Feedback on the simulation

[Schlitz]

Scoring definitely seems to be down since the update. I've played 6 games and only one team managed 5 goals in that span. The highest scoring game I have seen was a 3-4 OT loss to the Red Wings and I even got my very first shutout 2-0 over the Blue Jackets after 49 games. The downside is that with lower scoring my Panthers are really having a tough time of it now.

[saberhagen83]

I'm starting to wonder, does 2-0 leads after 1st = loss? Time and time again I see this happen both to me and to the AI. I can take no pleasure being up 2-0 after the first cause I know if we don't get that 3rd goal it 90% guarantee we'll loose cause so few games end 2-0 or 2-1, usually ends up either 2-4 or a late consolation goal making it 3-4. It's almost too predictable.

On the plus side, after the patch these wild west games of 13-1 or 9-2 seem to be a thing of the past finally. Even though I still see them all the time when I see scores flying by while simming, but in the NHL I see mostly 6-4 as the biggest result, with 5-3 and 6-2 being the usuall scores from high scoring games...maybe still a bit too much but at least better.

[elrune1988]

Backes has been my defensive stud this season. No respect on this team for that kind of game!

[krownroyal83]

I'm starting to wonder, does 2-0 leads after 1st = loss? Time and time again I see this happen both to me and to the AI. I can take no pleasure being up 2-0 after the first cause I know if we don't get that 3rd goal it 90% guarantee we'll loose cause so few games end 2-0 or 2-1, usually ends up either 2-4 or a late consolation goal making it 3-4. It's almost too predictable.

I've just looked at some of the box scores of the games and im not seeing what you're seeing. Teams with a 2 goal lead after one period won pretty much every game.

[batdad]

Yeah, the likelihood is your players are tiring out. Using too much of the same guys. My bet is your ice times are not balanced enough. Or you are just every team the Calgary Flames and Vancouver Canucks played all season long. That will be an interesting series. Who is gonna score first?

[CeeBee]

Yeah, the likelihood is your players are tiring out. Using too much of the same guys. My bet is your ice times are not balanced enough. Or you are just every team the Calgary Flames and Vancouver Canucks played all season long. That will be an interesting series. Who is gonna score first?

LOL they will both be giving the puck away and hoping the other team scores first so they can get going.

[CeeBee]

The types of injuries are still pretty weird. I've had more fractured shins than anything else and to be honest I'm not sure I've ever heard of that IRL. Should be more groin and muscle pulls and tears and more broken fingers, hands and feet. A lot less of the rare and truly weird ones would make the game more realistic.

[pjfoster13]

Since the 15.0.1b update:
1) goalie statistics/success have been scaled a bit too far, a sample season shows 11 NHL goalies at .920 or better and 18 goalies at 2.40 or better. Their adjustment worked but whatever change was made should be re-adjusted again so that it hits somewhere in the middle of where it started and where it is now.
2) nowhere near enough volume of hits for defensemen, contributing to comparatively-low overall ratings. In a sample season, the top 24 in hits range from 60 to 86, whereas in real-life 2015 the top 24 range from 170 to 306. The SIM's hit-rate is somehow mis-calibrated by almost a full factor of three (300%) which is very drastic.
3) by and large, sim favors small, offensive defensemen ahead of more well-rounded and/or defensive-minded (in terms of average rating). The John Klingbergs of the world flourish while the Cody Fransons tend to flounder
4) AI has flawed logic pertaining to who does and does not constitute a worthy NHL-er. Abnormally-high amounts of sub-optimal players on computer-controlled teams, relative to the free agent pool, the european pool, and sometimes even its own depth chart. Team selections contain inaccuracies

5) A high percentage of players (especially in drafts 2016-2025) have polarized attribute sets. In other words, players who have a series of 18-20s and a series of 6-8s. Must have more realistic player-type templates and a development model favoring well-roundedness. Also, those draft classes looked very, very, unrealistically weak.

SIM logic must solve for ways to improve bravery, determination, influence, and work rate. Some suggestions might include: similar to FM, player mentoring could be a way to mutually improve determination, influence could perhaps simply go up naturally over time by being a veteran, being a captain or assistant, mentoring, going deep in playoff series, etc. Bravery could maybe somehow be tied to injury recovery, like how in real life guys get hit in the mouth with pucks and dont miss a shift, things like that, I don't know if there's a way to code for that kind of thing but it's an idea.

6) AI has flawed logic when it gets into a cap crunch situation. for example when the Blackhawks' inevitable crunch happened, it didn't know how to identify the "bad contracts", it started by waiving the cheap guys like Brandon Saad and Andrew Shaw and then the most important players like Kane and Keith instead of starting with older, less important players such as Hossa, Sharp, Crawford, etc.

7) as previously mentioned by other comments, progression of physical attributes is extremely top-heavy, especially when compared to hockey sense and intangibles which develop very little. This tended to be a blind spot in EHM 2007 also. Ever since FM 2008, the FM model has tended to restrict physical attributes in a very specific way so that 16s and 17s are very very good and only the very best-of-the-best get 20s, very rare.

The overall player pool is incredibly confusing in this respect. When I search my current NHL player pool, amongst defensemen there are 19 players with 20 speed, yet there are 0 players with 20 anticipation, 0 players with 19, 1 player with 18, and 4 players with 17. Amongst forwards there are three and a half full pages of 20 speeds, yet 0 players with 20 anticipation, three with 19, six with 18, and five with 17.

Not sure if there's a way for SI to re-distribute the way that p.a. is portioned so that skill and mental get a higher percentage of the development as opposed to filtering so heavily into physical

[Job]

Since the 15.0.1b update:
1) goalie statistics/success have been scaled a bit too far, a sample season shows 11 NHL goalies at .920 or better and 18 goalies at 2.40 or better. Their adjustment worked but whatever change was made should be re-adjusted again so that it hits somewhere in the middle of where it started and where it is now.

stats from NHL.com tell me 15 starting goaltenders have a SV% of .920 or better and 18 goalies have a GAA of 2.40 or better. I'm not quite sure what you think is wrong with the stats you got. Unless I misunderstood you.

[Peter_Doherty]

I mentioned earlier (in another thread?) the problem with the physical attributes being way too high later on in the game, there are 75 Hagelins in my game and about 40 Charas (str & balance)

They either need to change the way physical attributes develop or we need to set them down in the DB at the start...

[nickflyers]

Since the 15.0.1b update:
1) goalie statistics/success have been scaled a bit too far, a sample season shows 11 NHL goalies at .920 or better and 18 goalies at 2.40 or better. Their adjustment worked but whatever change was made should be re-adjusted again so that it hits somewhere in the middle of where it started and where it is now.
2) nowhere near enough volume of hits for defensemen, contributing to comparatively-low overall ratings. In a sample season, the top 24 in hits range from 60 to 86, whereas in real-life 2015 the top 24 range from 170 to 306. The SIM's hit-rate is somehow mis-calibrated by almost a full factor of three (300%) which is very drastic.
3) by and large, sim favors small, offensive defensemen ahead of more well-rounded and/or defensive-minded (in terms of average rating). The John Klingbergs of the world flourish while the Cody Fransons tend to flounder
4) AI has flawed logic pertaining to who does and does not constitute a worthy NHL-er. Abnormally-high amounts of sub-optimal players on computer-controlled teams, relative to the free agent pool, the european pool, and sometimes even its own depth chart. Team selections contain inaccuracies

5) A high percentage of players (especially in drafts 2016-2025) have polarized attribute sets. In other words, players who have a series of 18-20s and a series of 6-8s. Must have more realistic player-type templates and a development model favoring well-roundedness. Also, those draft classes looked very, very, unrealistically weak.

SIM logic must solve for ways to improve bravery, determination, influence, and work rate. Some suggestions might include: similar to FM, player mentoring could be a way to mutually improve determination, influence could perhaps simply go up naturally over time by being a veteran, being a captain or assistant, mentoring, going deep in playoff series, etc. Bravery could maybe somehow be tied to injury recovery, like how in real life guys get hit in the mouth with pucks and dont miss a shift, things like that, I don't know if there's a way to code for that kind of thing but it's an idea.

6) AI has flawed logic when it gets into a cap crunch situation. for example when the Blackhawks' inevitable crunch happened, it didn't know how to identify the "bad contracts", it started by waiving the cheap guys like Brandon Saad and Andrew Shaw and then the most important players like Kane and Keith instead of starting with older, less important players such as Hossa, Sharp, Crawford, etc.

7) as previously mentioned by other comments, progression of physical attributes is extremely top-heavy, especially when compared to hockey sense and intangibles which develop very little. This tended to be a blind spot in EHM 2007 also. Ever since FM 2008, the FM model has tended to restrict physical attributes in a very specific way so that 16s and 17s are very very good and only the very best-of-the-best get 20s, very rare.

The overall player pool is incredibly confusing in this respect. When I search my current NHL player pool, amongst defensemen there are 19 players with 20 speed, yet there are 0 players with 20 anticipation, 0 players with 19, 1 player with 18, and 4 players with 17. Amongst forwards there are three and a half full pages of 20 speeds, yet 0 players with 20 anticipation, three with 19, six with 18, and five with 17.

Not sure if there's a way for SI to re-distribute the way that p.a. is portioned so that skill and mental get a higher percentage of the development as opposed to filtering so heavily into physical

Agreed. Mental stats are wonky in the later years. Have you posted this over at SI?

[paul4587]

Since the 15.0.1b update:
1) goalie statistics/success have been scaled a bit too far, a sample season shows 11 NHL goalies at .920 or better and 18 goalies at 2.40 or better. Their adjustment worked but whatever change was made should be re-adjusted again so that it hits somewhere in the middle of where it started and where it is now.
2) nowhere near enough volume of hits for defensemen, contributing to comparatively-low overall ratings. In a sample season, the top 24 in hits range from 60 to 86, whereas in real-life 2015 the top 24 range from 170 to 306. The SIM's hit-rate is somehow mis-calibrated by almost a full factor of three (300%) which is very drastic.
3) by and large, sim favors small, offensive defensemen ahead of more well-rounded and/or defensive-minded (in terms of average rating). The John Klingbergs of the world flourish while the Cody Fransons tend to flounder
4) AI has flawed logic pertaining to who does and does not constitute a worthy NHL-er. Abnormally-high amounts of sub-optimal players on computer-controlled teams, relative to the free agent pool, the european pool, and sometimes even its own depth chart. Team selections contain inaccuracies

5) A high percentage of players (especially in drafts 2016-2025) have polarized attribute sets. In other words, players who have a series of 18-20s and a series of 6-8s. Must have more realistic player-type templates and a development model favoring well-roundedness. Also, those draft classes looked very, very, unrealistically weak.

SIM logic must solve for ways to improve bravery, determination, influence, and work rate. Some suggestions might include: similar to FM, player mentoring could be a way to mutually improve determination, influence could perhaps simply go up naturally over time by being a veteran, being a captain or assistant, mentoring, going deep in playoff series, etc. Bravery could maybe somehow be tied to injury recovery, like how in real life guys get hit in the mouth with pucks and dont miss a shift, things like that, I don't know if there's a way to code for that kind of thing but it's an idea.

6) AI has flawed logic when it gets into a cap crunch situation. for example when the Blackhawks' inevitable crunch happened, it didn't know how to identify the "bad contracts", it started by waiving the cheap guys like Brandon Saad and Andrew Shaw and then the most important players like Kane and Keith instead of starting with older, less important players such as Hossa, Sharp, Crawford, etc.

7) as previously mentioned by other comments, progression of physical attributes is extremely top-heavy, especially when compared to hockey sense and intangibles which develop very little. This tended to be a blind spot in EHM 2007 also. Ever since FM 2008, the FM model has tended to restrict physical attributes in a very specific way so that 16s and 17s are very very good and only the very best-of-the-best get 20s, very rare.

The overall player pool is incredibly confusing in this respect. When I search my current NHL player pool, amongst defensemen there are 19 players with 20 speed, yet there are 0 players with 20 anticipation, 0 players with 19, 1 player with 18, and 4 players with 17. Amongst forwards there are three and a half full pages of 20 speeds, yet 0 players with 20 anticipation, three with 19, six with 18, and five with 17.

Not sure if there's a way for SI to re-distribute the way that p.a. is portioned so that skill and mental get a higher percentage of the development as opposed to filtering so heavily into physical

This is a great post. I agree with everything you've listed. I'm glad I'm not the only one who has the bigger, more defensive defensemen issue. In my 2007 DB guys like Pronger and Chara are very mediocre compared to the likes of Timonen and Visnovsky who tear the league apart.

[Alessandro]

pjfoster you should post this on the SI forums

[Alessandro]

In my save Grigorenko signed in the Russian third league in 2016

[Peter_Doherty]

Too many coaches get fired in NHL, i think it's mostly because every team think they should make the playoffs, should set realistic expectations for every team...

[trevok]

I've noticed in the simulation players score from the other end of the ice too much.

This happens a lot when the goalie is pulled, which isn't so bad, but twice I've had my players score with a shot taken from just in front of my own net. I don't think the goalie on the other team was Vesa Toskala either.

I could see this happening once as a fluke, but twice in 40 games is a problem.

[pjfoster13]

pjfoster you should post this on the SI forums

done. i think they're off to a good start but they just need to smooth some things out, and if it's possible i definitely hope they can borrow some of the FM code to set some new game triggers

[Koekenbakker]

Too many coaches get fired in NHL, i think it's mostly because every team think they should make the playoffs, should set realistic expectations for every team...

This!

[Duranium]

...that much that for the match engine - pretty normal tactical settings and this is how my penalties are looking almost every game - at 17:47 is my favorite piece, 2 penalties in one second, this also happens EVERY game

P.S. Also trying out Batdads conservative approach on settings for d-men - works great, game ratings are usually 5 or 6 for the whole defense

[bobmcgoo]

i simmed eight years of EHM:EA v15.1b with the TBL 7.3 rosters, against the last seven full NHL seasons. for anyone who cares, here are the major stat categories comparisons: http://s18.photobucket.com/user/bobmcgo ... t=3&page=1

in short, they've aced it. everything is very realistic except for what's been noted already in this thread: 1) defensemen shoot too much and forwards shoot too little, which means the total team shooting is a bit low overall, 2) the lower end goalies are not as good on the game as they are IRL, and 3) PPO are too high. i'm hoping all of these would be reasonably easy to change within the gamecode/rosters (if necessary), and they're not close to gamebreaking differences.

this is only from the standpoint of simming one game for stats purposes though, and it's only NHL stats as well. but overall, i'm seriously impressed.

[bobmcgoo]

- Goalies play way too many games each season and are not being replaced with poor seasonal performances. I find it alarming that after 6 seasons no new goalies have taken over the number 1 position anywhere.

this is a big problem. the game still relies too much on reputation rather than how well a player is actually playing. just to test it i introduced 8 20-year-old fictional goalies who all had 200 as their PA and various CAs between 140 and 170, but let the game generate their three reputation "attributes". after about six seasons only one(!) of them had the starter's job on his team despite that fact that they were all completely outplaying their rival goaltenders and had been doing so for years. some of them were still stuck in the minors! only when i went back and set their reputations to numbers to match their CAs did the game start picking them regularly. either reputation needs to be more flexible so it changes more frequently based on performance, or teams need to choose their rosters based on avr/form rather than reputation.

ASG roster selection is borked.

Players selected in Season 2:
Ryan Clowe (17 points in 43 games)
Semyon Varlamov (3.11 GAA, .903Sv%)
Andrew Ladd (28 points in 45 games)

Players NOT selected:
Gustav Nyquist (63 points in 50 games)
Tyler Toffoli (57 points in 48 games)
Ben Bishop (2.16GAA, .921 Sv%, best stats in league for goalie)

Since the 15.0.1b update:
4) AI has flawed logic pertaining to who does and does not constitute a worthy NHL-er. Abnormally-high amounts of sub-optimal players on computer-controlled teams, relative to the free agent pool, the european pool, and sometimes even its own depth chart. Team selections contain inaccuracies

probably both explainable by the reason above: reputation over performance.

Since the 15.0.1b update:
5) A high percentage of players (especially in drafts 2016-2025) have polarized attribute sets. In other words, players who have a series of 18-20s and a series of 6-8s. Must have more realistic player-type templates and a development model favoring well-roundedness. Also, those draft classes looked very, very, unrealistically weak.

i have noticed this too, the same as 2007 IIRC.

What i have noticed is that scores in the playoffs are alot lower.. .whether thats due to the better teams playing each other or another factor im not sure.

this was certainly the case with 2007 and it still is true by the looks of things.

and finally

Quicksim gives really different results. You cant compare quicksim simulated stats and stats what you get when you actually managing team with fullsim mode

+1... this is very important!

[Manimal]

Good stuff, bobmcgoo!

[Peter_Doherty]

Polarized attributes should get better when Player Roles get implemented i would imagine...

[batdad]

Awesome Mr. McGoo. CAn always be sure you will have good numbers for us. Thanks for posting that, so I can just refer to it instead of finding and posting the #s myself.

[SirMichaelJordan]

Hits & injuries are WAY off, any analysis on that?