Davis Bertans meets The Doppelgänger Machine

Before I drop Davis Bertans into the grinding gears of The Statistical Doppelgänger Machine, I thought I’d take some guesses at the type of player who might turn up. Then I’ll compare what I guessed against what the machine spat out.

My predictions:

1. Bertans probably won’t have close matches. He’s a specialist, and I suspect there are relatively few players who can match his volume of three-point attempts.
2. While he needs to play PF because he can’t guard threes or fives, and he doesn’t rebound much, I think we’re going to see SGs and SFs in his mix. When I think “three-point specialist,” players like Kyle Korver, J.J. Redick and Eric Gordon (after he got to Houston, but before he signed his current contract) spring to mind.
3. Typically with the doppelgänger machine, players tend to land within a relatively narrow PPA window. If a player has a 105 PPA (as Bertans did this season), it’s rare for him to get comps with a 200 PPA or replacement level. Because Bertans is fairly unique in terms of three-point volume, accuracy and overall efficiency, I think we may see that break down. I don’t think we’ll see Stephen Curry on the list (Curry is best known for his shooting, but at his peak actually contributed well across the stat sheet), but we could see a great three-point shooter and all-around excellent player like Ray Allen.
4. His closest comps will come from the past five years. There were some players in the 1990s and early 2000s who shot lots of threes, but prodigious launching is still only a few years old.
5. I expect some Euros.

Enough predictions, let’s get to the machine.

The Statistical Doppelgänger Machine compares a player to every other player season since 1977-78 across 14 categories, including age, playing time, my PPA metric and per possession box score stats. The machine does some math and spits out a list of comparable player season — similar production at a similar age.

Here’s what The Machine produced as comps for Bertans:

1. Allen Crabbe, SG, 2017-18, Brooklyn Nets, age 25 — Ugh. The idea of Crabbe was way better than the player. This was the best season of his career and he posted an 89 PPA. The stats are freaky similar to Bertans’. Bertans shot better from three point range so he scored a little more on near-identical usage. And Bertans was two years older. That’s the difference.
2. Ryan Anderson, PF, 2016-17, Houston Rockets, 28 — I should have put Anderson in my predictions. The per possession numbers are very similar. The biggest difference: Bertans shot better from two-point range. That’s about it.
3. Bojan Bogdanovic, F, 2017-18, Indiana Pacers, 28 — This was Bogdanovic’s first above average season (108 PPA), which matches up with this past year being the first above average season for Bertans. Their per possession numbers are pretty close, except that Bertans actually did a better job getting rebounds. This is an apt comp because I’ve thought — and continue to think — Bogdanovic’s contract is the level where Bertans will sign.
4. Eric Gordon, SG, 2016-17, Houston Rockets, 28 — My thinking that Bertans was unique isn’t holding up. Gordon didn’t shoot as well from three-point range, and didn’t rebound as well, but the rest of their numbers are quite similar. No shock that this is the second player from Houston.
5. Voshon Leonard, SG, 2000-01, Denver Nuggets, 27 — First name from outside the past five seasons. Leonard was a three-point specialist almost before there were three-point specialists. He fits the profile of each of the other comps (and Bertans) — volume threes at above average accuracy and not a lot else.
6. Robert Covington, SF, 2014-15, Philadelphia 76ers, 24 — This comp fascinates me because when I think of Covington, I think of defense. But, this was in analytics-driven Philly and Covington launched a high volume of threes at a decent percentage (.374) and the rest of their numbers were pretty similar. The big differences: Covington couldn’t make twos (.421) and had a steals rate double that of Bertans.
7. Morris Peterson, G/F, 2004-05, Toronto Raptors, 27 — While Peterson’s three-point volume was low by today’s standards, it was on the high side for that season. Yet another blow to my idea that Bertans was unique in league history. The numbers here are pretty similar.
8. Voshon Leonard, SG, 1996-97, Miami Heat, 23 — I should have predicted Bertans would have a comp from the three seasons when the NBA moved the three-point line a little closer to the basket. After 1996-97, they moved it back out. Leonard is the first repeat comp for Bertans.
9. Al Harrington, F, 2007-08, Golden State Warriors, 27 — Ah yes, the criminally overrated Harrington. This was his best season — a 110 PPA. He was slightly above average the following season and below average the rest of his career. The similarities are breaking down a bit at this point on the list. Harrington shot fewer threes and rebounded more. But the per possession scoring and overall quality is close.
10. C.J. Miles, SF, 2014-15, Indiana Pacers, 27 — Miles fits the overall Bertans profile, at least for this season. He took a bunch of threes, didn’t rebound much or generate assists. His defense was better than Bertans. This was probably his third best season (90 PPA). The big difference between the two: accuracy — Bertans shot .424 from three-point range; Miles .345 in 2014-15.

The top comp from Wizards/Bullets history was Martell Webster in 2012-13. This was the best season of Webster’s career (114 PPA). He hurt his back the following season and was never the same again.

How’d my predictions do?

1. Bertans won’t have close matches. Verdict: Totally wrong. There were definitely differences — generally three-point accuracy — but looked at across all categories, the doppelgänger machine found lots of guys who hoisted threes and didn’t do a lot else.
2. We’re going to see SGs and SFs in the mix. Yep — seven of the top 10 comps were SGs or SFs. I also tossed out names like Korver, Redick and Gordon as prototypical three-point specialists. Gordon made the list, but Korver and Redick were both solidly outside Bertans’ top 10 matches. Of the two, Redick in 2011-12 comes closest at 17th most similar.
3. The overall quality grouping (as measured by PPA) will break down because Bertans is unique. Totally wrong. The biggest PPA gap was 24 points to Gordon. I had to go 83 deep on the list to get to the kind of quality difference I anticipated — to Danny Green’s career-best 167 PPA in 2014-15.
4. His closest comps will come from the past five seasons. That was correct for his top four but fell apart for the next six. Two of those were Voshon Leonard but the list also included Peterson and Harrington from the early 2000s.
5. Euros! Bogdanovic made the list but I still count this as a miss because he was the only one.

Overall, the list seems to be comprised of decent role players who contributed in fairly narrow ways, but had a tendency to get overrated and overpaid, which became a problem when they got hurt and declined. Hmm, something about that seems familiar.