How Hockey ChangedBodies
Picks Six to Ten Got Safer. After Pick 30, Nothing Did.
In the 1985-2004 drafts, about three top-10 picks in four played 200 NHL games within seven seasons. Since 2005, nearly nine in ten have, close to the early-1980s drafts. The top five have made it about nine times in ten all along, and after pick 30 a team’s odds are no better than they were in 1969 (a little worse on the main count).
By Edgehalla ResearchPublished 9 min read
At the 1988 draft, the Quebec Nordiques used the fifth pick on Daniel Dore, an 18-year-old forward. He played 17 NHL games.
Five picks later, the Winnipeg Jets took Teemu Selanne, a 17-year-old from Finland. He did not play an NHL game until 1992-93, four seasons later.
In 2015, 27 years later, every one of the first ten picks in the draft made it in the NHL, from Connor McDavid at No. 1 to Mikko Rantanen at No. 10.
On this page
Top-five picks have made it about nine times in ten since 1969
So have NHL teams got better at drafting? Go back to 1988. Dore went fifth and played 17 NHL games. One pick later, Toronto took Scott Pearson, who played 262 in his first seven seasons. Pearson made it. Dore didn’t. Our line is 200 NHL regular-season games in a player’s first seven seasons, about two and a half full seasons. Every pick gets the same seven seasons. Mikko Rantanen, drafted in 2015, played 408 games in his. Doug Gilmour, drafted in 1982, played 456 in his, and that is the number we use, not his 1,474-game career.
Rounds can’t line up a 1969 pick with a 2019 pick, because the draft kept changing size. There were 84 picks in 1969 and 247 in 1974. Every draft of the late 1980s had 252, and 2000 had 293. In 2005 the draft was cut to seven rounds, and by 2019 it had 217 picks. A fourth-round pick meant something different in each. So we go by the overall pick. Gilmour went 134th, and he is compared with every other 134th pick, whatever round it fell in.
Dore’s miss was rare. Top-five picks made it about nine times in ten in the drafts from 1969 to 1984. They did the same from 1985 to 2004, and a little better since 2005. Their best decade was the 1980s drafts, at 95%. In the five drafts from 1980 to 1984, all 25 top-five picks made it.
Picks six to ten made the one clear jump, in the 2010s drafts
Share of picks who played 200 NHL games within 7 seasons, by overall pick, NHL drafts 1969-2019, grouped by draft era
Picks 1-5 hit 95% in the 1979-89 drafts and 92% in 2010-19, picks 31-60 24% and 24%; picks 6-10 were 64% to 73% in every era before the 2010s and 86% in 2010-19.
- 1969-78 drafts
- 1979-89 drafts
- 1990-99 drafts
- 2000-09 drafts
- 2010-19 drafts
Read across the bars and most of them wander from era to era. Picks 11 to 20 slid from the 1970s drafts to the 1990s and then climbed back. One bar jumps and stays up: picks six to ten, in the 2010s drafts.
So did teams get better at ordering the whole board, or only the very top? Look past the top 10 to the first 60 picks. Take any two of them, and usually the one taken earlier played more NHL games. We scored each draft from 0 to 1 on how closely the pick order matched games played. A draft scores 1 if every earlier pick out-played every later one. It scores 0 if the names might as well have come out of a hat. The 2014 draft came closest to perfect, at 0.73. The 2016 draft, the one where Vancouver took Olli Juolevi fifth, scored 0.39. Those two drafts, two years apart, sit 0.34 apart. The 1979-89 drafts averaged 0.51. The 2010-19 drafts averaged 0.57. One draft to the next swings more than that. In either decade, a typical draft lands about 0.1 above or below the decade’s average. So across the first 60 picks, teams order the board about as well as they did in the 1980s. What changed is five slots, six through ten.
Picks six to ten are where teams got better
Pearson, one pick after Dore, went in the slot where the odds moved. In every decade of drafts from the 1970s to the 2000s, picks six to ten made it between about 64% and 73% of the time. In the 2010s drafts they made it 86% of the time.
Split the drafts at 2005 instead, the year the draft shrank to seven rounds. Picks six to ten made it about six times in ten from 1985 to 2004. Since 2005 they have made it more than eight times in ten. Count the players: 58 of 97 picks six to ten made it in the earlier drafts, and 62 of 75 in the later ones. The earlier count is 97, not 100, because three of those picks went unsigned and were drafted again. Washington took Nick Boynton ninth in 1997, and Boston took him again at 21 in 1999. Lance Ward (New Jersey, 10th in 1996) and Mike Rupp (the Islanders, ninth in 1998) did the same. Each counts once, at his later pick. The top five moved only from 90% to 93% over the same years. So picks six to ten account for almost all of the top 10’s rise, from 75% to 88%.
Could the 2005-19 drafts just have been lucky? Flip a coin 200 times and you will rarely get exactly 100 heads. You will often land about 4 heads off in every 100. The 197 top-10 picks from 1985 to 2004 and the 150 since drift apart by luck the same way. The real gap is about 13 picks in every 100, roughly three times that drift. Coin flips almost never miss by that much. Cut the recent drafts in half and the rise is there in both. Top-10 picks made it 89% of the time in the 2005-11 drafts and 88% in 2012-19. Picks six to ten made it 80% of the time, then 85%.
The worst years came in the late 1980s. In 1988, Dore and Selanne were the two top-10 picks who missed. In 1989, four of the top ten missed. Vancouver took defenceman Jason Herter eighth. He played one NHL game, in 1995-96. Pool the 1985 to 1989 drafts and only about 70% of top-10 picks made it, the lowest of any five drafts in a row since 1969.
The misses kept coming through the 1990s. Washington took Alexander Volchkov fourth in 1996. He played three NHL games. The New York Rangers took Pavel Brendl fourth in 1999, and he played 78 in his first seven seasons.
Top-10 picks bottomed out in the late 1980s and have run near nine in ten since the late 2000s
Share of picks with 200+ NHL games within 7 seasons, by overall pick range, five-draft rolling average
Show data
| Year | Picks 1-10 | Picks 11-30 | Picks 31-100 | Picks 101-210 |
|---|---|---|---|---|
| 1971 | 78% | 53% | 11% | 4% |
| 1972 | 86% | 55% | 13% | 3% |
| 1973 | 80% | 51% | 15% | 3% |
| 1974 | 78% | 50% | 17% | 4% |
| 1975 | 80% | 49% | 19% | 4% |
| 1976 | 78% | 52% | 20% | 4% |
| 1977 | 80% | 58% | 23% | 6% |
| 1978 | 88% | 62% | 26% | 8% |
| 1979 | 94% | 60% | 24% | 9% |
| 1980 | 94% | 61% | 24% | 8% |
| 1981 | 96% | 59% | 25% | 8% |
| 1982 | 94% | 50% | 20% | 7% |
| 1983 | 88% | 43% | 15% | 6% |
| 1984 | 84% | 42% | 15% | 5% |
| 1985 | 80% | 39% | 13% | 5% |
| 1986 | 76% | 31% | 12% | 4% |
| 1987 | 70% | 31% | 12% | 4% |
| 1988 | 72% | 31% | 13% | 4% |
| 1989 | 72% | 34% | 14% | 5% |
| 1990 | 72% | 32% | 16% | 6% |
| 1991 | 76% | 40% | 16% | 7% |
| 1992 | 76% | 39% | 17% | 7% |
| 1993 | 78% | 41% | 17% | 7% |
| 1994 | 76% | 39% | 16% | 7% |
| 1995 | 79% | 38% | 14% | 7% |
| 1996 | 79% | 40% | 15% | 6% |
| 1997 | 79% | 39% | 14% | 6% |
| 1998 | 75% | 42% | 13% | 6% |
| 1999 | 73% | 44% | 14% | 6% |
| 2000 | 74% | 52% | 15% | 6% |
| 2001 | 74% | 54% | 16% | 6% |
| 2002 | 76% | 58% | 17% | 7% |
| 2003 | 78% | 55% | 17% | 7% |
| 2004 | 86% | 53% | 17% | 6% |
| 2005 | 84% | 47% | 15% | 6% |
| 2006 | 82% | 40% | 13% | 6% |
| 2007 | 86% | 39% | 13% | 6% |
| 2008 | 90% | 42% | 12% | 5% |
| 2009 | 90% | 41% | 14% | 5% |
| 2010 | 90% | 48% | 16% | 6% |
| 2011 | 92% | 50% | 18% | 5% |
| 2012 | 90% | 55% | 16% | 5% |
| 2013 | 92% | 57% | 17% | 6% |
| 2014 | 88% | 59% | 16% | 5% |
| 2015 | 88% | 56% | 14% | 4% |
| 2016 | 86% | 55% | 13% | 4% |
| 2017 | 88% | 50% | 15% | 4% |
- 2005: Draft cut to 7 rounds
Follow the top line. It peaks around the 1980 drafts, sinks through the late 1980s and 1990s, and climbs back after 2005. The lines for later picks barely move.
In 2015, all ten of the top ten made it
Since the late 2000s, about nine top-10 picks in ten have made it in every five-draft stretch. Then came 2015. Edmonton took Connor McDavid first and Buffalo took Jack Eichel second. All ten of the first ten picks played at least 273 NHL games in their first seven seasons: Connor McDavid, Jack Eichel, Dylan Strome, Mitch Marner, Noah Hanifin, Pavel Zacha, Ivan Provorov, Zach Werenski, Timo Meier and Mikko Rantanen.
A perfect top 10 was not new. It happened four times in the drafts from 1969 to 1984. It happened only twice in the 20 drafts from 1985 to 2004. Since 2005 it has happened four times, in 2006, 2011, 2013 and 2015.
Two of the 1988 top 10 fell short of 200 games; none of the 2015 top 10 did
| Pick | 1988 draft | Games, 7 seasons (career) | 2015 draft | Games, 7 seasons (career) |
|---|---|---|---|---|
| 1 | Mike Modano (MNS) | 423 (1,499) | Connor McDavid (EDM) | 487 (794) |
| 2 | Trevor Linden (VAN) | 529 (1,382) | Jack Eichel (BUF) | 409 (690) |
| 3 | Curtis Leschyshyn (QUE) | 434 (1,033) | Dylan Strome (ARI) | 273 (598) |
| 4 | Darrin Shannon (PIT) | 303 (506) | Mitch Marner (TOR) | 427 (738) |
| 5 | Daniel Dore (QUE) | 17 (17) | Noah Hanifin (CAR) | 517 (829) |
| 6 | Scott Pearson (TOR) | 262 (292) | Pavel Zacha (NJD) | 386 (706) |
Continues below
| Pick | 1988 draft | Games, 7 seasons (career) | 2015 draft | Games, 7 seasons (career) |
|---|---|---|---|---|
| 7 | Martin Gelinas (LAK) | 368 (1,273) | Ivan Provorov (PHI) | 450 (778) |
| 8 | Jeremy Roenick (CHI) | 458 (1,363) | Zach Werenski (CBJ) | 403 (642) |
| 9 | Rod Brind'Amour (STL) | 450 (1,484) | Timo Meier (SJS) | 394 (698) |
| 10 | Teemu Selanne (WIN) | 180 (1,451) | Mikko Rantanen (COL) | 408 (716) |
Read down the games columns. In 1988, Dore and Selanne sit under 200. In 2015 nobody does. The closest call was Dylan Strome, at 273.
The gain holds however you count it
Selanne shows the trouble with any line. He missed ours and still played 1,451 NHL games. Goalies get missed too, since a team plays one starter most nights. Some seasons were short, like the 48-game seasons in 1994-95 and 2012-13. So we checked the top-10 rise five more ways. We counted games as a share of each season’s schedule, so 24 games in the 48-game 1994-95 season count the same as 41 in an 82-game season. We followed players to age 25 instead of for seven seasons. We lowered the line to 100 games. And we left out goalies, then everyone but skaters born in Canada or the United States.
Every way we counted, top-10 picks since 2005 made it more often
Share of top-10 picks who made it under six definitions, NHL drafts 1969-84, 1985-2004 and 2005-19
Under every definition the 2005-19 top-10 picks hit more often than the 1985-2004 and 1969-84 ones; skaters only, 77% in 1985-2004 against 88% in 2005-19.
- 1969-84 drafts
- 1985-2004 drafts
- 2005-19 drafts
In every group of bars, the right-hand bar, the 2005-19 drafts, is the tallest, even with goalies left out or the line dropped to 100 games.
Goalies explain part of the old misses. Teams took 12 goalies in the top 10 from 1985 to 2004, and one since. Skaters alone still went from about three in four to nearly nine in ten. That gap is about two and a half times the drift that coin-flip luck usually opens, so it is very likely real. Selanne stayed in Finland for four seasons after his draft. Skaters born in Canada or the United States had no pro league outside the NHL to stay home in the way Selanne did, and their top-10 picks rose almost as much. Europeans who stayed home explain part of the rise at most.
We chose 1985 as the dividing year after we had seen that the late 1980s were weak. That choice could flatter the recent drafts. So we tried every other starting year from 1969 to 1999, always ending in 2004. The 2005-19 drafts beat every one of those spans. Out of every 100 top-10 picks, they put 7 to 15 more in the NHL. That is roughly one extra top-10 pick who made it in each draft. Lump every draft from 1969 to 2004 together and 79% of top-10 picks made it.
The drafts around 1980 were as good as today’s. Pool five drafts at a time, centred on 1979, 1980, 1981 or 1982, and 94% to 96% of top-10 picks made it, higher than any five-draft stretch since. In the 1980-84 drafts, picks six to ten made it 88% of the time, level with the 2010s. Then came Herter in 1989, Volchkov and Brendl in the 1990s, and a dip that lasted into the early 2000s. Since 2005 the top 10 has climbed back close to the early-1980s level. So the gain in our headline is measured against the Herter-to-Brendl years. Against all of the 1969-84 drafts, 84% then and 88% now, the gain could be luck.
Draft picks got younger, and the gain held
The 1970s drafts had a head start of their own. Their picks were older. A 20-year-old is closer to NHL-ready than an 18-year-old. So a 1970s draft could look better at picking when its players were simply further along. The first 30 picks in 1977 averaged 20.1 years old on draft day. By 1981 they averaged 18.4, and by 1983 18.3. The early-1980s drafts were already picking 18-year-olds, like today’s. Every draft from 1984 on has averaged between 18.2 and 18.6, except 1992 (19.1), a class with more older European picks.
The first 30 picks got nearly two years younger between 1977 and 1983, so the 1970s drafts had a head start
Average age on draft day, picks 1-30, NHL drafts 1969-2019
years
Show data
| Year | Average age, picks 1-30 |
|---|---|
| 1969 | 19.9 years |
| 1970 | 19.9 years |
| 1971 | 19.9 years |
| 1972 | 20.0 years |
| 1973 | 20.0 years |
| 1974 | 19.2 years |
| 1975 | 20.0 years |
| 1976 | 19.9 years |
| 1977 | 20.1 years |
| 1978 | 20.1 years |
| 1979 | 19.9 years |
| 1980 | 19.0 years |
| 1981 | 18.4 years |
| 1982 | 18.5 years |
| 1983 | 18.3 years |
| 1984 | 18.2 years |
| 1985 | 18.3 years |
| 1986 | 18.3 years |
| 1987 | 18.2 years |
| 1988 | 18.3 years |
| 1989 | 18.4 years |
| 1990 | 18.5 years |
| 1991 | 18.5 years |
| 1992 | 19.1 years |
| 1993 | 18.4 years |
| 1994 | 18.4 years |
| 1995 | 18.4 years |
| 1996 | 18.6 years |
| 1997 | 18.4 years |
| 1998 | 18.4 years |
| 1999 | 18.4 years |
| 2000 | 18.6 years |
| 2001 | 18.5 years |
| 2002 | 18.6 years |
| 2003 | 18.5 years |
| 2004 | 18.4 years |
| 2005 | 18.4 years |
| 2006 | 18.4 years |
| 2007 | 18.3 years |
| 2008 | 18.4 years |
| 2009 | 18.4 years |
| 2010 | 18.3 years |
| 2011 | 18.3 years |
| 2012 | 18.3 years |
| 2013 | 18.4 years |
| 2014 | 18.3 years |
| 2015 | 18.3 years |
| 2016 | 18.3 years |
| 2017 | 18.3 years |
| 2018 | 18.3 years |
| 2019 | 18.3 years |
The line drops off a ledge around 1980 and stays flat after. That head start may help explain why 1970s top-10 picks made it as often as they did. To take it away, follow every player to age 25, so a 1977 pick and a 2015 pick are judged at the same age. On that count, top-10 picks made it 91% of the time in the 2005-19 drafts, 83% in 1969-84 and 77% in 1985-2004, the same order as before. The early-1980s peak holds on that count too. Of the 1980-84 drafts’ top-10 picks, 94% made it by 25, more than any other five-draft group from 1970-74 to 2015-19.
Late first-rounders nearly all get an NHL game now, and about half stick
If scouts simply got sharper, the rest of the first round should have improved too. Picks 11 to 30 made it about four times in ten in the 1985-2004 drafts, and about half the time since 2005. That gain, between the two long spans, is small enough that it could be luck.
The bigger swings come from class to class. Inside the recent drafts, picks 11 to 30 made it 40% of the time in 2005-11 and 55% in 2012-19. That jump, between two sets of recent drafts, is bigger than luck usually makes.
Take them five drafts at a time. The 1975-79 drafts were the high: 58 of every 100 picks made it. The 1985-89 drafts were the low, at 31 in 100. The 2000-04 drafts rose to 58, the 2005-09 drafts fell back to 39, and the 2010-14 drafts climbed to 55. Across all the groups the average is about 47, and a typical group lands about 10 above or below it. If every class were equally deep, luck alone would move a group of 100 picks only about 5 either way. So the 1975-79 and 2010-14 classes really were deeper than the 1985-89 and 2005-09 ones. Deep and thin classes take turns. None of it climbs over time.
The 1988 draft shows how low a class can go. Of its picks from 11 to 30, only three of 20 made it, the fewest of any draft from 1969 to 2019.
Teams do give these players more chances now. Of the picks 11 to 30 from the 2012-19 drafts, 98% played at least one NHL game in their first seven seasons. From the 1985-2004 drafts, 87% did. Count the 2012-19 picks: 156 of 159 got a call-up, and 87 lasted 200 games. Across all the drafts since 2005, about half have lasted that long.
After pick 30, the odds are no better than in 1969
St. Louis found Gilmour at No. 134 in 1982. Sharper scouting should turn up more Gilmours deep in the draft. The odds there have gone slightly the wrong way. Picks 31 to 100 made it about one time in six in the drafts from 1969 to 1984, and about one in seven since. Picks after 100 have made it about one time in twenty in every era.
After pick 30, nothing improved and a little got worse
Picks 31-100 hit 18%, 15% and 14% across the 1969-84, 1985-2004 and 2005-19 drafts; picks 101-210 hit 6%, 6% and 5%.
Read the top row left to right: 18%, 15%, 14%. The bottom row: 6%, 6%, 5%.
The newest late picks may just be taking longer to arrive. New Jersey took Jesper Bratt 162nd in 2016, inside picks 151 to 210. Those slots made it 2.8% of the time in the 2010s drafts. The same slots made it 5.7% of the time in the 2000s drafts. Widen it to every pick from 101 to 210. The 2010s drafts put 4.1% of them past 200 games within seven seasons, and the 2000s drafts 6.0%.
Now follow those same picks 101 to 210 to age 25, and the order flips. The 2010s drafts reach 5.5%. The 2000s drafts reach 4.7%. Counting games as a share of each season’s schedule helps anyone whose window took in a short season, like the 56-game 2020-21. On that count the 2010s drafts still trail, 4.6% to 6.2%. Most of the 2010s gap looks like late arrival. A seven-season count can’t settle it, and some 2019 picks haven’t turned 25 yet. Recent picks 31 to 100 catch up the same way at 25. The 1969-84 drafts reach 17.4%, the 1985-2004 drafts 14.5% and the 2005-19 drafts 17.1%.
Late picks also face less competition than they used to. In 1985 there were 12.0 draft picks for every NHL team the next season: 252 picks and 21 teams. By 2019 there were 7.0: 217 picks and 31 teams. Fewer draftees were chasing each NHL roster spot, and the late odds still didn’t rise. Picks after 100 from the 2005-19 drafts still made it 4.6% of the time, against 6.0% for 1985-2004.
Every era had its top-five busts and its late finds
Late finds still happen, about as often as in 1969. Gilmour, the 134th pick in 1982, played 456 games in his first seven seasons and 1,474 in his career. Calgary took German Titov 252nd in 1993, and he played 487 in seven seasons. Montreal took Mark Streit 262nd in 2004. He played 443.
Vancouver and New Jersey both picked in 2016. Juolevi, the defenceman Vancouver took fifth, played 41 NHL games in seven seasons. New Jersey took Bratt 162nd. He played 389.
Every decade of drafts had a top-five pick who barely played and a late pick who played hundreds of games
| Draft era | Top-five picks with fewest games | Picks 101+ with most games |
|---|---|---|
| 1969-78 | Ray Martyniuk (1970, No. 5, 0 games); Bjorn Johansson (1976, No. 5, 15 games); Frank Spring (1969, No. 4, 35 games) | Larry Patey (1973, No. 130, 467 games); Garry Howatt (1972, No. 144, 449 games); Risto Siltanen (1978, No. 173, 431 games) |
| 1979-89 | Daniel Dore (1988, No. 5, 17 games); Neil Brady (1986, No. 3, 84 games); Dave Chyzowski (1989, No. 2, 118 games) | Gaetan Duchesne (1981, No. 152, 531 games); Doug Gilmour (1982, No. 134, 456 games); Gary Suter (1984, No. 180, 441 games) |
| 1990-99 | Alexander Volchkov (1996, No. 4, 3 games); Pavel Brendl (1999, No. 4, 78 games); Jason Bonsignore (1994, No. 4, 79 games) | Andreas Dackell (1996, No. 136, 553 games); Jaroslav Spacek (1998, No. 117, 496 games); German Titov (1993, No. 252, 487 games) |
Continues below
| Draft era | Top-five picks with fewest games | Picks 101+ with most games |
|---|---|---|
| 2000-09 | Thomas Hickey (2007, No. 4, 121 games); Alexander Svitov (2001, No. 3, 179 games); Stanislav Chistov (2001, No. 5, 196 games) | Lubomir Visnovsky (2000, No. 118, 499 games); Lee Stempniak (2003, No. 148, 456 games); Mark Streit (2004, No. 262, 443 games) |
| 2010-19 | Griffin Reinhart (2012, No. 4, 37 games); Olli Juolevi (2016, No. 5, 41 games); Michael Dal Colle (2014, No. 5, 111 games) | Andrew Shaw (2011, No. 139, 441 games); Jesper Bratt (2016, No. 162, 389 games); Viktor Arvidsson (2014, No. 112, 385 games) |
Pair the first row with the last. Montreal took Ray Martyniuk 5th in 1970, and he never played an NHL game. The Islanders took Griffin Reinhart 4th in 2012, and he played 37. The California Golden Seals found Larry Patey at No. 130 in 1973, and he played 467 games in seven seasons. Chicago found Andrew Shaw at No. 139 in 2011, and he played 441.
These numbers can’t say why picks six to ten got safer
Selanne made it too late for our test
Go back to 1988. A fifth pick like Dore’s made it about nine times in ten then, and still does. A pick in Selanne’s range, sixth to tenth, now makes it more than eight times in ten. A 134th pick like Gilmour’s still makes it about once in twenty, as it did in 1969.
Selanne himself still counts as a miss. He did not arrive until 1992-93, and he played 180 NHL games in his first seven seasons after the draft, short of 200. Then our seven seasons ran out.
He went on to play 1,451 NHL games.
Share this article
Draft and records tools
Live Edgehalla tools on the same subject.
Show the work
- Tests written down
- Not in advance (exploratory)
- Results computed
All 11,374 picks from the 1969-2019 NHL drafts (records.nhl.com), joined by player id to NHL regular-season games from the stats API, each measured over his first 7 NHL seasons (the cancelled 2004-05 season is skipped). The 2020 to 2026 drafts have not had seven seasons and are left out, and so are the 1963 to 1968 drafts (121 picks in six years, many of them 16-year-olds, before the universal draft of 1969). Players drafted twice count once, at their final selection (95 earlier picks dropped). Groups use overall pick ranges, not rounds. The top-10 change is checked in two halves of 2005-19 and under six definitions. Luck checks are two-proportion tests with unpooled standard errors (top 10, 1985-2004 vs 2005-19: 3.2 SE; picks 6-10: 3.5 SE; skaters only: 2.6 SE; picks 11-30: 1.6 SE across spans, 2.6 SE across the 2005-11 and 2012-19 halves; 1969-84 vs 2005-19: 0.9 SE; split-year check 1.8 to 3.2 SE). The picks 11-30 block spread is the standard deviation of ten five-draft blocks (10 points) against the binomial standard error of a 100-pick block (about 5), chi-square 32.6 on 9 degrees of freedom. Draft order is the Spearman rank correlation between pick number and games in seven seasons among the first 60 picks, averaged over draft years (SD across drafts 0.08 in 1979-89, 0.10 in 2010-19). Picks per team = picks ÷ teams in the next season’s league.
Continues below
The rolling top-10 rate pools five drafts centred on each year.
| Test | Result | Verdict |
|---|---|---|
| Top-five picks pan out more often now than in the 1980s | 95% (1979-89) against 92% (2010-19), n = 55 and 50; spans 90.0% / 90.0% / 93.3% | No change |
| Top-10 picks hit more often in 2005-19 than 1985-2004 | 75% (148 of 197) to 88% (132 of 150), 3.2 SE; both halves 89% and 88% | Holds |
| The top-10 gain came at picks 6-10 | Picks 6-10: 59.8% to 82.7% (3.5 SE); picks 1-5: 90.0% to 93.3%; by era 6-10 72 / 73 / 64 / 68 / 86% | Holds |
| Top-10 picks improved on the 1969-84 drafts | 84.4% to 88.0% (0.9 SE); picks 6-10 78.8% to 82.7% | Too small to call |
| Picks 11-30 improved | 40% then 55% inside 2005-19; five-draft blocks 31% to 58%, up and down (SD 10 points against about 5 from sampling; chi-square 32.6 on 9 df) | No trend |
More tests below
| Test | Result | Verdict |
|---|---|---|
| The 1988 draft had the worst picks 11-30 | 3 of 20; next lowest 1985 (5 of 20) | Holds (one class: a scene, not a trend) |
| Picks after 30 improved | 31-100: 18% / 15% / 14%; 101-210: 6% / 6% / 5% | Does not hold |
| A seven-season window can miss a star | Selanne (1988, No. 10): 180 games in the window, first NHL season 1992-93, 1,451 career | Known limit |
Frequently asked questions
What percentage of NHL first-round picks make it?
In the 2005-19 drafts, 88% of top-10 picks and 48% of picks 11-30 played 200 NHL games within seven seasons. 93% of picks 11-30 played at least one game.
Has NHL drafting gotten better over time?
Only near the top. Top-10 picks hit 88% of the time in 2005-19 against 75% in 1985-2004, and almost all of that came at picks six to ten (60% to 83%). Top-five picks were about 90% in each of the 1969-84, 1985-2004 and 2005-19 spans, and picks after 30 have not improved since 1969.
What are the odds a late-round NHL draft pick plays 200 games?
About 1 in 20. Picks 101-210 reached 200 games within seven seasons 4.6% of the time in the 2005-19 drafts and 5.7% in 1969-84.
Did every top-10 pick from the 2015 NHL draft make it?
Yes. All ten, from Connor McDavid to Mikko Rantanen, played at least 273 NHL games in their first seven seasons. Drafts since 2005 with a perfect top 10: 2006, 2011, 2013, 2015.
How do you measure an NHL draft pick’s success fairly across eras?
Use a fixed window after the draft (here, the first seven NHL seasons) so recent drafts are not judged on unfinished careers, and compare overall pick ranges rather than rounds, because the draft has had anywhere from 6 to 25 rounds. We also checked a schedule-adjusted version and an age-25 window.
More in Bodies
- Verdict: Data story
Nobody Drafted Martin St. Louis
- Verdict: Data story
Major Junior Lost Half Its Share of the NHL Draft
- Verdict: Data story
The 1,000-Game Club Had Its Biggest Year. Getting In Is No Easier.
- Verdict: Data story
How Much Lighter Did the NHL Really Get?
- Verdict: Data story
The NHL Is Getting Older Again
- Verdict: Data story
Connor McDavid Got Faster. The League, Not So Much.
Sources and method
- NHL Records: draft history API, every pick since 1963 (2019 draft shown)
- NHL stats (NHL.com), every season since 1917-18
- NHL stats API: skater and goalie season summaries (games played by player)
- World Hockey Association (1972-79)
Research and analysis, not betting advice. Past results do not guarantee future returns; bet only what you can afford to lose.