An Analysis of Draft Classes

[greenOak]

So, I decided to look more specifically at the draft, and how draftees performed over the course of their career.

The tl;dr version is:

Draft classes are very top heavy, except with respect to goalies. All superstars or even top-six forwards / top-four defenseman come from rounds 1-2 (mostly round 1).

The long version is:

I looked at three different draft classes: the 2020-2021, 2021-2022, and 2022-2023. These years were chosen as they are the first draft classes composed entirely of CPU generated players (ie. those not found in the initial database). The 2015-2016 through to the 2019-2020 drafts seemed at first glance to be very deep with talent. Especially the 2015-2016 draft.

I checked up on these draft classes in 2028, where the players in each draft class are 26,25, and 24 respectively. A point in their careers, where they should theoretically be entering their prime. I checked up on them again in 5 years, when they have become established veterans, or even retired in some cases. I didn't want to take data for all 210 players, so instead I took the top 20 in games played in each draft class plus all players scouted at 4 stars or better. The raw data can be found in the document below:

https://docs.google.com/spreadsheets/d/1AKTzeaCVjaP3uI90pfphtuLZs7ma6nhY40tWy-QPt8Q/edit#gid=1123124997

Pos = Player position, DraftPos = position where drafted (overall), Rating = scouts rating (2028) in stars, GP = Games played, NHL PPG = points per game over NHL career, Peak PPG = highest PPG season (min 50 games), NHL GR = career game rating in NHL, Peak GR = highest game rating for a season (min 50 games for skaters, min 35 games for goalies), NHL Sv% = Career save percentage in the NHL, Peak Sv% = Highest save percentage in a season (min 35 games).

ANALYSIS

It appears as though one can find quality goaltenders quite deep into the draft in FHM2. As for skaters:

SKATERS ONLY

# of scouted stars / range of position drafted / average draft position
5, 1-7, 3.25 (4 Players)
4.5, 1-36, 11.75 (4 Players)
4, 1-33, 11.72 (22 players)

We can see immediately, that there isn't a single player rated 4 stars or that was picked after 36. In fact, there were only two players on this list picked in the 2nd round (pick 33, and pick 36). Perhaps interesting to note is that both players were Russian.

Now, scouts could be wrong, and there could be some very productive players who are rated poorly by scouts. Looking at NHL games played seems to support this theory.

Looking at players drafted in rounds 3-7, we see a number of interesting players:

Alfred Pruitt,Pick 114, LW, 604GP
Joel Sjorgen, Pick 138, LW, 364GP
David Redford, Pick 70, LD, 586GP
Adam Kurceba, Pick 86, LD, 417GP
Devin Miller, Pick 130, LW, 329GP
John Cline, Pick 88, LD, 368GP
Conor McAuley, Pick 62, RD, 208GP
Robert Tuer, Pick 65, LW 438 GP
Justin Elik, Pick 129, RW, 712GP
Zachary Bridges, Pick 90, LW, 387GP
Walker Claffey, Pick 132, LD, 220GP
Brayden Siebert, Pick 80, LW, 586GP

However, upon deeper inspection, it appears that none of these players were all that effective, using both PPG and GR as proxies for effectiveness. Joel Sjorgen, and Alfred Pruitt had the highest career PPG stats at 0.365 and 0.359 respectively. Joel Sjorgen had the highest NHL GR at 64. There were a few good individual seasons put up by these players, but these appeared to be the exception not the rule. To put things into further perspective:

Avg Draft Position of players with career PPG over 0.7: 5.7 (Range: 1-9, Median: 7)
Avg Draft Position of players with career PPG 0.6-0.7: 7.6 (Range: 1-29, Median: 4.5)
Avg Draft Position of players with career PPG 0.5-0.6: 12.1 (Range: 1-33, Median: 11.5)

This measure will be biased towards showing forwards. The top 5 scoring defenseman were drafted at 33,10,16,49,2. This spread is somewhat encouraging.

Avg Draft Position of player with career GR over 80: 5.7 (Range: 1-9, Median : 7)
Avg Draft Position of player with career GR between 75-80: 13.7 (Range: 1-33, Median: 7.5)
Avg Draft Position of player with career GR between 70-75: 11.8 (Range: 1-49, Median: 10)

For defenseman, the top 5 in terms of NHL GR were drafted at 2,33,10,16,49.

The last thing I want to show is an in depth look at the 2020-2021 draft where players Offensive Potential and Defensive Potential (the values in the player editor which are hidden when not in commissioner mode) relate to draft position. The raw data is found in the second sheet in the link previously posted.

A FEW FACTS:

-9 of the top 10 players went in the top 10.
- 27 of the top 30 players went in the top 30.
-A total of six players from rounds 3-7 were better than the worst player from round 1.

I would argue, that the real NHL is not nearly this top heavy when it comes to drafting.

[Mecza]

Appreciate the insight. I would like to throw this nugget your way regarding the top heaviness of draft classes:

Analyzing the value of NHL draft picks - Sportsnet.ca

Talent outside of a top 5 pick falls drastically. Anything outside the second round blurs together. It seems like the game is handling NHL talent relatively well at first glance. Though to be honest, I'm still struggling with the idea that a 3* player in the NHL is a bottom of the roster player. Relative to the other leagues it does make sense but it's still weird to my brain

[greenOak]

Tbh I disagree with your conclusion. It's true the NHL draft is a crap shoot after the first 20 picks or so, but in real life, you can still find top end talent from 20-210 (albeit rarely). In FHM2 the 20-40 picks have quite a bit of top end talent, while the 60-210 picks have absolutely none.

Don't think the Zetterberg's, Datsyuks, or St. Louis' are going to exist in FHM2.

[JeffR]

Originally Posted by greenOak

Tbh I disagree with your conclusion. It's true the NHL draft is a crap shoot after the first 20 picks or so, but in real life, you can still find top end talent from 20-210 (albeit rarely). In FHM2 the 20-40 picks have quite a bit of top end talent, while the 60-210 picks have absolutely none.

Don't think the Zetterberg's, Datsyuks, or St. Louis' are going to exist in FHM2.

No, it's not unpredictable to that point yet. Definitely something we want to improve on in the long run. First thing I'll probably try is making the amount by which a player can be underrated by scouts considerably larger. That should push the occasional potential star down into the lower rounds. But I want to avoid doing it in a way that'll make it obvious to a human player that's what's going on with the player.

[Fifabing]

Why not give players a range in their potential attributes? I hate to bring a rival developer and game into the lime light, but they give their young players a range for over potential in the future, which adds to randomness and allows lowly drafted players to potentially have a higher overall potential that goes unnoticed.

[glacius]

Well in my current game the AI finds some gems past 120 ~ picks , for example i had a guy that i really wanted with my 150 pick , but Anaheim snapped him at 148 and he had a good rookie season that made me trade for him giving 2 mid picks and a kerby rychel that had a good season for my nordiques.

Screenshot by Lightshot


Sean Montgomery was picked by Coyotes at 183 and he helped them win a Stanley Cup with an incredible postseason .


It does seem that mid to late 1st rounders tend to fail in my game i am in 2020/2021 season btw.

[josephyw]

We're in an online league that has made it through to 2022 and have noticed the draft classes are nearly 85-90% European skaters. That causes problems because Nordic players seem to often decide that they do not want to play in America, and a lot of the Russian players end up in the KHL or VHL (which doesn't allow us to sign players), and their contracts don't expire within the signing window. I recently had an entire draft class, minus one player, that I was unable to sign to ELCs due to this issue.

Any ideas as to how we can mitigate this? In 2016, the draftable prospect list on NHL.com shows 210 North American skaters and 133 European skaters.

Yet, after a few years, my current prospect pool--after drafting best available--looks as follows:

Czech Republic: 6
Finland: 2
France: 1
Russia: 8
Sweden: 9
Switzerland: 2
United States: 1

Is this other people's experience as well? Are there settings that would help us balance this out?

[coyotes_uk]

Originally Posted by josephyw

Yet, after a few years, my current prospect pool--after drafting best available--looks as follows:

Czech Republic: 6
Finland: 2
France: 1
Russia: 8
Sweden: 9
Switzerland: 2
United States: 1

Is this other people's experience as well? Are there settings that would help us balance this out?

After a few years of building up prospects I found the same, so I have just tried to pick more Canadian/US players in each round. There does seem to be NA options within each round but, a higher number of EU players strangely!

[pens66]

Same problem here.

[josephyw]

Originally Posted by josephyw

We're in an online league that has made it through to 2022 and have noticed the draft classes are nearly 85-90% European skaters. That causes problems because Nordic players seem to often decide that they do not want to play in America, and a lot of the Russian players end up in the KHL or VHL (which doesn't allow us to sign players), and their contracts don't expire within the signing window. I recently had an entire draft class, minus one player, that I was unable to sign to ELCs due to this issue.

Any ideas as to how we can mitigate this? In 2016, the draftable prospect list on NHL.com shows 210 North American skaters and 133 European skaters.

Yet, after a few years, my current prospect pool--after drafting best available--looks as follows:

Czech Republic: 6
Finland: 2
France: 1
Russia: 8
Sweden: 9
Switzerland: 2
United States: 1

Is this other people's experience as well? Are there settings that would help us balance this out?

Just knocked out another year and did the math for the 2023 draft: only 14% of draftable prospects were North American. Also, 100% of the prospects I drafted years prior that I have to sign or lose the rights to (mostly Swedish, with some Czech players) aren't "interested in playing in North America right now."

While we dedicate a lot of deserved time and attention to this game--and have a lot of fun doing so--this is becoming issue for us that requires a bunch of time and attention (having to craft the contracts then watch all players to make sure they don't get claimed on waivers) to manage.

Just to reiterate, NHL.com lists 239 North American prospects vs. 148 European prospects, which is 62% (a ratio that is fairly consistent over the last several years). Also, I could be wrong, but I can't think of any recent examples of high-profile European players (not including Russian) spurning the NHL at the time of their rights expiring, especially with the alarming frequency that it has been happening in this game.

I really hope this can be adjusted with the next patch and look forward to continuing with this game in the meanwhile!

[pens66]

I hope this gets fixed in the next (and last?) patch. Or the game would be in an eternal state of unplayableness (is that a word? ).

Stupid question maybe: But was this distribution of prospects in the game since it was released? I only ever played historical leagues at the beginning and only just recently started playing long term modern games. So I came across this only recently. Now that this mess with 2nd GMs also started to appear I've stopped playing completely. It's sad really.

[benjamin439]

Originally Posted by JeffR

No, it's not unpredictable to that point yet. Definitely something we want to improve on in the long run. First thing I'll probably try is making the amount by which a player can be underrated by scouts considerably larger. That should push the occasional potential star down into the lower rounds. But I want to avoid doing it in a way that'll make it obvious to a human player that's what's going on with the player.

In my opinion, I don't think potential should be a stat in this game or any simulation based game for that matter. It gives far to much information to the GM. In real life scouts use past performances to dictate future growth. They don't have the advantage of truly knowing a players ceiling. In FHM 2 and almost every other simulation game Potential starts hidden at it's true value and than gets revealed based on scouting. Sure bad scouting might get a 4.5 star potential to fall to 4 stars but it doesn't make them fall to 3. If a system makes the potential stat to unreliable than there really isn't a point in having that stat at all.

For instance in most drafts there are a handful of 3 star ability players. These players are frequently PPG players at star or franchise roles. In real life these players would be the most likely candidates in the first round. But because of the potential stat these players frequently get passes up for players on second and third lines who have higher potential. One 3 star ability player was a PPG for his past 2 years and was the cornerstone of his team. In real life this kid would have gone in the first round but because of his potential stat he didn't get drafted till the end of the draft. Sure he would have become a first round bust, but that frequently happens in real life.

Without the Potential stat, GMs would be forced to draft based on players ability scores and the additional information scouting brings. Late blooming players with lower ability scores should be available in later rounds, while high ability low potential players would get picked up in earlier rounds.

In real life every draft pick is a lottery, teams curb their odds by choosing players with the right personality and/or role that fits their team.

[JeffR]

Originally Posted by benjamin439

In my opinion, I don't think potential should be a stat in this game or any simulation based game for that matter. It gives far to much information to the GM. In real life scouts use past performances to dictate future growth. They don't have the advantage of truly knowing a players ceiling. In FHM 2 and almost every other simulation game Potential starts hidden at it's true value and than gets revealed based on scouting. Sure bad scouting might get a 4.5 star potential to fall to 4 stars but it doesn't make them fall to 3. If a system makes the potential stat to unreliable than there really isn't a point in having that stat at all.

For instance in most drafts there are a handful of 3 star ability players. These players are frequently PPG players at star or franchise roles. In real life these players would be the most likely candidates in the first round. But because of the potential stat these players frequently get passes up for players on second and third lines who have higher potential. One 3 star ability player was a PPG for his past 2 years and was the cornerstone of his team. In real life this kid would have gone in the first round but because of his potential stat he didn't get drafted till the end of the draft. Sure he would have become a first round bust, but that frequently happens in real life.

Without the Potential stat, GMs would be forced to draft based on players ability scores and the additional information scouting brings. Late blooming players with lower ability scores should be available in later rounds, while high ability low potential players would get picked up in earlier rounds.

In real life every draft pick is a lottery, teams curb their odds by choosing players with the right personality and/or role that fits their team.

One thing to remember about potential in FHM: it's absolutely not a hard-cap number the way it is in some other games. Only offensive and defensive skills have individual upper bounds for each player, and even then it's possible (to a limited degree) for the player to exceed those. Physical and mental attributes don't get a cap, other than making it somewhat difficult for them to wildly exceed all of the player's other ratings. The potential rating that's given in the game is a product of the potential caps on offensive and defensive skills, some projections based on the player's current age and level of mental and physical skills, and a factor that approximates the player's reputation amongst the scouting community (over- or under-rated), which can be adjusted by a variety of things, most notably the amount and quality of scouting attention the viewer has given the player. There's a lot of projection involved, and what the scout sees in the player's future may not match what the player's junior coach can get out of him right now.

I don't agree that junior performance correlates that strongly with a player's NHL prospects. That's just not the way it works in real life. If that were the case, Conor Garland, who ran away with the QMJHL scoring title last year and is doing the same this season, wouldn't have waited until the 5th round to get picked - after being passed over completely the year before and not even being on Central Scouting's list for most of the season. Other Q forwards with a third of his production but who project as better NHLers went earlier than he did. His offensive skills may make him an elite player in the company he's in right now, but when he moves up, his size and lack of defensive development are going to be very tough to overcome at the next level.