Is the implication that Jordan won't be #1 on shooting guards list?

the margin of error here is about 5%

the "margin of error". It's a method pulled from someone's asshole and it's underlying stats are also based on bullshit. But more importantly, there is nothing to compare it to in order to come up with a margin of error.

Keep in mind my margin of error here is 3.4%

Bruce, how did you calculate the margin of error?

If the underlying stats are based on "bullshit" how do you explain it coming up with a list that is not very dissimilar from the lists that you idiots pull out of your asses?

for one, the similarities in all of these things- the observations, the advanced stats, the traditional stats, opinions of peers, etc- happen because underneath all of it is "production". points, rebounds, assists, steals, blocks, fgs, fts, 3s, etc. production plays a huge role in how players are seen, as does efficiency. and its fair, production and efficiency on offense are hugely important pieces of information. every individual component of most advanced stats is- by itself- useful information.


the problem i have with WinShares and the other various advanced stats is they arbitrarily assign weights to different examples of production, sprinkle in the efficiency to varying degrees, leave out a ton of stuff in the process, add in priors that have their own issues that distort the end result (not in the case of WinShares of course), and then people like you treat the end result as gospel because you think basketball is baseball when its not. its not because I think its impossible for WinShares or PER or VORP or Box+/- or RAPM or whatever to hint at great play, because they are very capable of doing that. after all, they are based on stats that are independently useful.

but WinShares has all kinds of issues, especially when you go back to the 60s. and the margin of error basketball-reference claims proves WS's usefulness is also flawed. they cite a +/- 2.74 win average error. but that's not a 2.74% error like you'd see in a poll. that's the number of wins that it is off on average, which on average for the entire NBA is 41 per team (that's 6.6% off). more importantly, half of the equation (the defensive half) is literally based on the result and worked backwards. when you start with the result and work backwards, of course all the pieces will add back up to the result. that doesn't prove that the way the divide up the defensive credit was accurate at all (and spoiler, it isn't). so they are 6.6% off despite literally being gifted 50% from the start. having a 6.6% margin of error on 50% of the data is not impressive at all.

If these numbers are useful but Win Shares is doing it wrong., why aren't YOU coming up with the right way to use them?

Because the right way to use them isn't to combine them into one pointless metric that attempts to be the only stat needed for analysis.

What if I told you that you could simply use individual stats to analyze that specific thing within the larger context of what's happening on the court? Shocking I know.

There is nothing "advanced" about one-number metrics. There is a new hip one every few years and then we all see how flawed it is and move on to the next one. It happened with PER a long time ago, and win shares after that. It'll happen with RPM soon too. Because that's not how basketball works.