Theory about AI draft selections

[Cygnus_X1]

FHM 7 brought in a new mock draft feature. I love the idea of this in theory. It gives a larger sense of immersion in the draft experience. The idea of a mock draft giving teams a rough idea of where players are going to fall, within a general margin of error of course (we have seen plenty of draft day falls and gains before irl) is brilliant in theory. However, i have seen an issue in execution's. I don't know if this happens with anyone else, or if the AI in my game is just acting funky, so this is very much anecdotal evidence i am bringing here, but something is up. It seems like the AI does not consider the mock draft at all when it comes to making selections, rather just going by star potential rating. This is obviously logical in theory, select the player with the most potential with your high draft picks. my issue is that this completely eliminates any chances of hidden gems in the draft class. Case and point, Detroit has the third overall pick in my game. The player rated 3rd on the mock draft is a player named Paul Heber. He is a player with a high floor but low ceiling, but the game puts his star ability rating at the highest of anyone in the draft, 1.5 stars. Who does Detroit select 3rd? They end up going with a player named Brian Card. Now, this makes sense if you dont look at the mock draft. The guy has a star potential of 4 stars, which is great. he can be a very good player in the league. The issue is that he was ranked 123 in the mock draft. The potential fate of a player is not known on draft day. No GM in their right mind would select a player projected to go in the mid rounds with the 3rd overall pick, regardless of if that player becomes a star, because no one would have known that at the time. There is reaching for straws, and then there is reaching for thin air. This isnt the only example in the draft i just completed. Numerous players projected to go later on in the draft were selected with high value picks. This makes me think that the AI prioritizes potential over the mock draft WAY too much. There are obviously some off the board picks. Barrett Hayton in 2018 is an exact example. The issue there is Hayton was projected to go around the middle of round one when he was selected 5th. Card, in my game, was projected to go in the 4th but was selected with a lottery pick. As of now, I find the mock draft completely superfluous to the game because of the AI behavior. If changes were made to the AI drafting logic, it would be a complete benefit to the game. Not only would the game feel more realistic and give the mock draft meaning, but it would make hidden gems more prevalent. These are just random thoughts from a really tired guy who is burnt out from university mid terms, and this very well may not be an issue for other games, but I still find it peculiar to say the least. Thanks for reading,

[Jagorix]

Originally Posted by Cygnus_X1

FHM 7 brought in a new mock draft feature. I love the idea of this in theory. It gives a larger sense of immersion in the draft experience. The idea of a mock draft giving teams a rough idea of where players are going to fall, within a general margin of error of course (we have seen plenty of draft day falls and gains before irl) is brilliant in theory. However, i have seen an issue in execution's. I don't know if this happens with anyone else, or if the AI in my game is just acting funky, so this is very much anecdotal evidence i am bringing here, but something is up. It seems like the AI does not consider the mock draft at all when it comes to making selections, rather just going by star potential rating. This is obviously logical in theory, select the player with the most potential with your high draft picks. my issue is that this completely eliminates any chances of hidden gems in the draft class. Case and point, Detroit has the third overall pick in my game. The player rated 3rd on the mock draft is a player named Paul Heber. He is a player with a high floor but low ceiling, but the game puts his star ability rating at the highest of anyone in the draft, 1.5 stars. Who does Detroit select 3rd? They end up going with a player named Brian Card. Now, this makes sense if you dont look at the mock draft. The guy has a star potential of 4 stars, which is great. he can be a very good player in the league. The issue is that he was ranked 123 in the mock draft. The potential fate of a player is not known on draft day. No GM in their right mind would select a player projected to go in the mid rounds with the 3rd overall pick, regardless of if that player becomes a star, because no one would have known that at the time. There is reaching for straws, and then there is reaching for thin air. This isnt the only example in the draft i just completed. Numerous players projected to go later on in the draft were selected with high value picks. This makes me think that the AI prioritizes potential over the mock draft WAY too much. There are obviously some off the board picks. Barrett Hayton in 2018 is an exact example. The issue there is Hayton was projected to go around the middle of round one when he was selected 5th. Card, in my game, was projected to go in the 4th but was selected with a lottery pick. As of now, I find the mock draft completely superfluous to the game because of the AI behavior. If changes were made to the AI drafting logic, it would be a complete benefit to the game. Not only would the game feel more realistic and give the mock draft meaning, but it would make hidden gems more prevalent. These are just random thoughts from a really tired guy who is burnt out from university mid terms, and this very well may not be an issue for other games, but I still find it peculiar to say the least. Thanks for reading,

Sounds like the problem is that the mock draft isn't properly factoring potential into its rankings. It makes sense that GMs would value potential highly.

[Cygnus_X1]

Originally Posted by Jagorix

Sounds like the problem is that the mock draft isn't properly factoring potential into its rankings. It makes sense that GMs would value potential highly.

It does make sense, but my issue is that the AI reaching like this takes away from the realism of the game. We have seen players who where drafted earlier than expected and ended living up to that potential, but we have also seen many players from the late rounds develop into great players. One example of this is Johnny Gaudreau. He was selected early in the 4th round. That is a perfect example of a hidden gem, a diamond in the rough. We did not know at the time that he was going to develop the way he did. A lot of the time, the draft is really just luck of the draw. There is no guarantee players will develop. It takes time for players to grow and for spectators to see what they can truely become. Plus, you have to remember that the mock draft is supposed to show what position players are going to go at that time. Hindsight is 2020 on a lot of picks, Yakupov was projected near universally to go 1st in 2012. He did, and he was a bust. Connor Hellebyuck, who was selected round 5 in that same draft, just won a vezina last year. There lies my problem with the system. If you are going to have a mock draft, have it actually play a role on how the draft could realistically play out.

[KevT90]

Originally Posted by Cygnus_X1

[...] many players from the late rounds develop into great players. [...] A lot of the time, the draft is really just luck of the draw.

This !

Mock draft or not, I would love to see a scouting system/mechanic that makes the entry draft much more uncertain and enjoyable to take part in.

In reality, lots of 1st rounders become bust, many don't ever play a game in the NHL. On the other hand, many late rounders become great players, a few become superstars. Even undrafted players !

Just now, look at Kaprizov, he's the result of a 135th pick 5 years ago.

This would also give more value to late picks and avoid abusing the AI by bundling them together and trade for a higher draft pick. It would also take away of the sure value of 1st round picks and tanking abuse.