In part one, we introduced PER as the metric du jour for our draft study. In part two, we applied the study to the current Bucks roster, and found out that they're not all that good, at least in terms of PER. Now, in our thrilling conclusion, we present the results of the study conducted over the past few months, and hopefully get some direction on where the team might look for improvement.
With a record of 35-47, the Bucks were "good" enough for the 10th overall slot in Monday's draft lottery. Barring a trade, they'll likely stay at the same position they were at in 2009 when Brandon Jennings was brought aboard. Conventional thinking would leave us disappointed; after all, the real franchise cornerstones are found at the top of the draft, not in the late lottery. On a team full of capable role players, the last thing the Bucks need is another one, and that's exactly what the 10th overall pick offers. Or does it?
The groundwork for the draft study is the same as in part two: Players' performances over the course of their career peaks were displayed in what we call the "3-year PER peak". The metric is the average of the player's three highest consecutive qualifying seasons' PER measurements. If that doesn't quite make sense (and I can empathize if it doesn't), here's a refresher on what a qualifying season is:
By our standards, a qualifying season is one in which the player plays in at least 50 games and averages at least 12.0 minutes played per game. If one of those two conditions is not met, that player's season is rendered non-qualifying for the purposes of this study.
So we go over a player's career stats, eliminate any seasons that don't meet these benchmarks, then find and average the three remaining highest consecutive season PER scores. Simple, no?
Now that you're up to speed, here are some of the finer points of things:
The study reaches back to the 2000 NBA draft, when Kenyon Martin was taken first overall by the New Jersey Nets. We considered the top-16 draft choices each year; although the lottery technically cuts off at 14, we went two picks extra because a) it divides the top of the draft into four neat little tiers (picks 1-4, 5-8, 9-12, and 13-16), and b) there are usually players who are projected to go in the lottery but slip slightly below (Hedo Turkoglu, Al Jefferson, Rodney Stuckey), and going two picks further helps catch those players.
Below is the basic product containing just the 3-year PER peak average of all 176 players considered:
That's a lot of information in one table, and it's only the basic product. We went way deeper into the numbers, and the things we found are simply too robust to post here. Thanks to the power of Google, you can look over the spreadsheet yourself. Click here for the entire file.
While you're free to come to your own conclusions, here are a few things that we noticed that are interesting:
1) The draft is big. I don't mean this in terms of importance to rebuilding teams, or as a league event, or as a milestone for prospective players. I mean this in terms of positions drafted: 41% of draft picks over the past decade-plus were either power forwards or centers. The remaining 59% is split between guards and wings, with a slight advantage for the latter.
2) The draft is a gamble. Maybe not a gamble per se, but there's definitely some calculated risk involved here. By taking qualifying seasons and dividing them by the total possible seasons, we can get a percentage that reflects how much of a player's career will meet the minimum threshold (50 games played, 12.0 minutes per game). The table itself doesn't lend to a convenient cut/paste, but I can tell you it's on page 1 of the "SUMMARY" tab on the Google Doc.
|Qualifying probability||Overall Pick|
Surprisingly, the data shows that there is not as much consistency among the top tiers of draft picks. For example, the first overall pick does not have the highest "success percentage"; picks 3, 4, and 7(!) have a higher likelihood of meeting the aforementioned season marks. Another surprising find is that picks 11 and 12 both come with a success probability of less than 50%, making them the riskiest positions in the lottery.
3. Busts are every year, but not everywhere. For the purposes of our study, we defined a bust as a player who could not muster a 3-year PER peak greater than 10.0. (2009 and 2010 draftees were exempt, since they haven't really gotten a fair chance yet). Low expectations, sure, but the last eleven drafts averaged nearly 2 busts a year.
The busts, however, are definitely concentrated at the bottom of the lottery. Picks 1-8 yielded a total of five players who failed to meet the most rudimentary of benchmarks. Picks 9-16? Sixteen busts, including players like Robert Swift, Cedric Simmons, and Sean May.
4. GMs don't suck as much as we make them out to suck. This comes straight from Steven von Horn, our resident stat-man:
"In general, NBA scouting departments and GMs do a pretty good job of approximating the talent level in the draft and selecting the most talented and most successful players with the higher picks the majority of the time. This is pretty impressive stuff nearly any way you look at it:
- Of the 4 players with a three-year peak PER above 25 in the past decade, 2 have been selected with the 1st pick overall (LeBron James in 2003 and Dwight Howard in 2004). The two other players were also taken in the top 5 picks (Dwayne Wade at pick #5 in 2003 and Chris Paul at pick #4 in 2005).
- Of the 16 players with a three-year peak PER above 20 in the past decade, 12 have been drafted in the top 5 picks. Meanwhile, only 1 has been drafted outside the top 10 picks (Al Jefferson at pick #15 in 2004).
- Of the 78 players with three-year peak PER above 15 in the past decade, only 16 of those players have been drafted outside of the top 10 picks. That means 79.5% of the above-average players, according to PER, drafted in the past decade have been taken in the top 10 picks."