- Corey Pronman, ESPN Insider
This is the third and final piece in our series reflecting on where stats saw future success (or failure) on the horizon for draft picks. After first looking at forwards and then defensemen, we now look to the crease.
To review, many studies have shown there is notable predictive value in stat-based analysis for prospects. The main issues facing analytics for junior, collegiate and other levels are adjusting for context and trying to separate the relationship between production and the typical uncertainty/miss rate of projecting prospects. After all, stats can't tell you everything about prospects. They need to be adjusted for a variety of factors, they can't replace scouts, and they are subject to statistical noise. However, if used correctly for the NHL draft, stats can help teams find some overlooked sleepers and may prevent them from making a glaring mistake.
Goaltenders are notoriously difficult to project -- which is why teams have not shed too many tears after "wasted" first-round picks on players like Al Montoya and Jean-Francois Damphousse. Nevertheless, there have been some cases where a player's draft position could have been considerably different had teams given more credence to the metrics.
Lundqvist was chosen in Round 7 (205th overall) by the Rangers in the 2000 draft. He is widely acclaimed as one of the great late-round picks of the past 15 years, but it is debatable if he should have fallen that far.
Lundqvist, as a 16- and 17-year-old, played in the Swedish under-20 league -- for which stats are limited -- but there were some statistical signs of him standing out. A basic stat, games played, actually tells us a lot. Lundqvist played 29 games as a 16-year-old in the Swedish under-20 league in the 1998-99 season, which is rare. No other goaltender equaled that feat until Jhonas Enroth in 2005. One Swedish scout said that it is only in extremely rare cases that a 16-year-old goalie would start in that league. Lundqvist posted a .904 save percentage in the 1999-00 season, which was fourth in the league. No other 17-year-old goaltender had a SV% over .900 with regular appearances until Enroth again in 2006.
Evaluating across leagues and eras is tough, but relative analysis is a great way to do it. Lundqvist was given unique responsibility relative to most under-18 goaltenders in Sweden, an indication that his organization thought highly of him. Not only that, but he made the most of that opportunity by performing well. If teams believed in his tools, there's a good argument he should have gone at least 100 spots higher in the draft.
Bryzgalov was a bit of a late bloomer, who in his third draft-eligible season had a fantastic year. He worked his way up to the Russian Super League where, in 14 games as a 19-year-old goalie, he had a .930 SV%, raising that mark to a .947 in seven playoff games. He was also the top goaltender at the 2000 World Junior Championship and later started for Russia at the World Championships.
His sample size wasn't reliable for goalies, but keep in mind from then until the end of the RSL in 2008 that only three other under-20 goalies had been regular starters: Andrei Medvedev, Semyon Varlamov and Sergei Bobrovsky. For those unfamiliar, the first man on that list was a second-round pick of Calgary's. You can debate the relative merits of these players -- and I differ on the value of goalies in the marketplace -- but NHL teams have clearly valued players like Varlamov and Bobrovsky highly. Bryzgalov's 1999-00 season was superior, if not clearly superior to any of them.
I wouldn't have picked him in the top two rounds, but based on how NHL teams draft goalies, his selection (44th overall) seemed low given his size, raw tools and outstanding performance in an elite league, not to mention his international accolades.
It's tough to keep piling on this guy, but there was statistical evidence even back when DiPietro was taken No. 1 overall in the 2000 draft that this could be a poor choice, not even taking into account the high uncertainty that comes with projecting goalie prospects into the NHL.
DiPietro did not really have a long track record of sterling performance. He was 20th in the NCAA in his draft season in SV% among regular goaltenders, with a .913 in 29 games. That's good and impressive for his age, but not amazing. Among under-20 goaltenders overall in the college ranks, he was fifth, and in his conference, he was second. The prior season he was 11th in the USHL in SV%, playing for the USNTDP, with a .907 in 30 games. His backup, Chris Gartman, had a .912 SV% in 17 games. DiPietro was also poor at the World Under-18s that season.
On the other hand, DiPietro did have a fantastic performance at the WJC in his draft season, when he was named the best goaltender. Mike Milbury -- the Islanders' GM at the time -- indicated DiPietro's puck handling, goals-against average and win-loss record as reasons for the selection. The problem with that line of thinking is that puck handling does not have a significant effect on games and the two stats Milbury cited are not reliable analytical tools.
DiPietro ended up being a decent NHL goalie prior to his myriad injury issues and he has a lot of talent, but it was debatable if he had really shown he was a truly elite prospect leading into the draft, even without the extra positional risks.
Potvin was drafted in the middle of the second round by Toronto in 1990 out of the QMJHL. While available statistics were quite basic back then, there were still signs that Potvin could be a quality NHL goaltender.
The most telling signs were his usage and performance. Potvin was given an unusual amount of responsibility at a young age. He played 65 games as a 17-year-old and 62 games as an 18-year-old, which was something that simply didn't happen in that era. Martin Brodeur went over 50 games in only one of his three QMJHL seasons. Patrick Roy would be the closest at that age as he played 54 games at age 17 and 61 at age 18. In Potvin's 17-year-old season, he had a SV% of .868, and as an 18-year-old, his SV% was .885, but back then those were great numbers. For example, Potvin's .885 in 1989-90 was fourth in the QMJHL and everybody above him played notably fewer games.
The reason why the games played is so significant is that a goaltender's performance can be volatile in a short number of games. When goalies are playing 20-40 games in a season, the data isn't very reliable from a statistical standpoint. Now, 60 games isn't much better, but the extra data help provide more reliability in the performance, especially since it was reasonably strong performance.
Potvin shouldn't have gone in the top 10, but given how NHL teams valued (and continue to value) goaltending, he arguably should have been toward the middle or end of the first round.
Corey Pronman identifies four cases among drafted NHL goaltenders where metrics should have either raised a red flag or spotlighted a player's skill.