Diamonds in the Rough: Finding Elite Talent Late

Updated: July 11, 2016 at 2:00 pm by Byron Bader

The later round picks (round three on) in the NHL draft have provided very few true NHLers. From 1980 to 2015, 6,127 selections were taken after the second round. Of those, 815 of the 6,127 have gone on to play in the NHL in a meaningful way (150 games or more) thus far. That’s a success rate of roughly 13%. 

As a result, late round picks are not treated as a high commodity by teams. Late round picks are mostly used as filler in larger trades or used to acquire energy players or “room guys.” Or GMs take complete gambles on extreme low-probability players with those picks.  

The belief seems to be that finding an NHLer or even an elite NHLer late is a stroke of luck. You pick a player with little upside and a few years down the road, he’s turned a switch and turned into something. How can you predict that? 

However, if you look closely at the elite talent, namely forwards, selected late… there’s a few things that stand out which suggest it’s possible to improve on the extremely low probability (1.6%) of finding great talent later in the NHL draft. 


I’ve gathered all the forwards drafted between 1980 and 2015 who have played at least 150 games in the NHL to date and registered a points/gp rate of 0.60 or more over the course of their career. I’ve also looked at a number of factors, including: age drafted, height, weight, country of origin and junior/euro scoring/NHLE thresholds. In total, 100 players fit these parameters.

Interestingly, 75% of the players fit into three groups: small – 5’10 or smaller; drafted as an overager – drafted after their true 18-year-old season; and/or European – were born in Europe and played their pre-NHL hockey in Europe (Russia, Sweden, Finland, Czech Slovakia, etc.). Meanwhile, 25% were low probability picks that worked out (i.e., the lucky hit). Below is the full breakdown of the 100 players as they relate to each one of these groups, but in total 31% of the elite late surprises were small, 40% were overagers and 46% were European.

  • Overager/European – 17 players (17%)
  • European – 17 players (17%)
  • Small – 13 players (13%)
  • Overager – 11 players (11%)
  • Small/Overager – 7 players (7%)
  • Small/European – 5 players (5%)
  • Small/Overager/European – 5 players (5%)
  • Other – 25 players (25%)

As travel becomes easier and more data is readily available, the European affect is starting to wane. We’re now seeing players like Ovechkin, Malkin and Tarasenko taken very high, and rightfully so. As well talents like Panarin, although still undrafted, are being recognized earlier and earlier and are making their way to the NHL. These days, if you’re lighting up a men’s league in Europe at a very young age, NHL teams should know about you. But in the 80’s and 90’s, players like Alex Mogilny, Hakan Loob and Pavel Datsyuk slipped through most teams’ fingers because most teams had very little intel on them.

But teams are still incredibly hesitant with overage players and small players. Players like Doug Gilmour, Mark Recchi, Theo Fleury, Marc Savard and Johnny Gaudreau had the elite vision, talent and point production in their draft year to suggest they should be first round picks  yet they were skipped over hundreds of time until somebody “rolled the dice” on them. Despite the enormous success elite small players have had, for years and years, they are taken later than they should be (see Alex DeBrincat, Vitali Abramov and Matthew Phillips in the 2016 draft). 

The same story emerges with overagers. A player that registers 100 points in an OHL season in their true draft year will go in the first round, unless extremely tiny. However, a player who does so in their D+1Y will most likely go in rounds five through seven, if selected at all (one exception appears to be NCAA players – NCAA players that excel in their D+1 year sometimes go early – see Zach Parise, Mike Cammalleri and Paul Stastny). 

It is important to factor in age as at a certain point, especially in the CHL, it’s just too old to matter. The problem is there’s a middle area where the player is not too old yet (e.g., a Canadian playing in the CHL who’s a summer baby who is not even 19 by the time the D+1 draft has come and gone or a 21-year-old Russian putting up enormous numbers in the KHL, the second best league in the world).  

It’s easy to evaluate when the player is too old by looking at the history of overage selections. One example, regardless of league, if a player turns it on before age 19 (30+ NHLE) but not in their true draft year, their success rates and likely point production are nearly identical to players that do so in their draft year. Yet even players that are barely overagers are lower on the totem pole at the draft floor their second time around, teams preferring to take first-year eligible hulking wingers that registered few points.

But these players must look different from superstars drafted earlier in terms of point production, right?

Yes they do look different, for the most part, but not by much. Here I speak of how NHLE thresholds can be used to get a relatively accurate view of what type of scoring players will produce over the course of their NHL career. The big points that come out: nearly all elite impact players will hit a 30+ NHLE before turning 19; many, if they go back, will hit a 40+ NHLE by 20; and the absolute best of the best tend to have astronomically high NHLEs (50+) before entering the NHL.

If we look at these 100 elite late surprises here’s what they look like in terms of these factors:

  • Hit a NHLE of 30+ before turning 19 – 53 players (53%)
  • Hit a NHLE of 40+ before 20 – 45 players (45%)
  • Hit a NHLE of 50+ before turning 21 – 35 players (35%)
  • Hit all three of the thresholds above – 35 players (35%)
  • Hit a 40+ NHLE at some point before entering the NHL – 75 players (75%)
And comparing the 56 elite forwards drafted in the first two rounds between 2004 and 2014:

  • Hit a NHLE of 30+ before 19 – 37 players (66%)
  • Hit a NHLE of 40+ before 20 – 27 players (48%)
  • Hit a NHLE of 50+ before turning 21 – 9 players (16%)
  • Hit all three of the thresholds above – 8 players (14%)
  • Hit a 40+ NHLE at some point before entering the NHL – 31 players (55%)
There’s little difference. Late round surprises, if they’re not small, tend to hit the 30+ threshold a little later than elite talent that go in the first round, and they also tend to spend more time in their feeder league than the elite talent. Early drafted elite talent often make the NHL right after being drafted or a year after being drafted. You’ll be hard pressed to find a late round pick that does this, no matter how good they are (the late great Pavol Demitra is the only one of this group of 100).


Finding elite offensive talent is the most difficult thing to do at the draft and via trade, hence why extreme overpayment occurs on July 1 for the best offensive talent available. In the first round, teams appear to make a concerted effort to find offensive talent by taking the best scorers in every league that are first year draft eligible. However, in the later rounds teams seem to throw caution to the wind and start to make pick after pick on intangibles. 

How to improve one’s odds of finding elite talent late? It’s easy. Following three simple rules will probably increase a team’s chances of finding late elite talent by several hundred percent. 

  1. Extreme low scorer vs. smaller player with first round or second round worthy numbers… take the small player. 
  2. Extreme low scorer vs. 19-year-old CHL overager or early 20’s European that put up numbers comparable to the forwards picked in the top 10… pick the overager. 
  3. Extreme low scorer vs. 18-year-old Russian already playing in the KHL and scoring at a modest rate… pick the Russian! Even if the concern is that the player will never come over. Unless you’re positive he won’t come over (as in he has told you to his face he never will), especially if it’s late… take the gamble as they almost always come over when playing in the NHL is a real possibility. 
Bookmark and Share