using last year's results as the norm?)

FG% 3FG% FT% Pts/g 2008-09 .443 .334 .747 190.3 2009-10 .455 .363 .760 199.9 but beware that the conclusion is not as strong as this suggests-- early 08-09 was on the low side. from this thread: https://www.lvasportsboards.com/showthread.php?t=1294&highlight=rims There oughta be a way to merge these threads; is there a moderator reading this?

[QUOTE=rhinoceros;8410]FG% 3FG% FT% Pts/g 2008-09 .443 .334 .747 190.3 2009-10 .455 .363 .760 199.9 but beware that the conclusion is not as strong as this suggests-- early 08-09 was on the low side.[/QUOTE] I knew these numbers from the ESPN article but it's SDs that we need, not just the absolute change. THAT'S what's going to start giving us some insight.