Of the 98 players who were 3 time or greater NBA all stars and drafted after 5th,
8 of them ( Tony Parker, Shawn Kemp, Peja Stojakovic, Kobe Bryant, Dirk Nowitzki, Amare Stoudemire, Detlef Schrempf, and Jermaine O’Neil) were drafted at a time when high school and foreign players were routinely not correctly valued.
Exclude these players and the skewed numbers become even more skewed towards drafting in the top 5.
So I was playing with the data and my findings were as follows:
I organized the data by number of all-stars drafted per year. I left out the 3 most recent years as many of those players could still develop into All-Stars making the data incomplete.
The data covered 65 drafts and there were a total of 375 All-Stars drafted during that time. The Average number of All-Stars drafted per year was 5.77 and the Median number of All-Stars drafted was 6 and the standard deviation was 2.25. The Mode was 5, which I usually think is useless but, interesting in this particular exercise.
If I was to infer anything about this data it would be that every draft should be expected to produce 3-8 All-Stars and the likelihood of anything within that range is the same and that if you want to guarantee that you have an opportunity to pick one of them you would have to pick 3rd at worst.
Wanting more data I ran a regression on the data set (I'm really rusty in this area but forged ahead anyway) and the results suggest that the probabilities listed above are not statistically significant. I think this data set provides an interesting look into past drafts, I do not trust its ability to predict. I wouldn't balk if someone wanted to check my work or run this themselves, but these were my results.
If a team wants to "tank" I think the most reasonable way to approach it would be to base the tanking strategy (and targeted draft position) on scouting of the upcoming draft class to determine where you need to be to ensure a shot at the best players. I don't think using past drafts to determine a target draft position makes sense, especially when scouting data is available and considers the actual draftees.