Mafarfloune Posted April 9, 2015 Share Posted April 9, 2015 NOTE : I wasn't really sure where to put this. It's certainly not a question. And it's not exactly a bug. It's more of a pointing out something that doesn't seem quite right. Anyway. I ran a little test over at TBL and I've been encouraged to post the results here also. What I wanted to see in that test was just how different the simulation was handled when setting the Detail level to full detail (all games), and when setting it to Quick Sim. Test configuration - EHM:EA version 15.0.2b - TBL roster database - NHL and AHL as only active leagues. - No human player in control of a team - Vacation until April 12. - Ran test once with detail level set at all games (NHL) - Ran test once with detail level set at Quick Sim. Here are the results, as posted in the TBL forums. Team Standings and stats Enhanced Quicksim Analysis : Scoring seems to be higher by about 25 Goals/Team on enhanced. PP% way too low on quicksim. Scoring Leaders Enhanced Quicksim Analysis : Scoring leaders were way more 2015-ish in the quicksim test. Much fewer PPG players. Penalty minutes Enhanced Quicksim Analysis : In both cases, the leaderboard was filled with forwards, which is probably not right. The number of penalties seems a little bit more reasonable in quicksim mode (though still a bit high). +/- Enhanced Quicksim Analysis : For some reason, the leaderboard was almost exclusively filled with forwards on Enhanced, and mostly filled with D-men on quicksim. Leaders have quite high numbers in both cases (worse on quicksim). Shooting Enhanced Quicksim Analysis : Outside of Ovechkin, the leaderboard is exclusively D-men on Enhanced. In Quicksim, the leaderboard is filled entirely with forwards. Huge discrepancy. In both cases, a bit high when compared to actual numbers in the NHL (except for the Ovechkin outlier). Icetime (D-men) Enhanced Quicksim Analysis : Enhanced mode edges closer to reality, though it lacks the couple of outliers with huge ATOI numbers. Both modes seem to distribute icetime way differently, with quicksim giving about 3 minutes more per game to the top pairing. Faceoffs Enhanced Quicksim Analysis : Top centers FO% is WAY too high on enhanced. Quicksim seems to handle it better. Image limit reached, continued in next post... Link to post Share on other sites More sharing options...
Mafarfloune Posted April 9, 2015 Author Share Posted April 9, 2015 Edit : See Archi's post. Link to post Share on other sites More sharing options...
Archibalduk Posted April 9, 2015 Share Posted April 9, 2015 I know Job is having difficulty posting his thead, so I thought I'd copy it over from his original thread over on The Blue Line: http://www.ehmtheblueline.com/forums/viewtopic.php?f=3&t=13844&p=184842#p184636 Here is his post: So I've gone ahead and elected to sim the 2015 season twice : Once with the simulation set as enhanced, once set as quicksimming. Here are the results! Team Standings and stats Enhanced Quicksim Analysis : Scoring seems to be higher by about 25 Goals/Team on enhanced. PP% way too low on quicksim. Scoring Leaders Enhanced Quicksim Analysis : Scoring leaders were way more 2015-ish in the quicksim test. Much fewer PPG players. Penalty minutes Enhanced Quicksim Analysis : In both cases, the leaderboard was filled with forwards, which is probably not right. The number of penalties seems a little bit more reasonable in quicksim mode (though still a bit high). +/- Enhanced Quicksim Analysis : For some reason, the leaderboard was almost exclusively filled with forwards on Enhanced, and mostly filled with D-men on quicksim. Leaders have quite high numbers in both cases (worse on quicksim). Shooting Enhanced Quicksim Analysis : Outside of Ovechkin, the leaderboard is exclusively D-men on Enhanced. In Quicksim, the leaderboard is filled entirely with forwards. Huge discrepancy. In both cases, a bit high when compared to actual numbers in the NHL (except for the Ovechkin outlier). Icetime (D-men) Enhanced Quicksim Analysis : Enhanced mode edges closer to reality, though it lacks the couple of outliers with huge ATOI numbers. Both modes seem to distribute icetime way differently, with quicksim giving about 3 minutes more per game to the top pairing. Faceoffs Enhanced Quicksim Analysis : Top centers FO% is WAY too high on enhanced. Quicksim seems to handle it better. Link to post Share on other sites More sharing options...
Archibalduk Posted April 9, 2015 Share Posted April 9, 2015 And here is the second part of his post: Hitting Enhanced Quicksim Analysis : Mostly D-men on Quicksim, Only forwards on Enhanced. Both are off (IRL, in the top24, 17 are forwards). Numbers are a bit low on quicksim, and absurdly low on enhanced. (see hits leaders Giveaways/Takeaways Enhanced Quicksim Analysis : For giveaways, Enhanced have the numbers about right, but all the leaders are D-men, whereas in real life, it's about 50/50. Quicksim shows only forwards except for one players, but the numbers are way too low. For takeaways, numbers are about 30% too high on enhanced, and again heavily favors d-men, whereas in real life the leaderboard is filled with forwards. Quicksim is being downright ridiculous. (compare with Giveaways and Takeaways) Blocked Shots Enhanced Quicksim Analysis : Way too low in both cases. Discrepancy between best and average less ridiculous in Enhanced. However, Quicksim does have it right in making all leaders D-men. Average rating Enhanced Quicksim Analysis : Goalies and Forwards are the highest rated on enhanced (by far). D-men are the highest rated on quicksim (by not quite as far). Also of interest, a quick look tells me forwards and d-men with good defensive abilities are favored in quicksim, whereas Enhanced seems to favor players with high offensive abilities. Goaltending Enhanced Quicksim Analysis : Goalie stats on enhanced seem closer to reality on Enhanced (SV% being too low on Quicksim), which is weird considering more goals were scored on enhanced. I suspect lower-tier goalies are terrible on enhanced and somewhat better on quicksim. Also, notice the huge discrepancy between the Average rating of goalies in both simming modes. Link to post Share on other sites More sharing options...
Archibalduk Posted April 9, 2015 Share Posted April 9, 2015 And the third and final part: Some further pictures, because I had forgotten about a couple of important things. D-Men Scoring Enhanced Quicksim Analysis : A LOT more points from defencemen when simming in Full Detail as opposed to Quicksim. Especially more goals scored by D-men, too. (This probably correlates with our shooting stats.) Also notice that the average rating mean is about the same for both modes, but the highest rating are WAY higher in Quicksim. Team form Enhanced Quicksim Analysis : I chose the Canucks because they finished with the same number of points in both tests. Quicksim, as a whole, seems to be a lot more generous with its average rating (except for goalies), and that's a trend for most (every?) teams. Most players are over 7.00 in Quicksim. Most players are under 7.00 (With a lot of them hovering around 6.50) in Full Detail mode. For most teams, the mean seems to drop from 7.15 AVR to 6.80 AVR when switching from Quicksim to Full Detail, with the D-men suffering the most. The curious case of Zdeno Chara Enhanced Quicksim Analysis : In quicksim mode, Chara gets a crazy amount of hits, takeaways and blocked shots that he doesn't get in Enhanced simulation. That translates in a huge rating discrepancy, Chara transforming from the best rated player on the ice in Quicksim (8.84) rating, to a very average rated player in Full Detail mode (6.94 AVR). Also notice the huge amounts of shots he has in enhanced, which he doesn't have in Quicksim, and the fewer number of giveaways in quicksim. Link to post Share on other sites More sharing options...
Mafarfloune Posted April 9, 2015 Author Share Posted April 9, 2015 Thanks for your help. I guess the board just wouldn't let a guy who just signed up make a post with 30+ images in it... Test configuration - EHM:EA version 15.0.2b - TBL roster database - NHL and AHL as only active leagues. - No human player in control of a team - Vacation until April 12. - Ran test once with detail level set at all games (NHL) - Ran test once with detail level set at Quick Sim. Link to post Share on other sites More sharing options...
DrewsT Posted April 9, 2015 Share Posted April 9, 2015 Great work testing this! It appears the game is much more what were looking for when not in enhanced mode (lower scoring, more defensive, realistic numbers) Not sure the difficulty in finding what in code can cause this but if they can find a way to make enhanced play like the quick sim I think that would go a long way into helping the overall realism fans look for in a sim. Also Archie if you get a chance is it possible to edit the post lol lots of pics. Link to post Share on other sites More sharing options...
kidhander1983 Posted April 10, 2015 Share Posted April 10, 2015 Some good analysis there, I hope the devs will have some food for thought Link to post Share on other sites More sharing options...
Konnan51 Posted April 10, 2015 Share Posted April 10, 2015 Takeaways and hits seem to be the biggest contributing factor in ratings. Seems to be some sort of ratio differences in enhanced and quicksim that are causing DMen to be OP when quicksimming. Link to post Share on other sites More sharing options...
Mafarfloune Posted April 10, 2015 Author Share Posted April 10, 2015 It appears the game is much more what were looking for when not in enhanced mode (lower scoring, more defensive, realistic numbers) Not sure the difficulty in finding what in code can cause this but if they can find a way to make enhanced play like the quick sim I think that would go a long way into helping the overall realism fans look for in a sim. I'm not sure. Yes, scoring is (was?) too high in full detail (and then again PP% was 5-10% too low in Quick sim, correct that and scoring is probably the same), and hitting was too low. D-Men had too many shots and too few hits compared to forwards. But those seem like they could be tweaked. Nothing appears to be completely broken like the giveaway/takeaway ratio of Quick Sim mode. Link to post Share on other sites More sharing options...
VonCurry Posted April 10, 2015 Share Posted April 10, 2015 Sounds like those of us who just like to play GM and not coach (Me) should actually set everything to quick sim? Link to post Share on other sites More sharing options...
wazzaflow10 Posted April 11, 2015 Share Posted April 11, 2015 Thanks for your help. I guess the board just wouldn't let a guy who just signed up make a post with 30+ images in it...Test configuration - EHM:EA version 15.0.2b - TBL roster database - NHL and AHL as only active leagues. - No human player in control of a team - Vacation until April 12. - Ran test once with detail level set at all games (NHL) - Ran test once with detail level set at Quick Sim. Interesting idea and great work! Couple questions though. Did you just run the season once? Or did you do it multiple times and just post these results? I'm not at all saying that certain things don't look out of whack, but some could just be variance. It be worth running about 30 tests to see which statistics are statistically significantly different. fwiw I'd be willing to help with the significance testing. Link to post Share on other sites More sharing options...
Archibalduk Posted April 11, 2015 Share Posted April 11, 2015 A similar question about these tests was posted over on TBL. Mafarfloune's answer was as follows (and I think it's a pretty fair point): [...] I'm trying to evaluate/compare the average numbers players seem to get in one given game. I feel like a sample of 1,230 games (one full NHL season, all games combined) is enough for that, or at the very least enough to see the trend. Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.