|
|
Post by Dane-Hamilton on Nov 28, 2022 19:52:22 GMT -5
Wanted to wait until the sample size was large enough that it wasn't an anomaly but some of the stats I've been looking at makes me think we are starting to stray away from the NHL game.
Main issue is scoring, we do too much of it. We are currently scoring at a 20% quicker pace than the NHL this year which doesn't seem like a ton but when looking at goalie stats it's hard not to wonder if this additional scoring takes star goalies entirely out of the GHL.
I looked at the top goalies and their stats are average at best (none under 3 GAA, 5 below .900)
Shesterkin - .907 3.20
Andersen - .907 3.43
Markstrom - .904 3.29
Sorokin - .911 3.00
Jarry - .896 3.67
Swayman - .894 3.51
Kuemper - .890 4.00
Comrie - .891 3.00
Husso - .913 3.00
Ullmark - .891 3.88
The argument for these stats are maybe their D is trash, or they only play the toughest matchups but still they are all at least .020 away from their NHL stats and on different teams. I think if we could tweak it to where we reign in the scoring a bit GHL goalies may come closer to emulating their NHL selves.
Also looking at last two full seasons between both GHL and NHL 2018-19 and 2021-22 the closer the scoring the more realistic the top goalies seem to be.
2021-2022 - Outscored NHL by 14%. Top goalies Hill (58-65), Gustavsson (65-65), Oettinger (62-55), Fleury (75-73), Hellebuyck (54-60)
2018-2019 - Outscored NHL by 8%. Top goalies Miller (?), McElihinney (?), Hutton (82-87), Hellebuyck (71-78), Saros (72-80)
Not sure what if anything we can do to reign in scoring but think that may help with the goalie card vs stats issues and bring some realism back to vezina talks.
Obviously just discussion and wouldn't want any changes (if any) mid season
|
|
|
|
Post by Glenn-Philadelphia on Nov 28, 2022 21:38:25 GMT -5
Nice analysis Dane. I know scoring was up at the beginning of these season due to some off season tweaks in making the attempted shots/shots on goal a little more in line with what the NHL produces. This tweaking wont actually be noticeable until the games are broadcast on twitch again. That being the case the truer test of what scoring is only taking games into account that were played since the tweak I made to the scoring lookup tables/charts about a month or two back and as was discussed in the "80s like scoring" thread.
About a week ago I did a cursory review of some of latest games to make sure goalies were being loaded correctly and that these sub categories were being pulled properly as well.
As I have access to all of the game data in tables it makes it easier for me to review the data from a standpoint of how scoring has been since the tables were updated. Once done I will post my findings.
|
|
|
|
Post by Glenn-Philadelphia on Nov 29, 2022 10:19:18 GMT -5
OK, some preliminary results as a baseline:
NHL 21-22 Season
GP 1,312
GS 8,252
GPG 6.29
---
Off season fix was made to increase the amount of shots that once taken, would be on net (too many shows going wide)
Start of 22-23 GHL season
GHL Games prior to 10/19 (before fix 80's table scoring tweak below)
GP 56
GS 520
GPG 9.29
GHL Games on or after 10/19 (fix 80's table scoring tweak)
GP 296
GS 2,133
GPG 7.21
GHL Year to Date: (includes both pre and post fix games)
GP 352
GS 2,653
GPG 7.54
Adding last years cards/stats used for cards for reference
21-22 GHL Regular season goals per game: 7.17
20-21 NHL regular season goals per game: 6.07 (stats used to generate the 21-22 GHL cards)
18.2% GHL scoring higher than NHL (1.1 more goals per game)
Current cards/stats
22-23 GHL Regular season goals per game (post fix): 7.21
21-22 NHL regular season goals per game: 6.29 (stats used to generate the current GHL cards)
14.6% GHL scoring higher than NHL (.9 more goals per game)
Comparing last years GHL stats to this years stats tabulated after the 10/19 fix we see that scoring is up slightly .04 at goals per game. This equates to an increase of 0.6% year to date.
One thing we should consider when evaluating these numbers as an aggregate, is that they include games in the GHL where goalies who are starting are not NHL goalies per-se. Some teams need to use AHL caliber goalies as their starters because that is all they have in their organization.
Granted, the statement above will not have a bearing when I have the time to do a deeper dive into the goalies listed in Dane's original post.
All, please feel free to comment and add your thoughts as well.
|
|
|
|
Post by Jon-Seattle on Nov 29, 2022 12:20:41 GMT -5
I’m not sure if it matters but I feel like my entire teams a statistical anomaly. My best defensive forwards have the worst +\- and I only have 2 everyday defensemen (who are also two of my better defensive defensemen) with a negative +\- but I also am scoring the fewest goals in the league despite having one of the highest offensive aggregates so it all seems odd to me
Edit - I will add that I think I had played 2-3 less games than most other teams before Glenn’s tweak which seems to have mitigated the worst of the barn burner games but my last ten games still sit somewhere around 8 total goals per game.
Shesterkin also gave up another 7 goal game although I will give the caveat that it was at least up against the (take a screenshot Suffolk) offensive juggernaut that is the Stampede this year rather than the newfound lottery contender that is Cleveland (though if he’s not already looking for a 5th crown next year I won’t be surprised). But I’m getting sidetracked:
Shesterkins games played and Save %s
NHL 19-20: 12gp .932%.
GHL 20-21: 34gp .896%. -maybe just overplayed
NHL 20-21: 35gp .916%
GHL 21-22: 48gp .899% -blame the team?
NHL 21-22: 53gp .935%
GHL 22-23: 15gp .907%
it can’t all be due to the lineup ahead of him when on paper I should be just as good if not better than the majority of teams I’m playin against
|
|
|
|
Post by Glenn-Philadelphia on Nov 29, 2022 13:15:15 GMT -5
I did some further analysis over lunch. Part of the scoring increase may be because of teams PP efficiency as compared to the NHL. Last years NHL PP percentage average was 20.6%. Year to date in the GHL this year, teams are scoring on an average of 24.1% of their PP opportunities.
As time permits I will review the code handling the situations to determine if the off season change had any additional unintended downstream impacts.
|
|
|
|
Post by Glenn-Philadelphia on Nov 29, 2022 13:33:36 GMT -5
I’m not sure if it matters but I feel like my entire teams a statistical anomaly. My best defensive forwards have the worst +\- and I only have 2 everyday defensemen (who are also two of my better defensive defensemen) with a negative +\- but I also am scoring the fewest goals in the league despite having one of the highest offensive aggregates so it all seems odd to me Edit - I will add that I think I had played 2-3 less games than most other teams before Glenn’s tweak which seems to have mitigated the worst of the barn burner games but my last ten games still sit somewhere around 8 total goals per game. Shesterkin also gave up another 7 goal game although I will give the caveat that it was at least up against the (take a screenshot Suffolk) offensive juggernaut that is the Stampede this year rather than the newfound lottery contender that is Cleveland (though if he’s not already looking for a 5th crown next year I won’t be surprised). But I’m getting sidetracked: Shesterkins games played and Save %s NHL 19-20: 12gp .932%. GHL 20-21: 34gp .896%. -maybe just overplayed NHL 20-21: 35gp .916% GHL 21-22: 48gp .899% -blame the team? NHL 21-22: 53gp .935% GHL 22-23: 15gp .907% it can’t all be due to the lineup ahead of him when on paper I should be just as good if not better than the majority of teams I’m playin against I did a quick look at your team stats and the worst forwards on your roster +/- wise are Kase (-11), Rodriques (-5) and Perason (-5). Now I think it's safe to say these guys wouldn't be considered Selke candidates year in and year out as they have defense ratings of 43, 45 and 45 respectively. I did note that these guys have staminas of 5 and 4 as well. I would check the game time these guys are getting to make sure they aren't being over worked, and suffering from the depreciated performance because of this. You can modify playing time for your 4th line on the lineup page to try to find a remedy for your teams malaise. Of course, there also is always the line juggling option to try as well. I have often been puzzled by my own goaltenders issues regarding Vasilevskiy's performance. This year is a little more understandable because of the only above average card. I will continue to review the engine and will add some additional output to the debug tables for me to review to make sure there isn't anything that is not operating as it should. |
|
|
|
Post by Dane-Hamilton on Nov 29, 2022 19:37:06 GMT -5
I did some further analysis over lunch. Part of the scoring increase may be because of teams PP efficiency as compared to the NHL. Last years NHL PP percentage average was 20.6%. Year to date in the GHL this year, teams are scoring on an average of 24.1% of their PP opportunities. As time permits I will review the code handling the situations to determine if the off season change had any additional unintended downstream impacts. That would def account for some of the extra scoring for sure. PPG roughly 25% of all goals scored. Was also thinking are there different boosts to the different scoring zones? Wondering if there is one zone (like 14) that goalies no matter the card are just bad at. If the average Sv Per in zone 14 is like .600 for example even with the best goalie would be a huge drag on his stats. Outlier like that could explain how the stats for goalies got more even. Hope that makes sense, theory is if Shersterkin sv per is .935 in all other zones but .800 in zone 14 could explain the stats. At that point it wouldn't be so much the card determining the outcome as # of shots from zone 14. Don't get me wrong it should be easier to score from zone 14 but if it's too easy I think it puts goalie stats more to random chance than card. |
|
|
|
Post by Jon-Seattle on Nov 30, 2022 0:50:11 GMT -5
I’m not sure if it matters but I feel like my entire teams a statistical anomaly. My best defensive forwards have the worst +\- and I only have 2 everyday defensemen (who are also two of my better defensive defensemen) with a negative +\- but I also am scoring the fewest goals in the league despite having one of the highest offensive aggregates so it all seems odd to me Edit - I will add that I think I had played 2-3 less games than most other teams before Glenn’s tweak which seems to have mitigated the worst of the barn burner games but my last ten games still sit somewhere around 8 total goals per game. Shesterkin also gave up another 7 goal game although I will give the caveat that it was at least up against the (take a screenshot Suffolk) offensive juggernaut that is the Stampede this year rather than the newfound lottery contender that is Cleveland (though if he’s not already looking for a 5th crown next year I won’t be surprised). But I’m getting sidetracked: Shesterkins games played and Save %s NHL 19-20: 12gp .932%. GHL 20-21: 34gp .896%. -maybe just overplayed NHL 20-21: 35gp .916% GHL 21-22: 48gp .899% -blame the team? NHL 21-22: 53gp .935% GHL 22-23: 15gp .907% it can’t all be due to the lineup ahead of him when on paper I should be just as good if not better than the majority of teams I’m playin against I did a quick look at your team stats and the worst forwards on your roster +/- wise are Kase (-11), Rodriques (-5) and Perason (-5). Now I think it's safe to say these guys wouldn't be considered Selke candidates year in and year out as they have defense ratings of 43, 45 and 45 respectively. I did note that these guys have staminas of 5 and 4 as well. I would check the game time these guys are getting to make sure they aren't being over worked, and suffering from the depreciated performance because of this. You can modify playing time for your 4th line on the lineup page to try to find a remedy for your teams malaise. Of course, there also is always the line juggling option to try as well. I have often been puzzled by my own goaltenders issues regarding Vasilevskiy's performance. This year is a little more understandable because of the only above average card. I will continue to review the engine and will add some additional output to the debug tables for me to review to make sure there isn't anything that is not operating as it should. I was assuming mid 40s defensive rating for a forward was pretty good but looking back it does seem to be more of an average rating, plus all had a positive +\- last year for what that’s worth, but I had been monitoring ice time heavily and all have played at least 2 minutes less than last years average TOI: Kase: GHL-11:40 NHL-13:59 Pearson GHL-9:36 NHL-16:03 Rodrigues GHL-13:15 NHL-15:50 Best pure forward defensive rating is your own Patrice Bergeron (92-73 card) and it falls off quickly after that with a 68, 64, 62 and mid 40s seems to be more average-above average for 3rd-4th liners which is what I meant to imply (and did so terribly) Last week I was looking at when goals were scored against me by period and there doesn’t seem to be any clear indicator that it was a fatigue issue, at least from 1st to third period, as it was a fairly flat rate. I didn’t have time to check early vs late in a period though. I do know that a significant portion of my goals against were/are from net front shots. I think my biggest point is that there’s no glaring explanation for why these guys are allowing so many goals against What do I fix to clean this up? Particularly when the GHLs leading goaltender (Jake Allen/NJ) has 8 games under his belt, a .929save%, 5-6 fort-stamina, and doesn’t have a single defenseman above an 80 defensive rating. Only his first line would beat out those three guys I have listed above. In fact Kase (and his 86-43 card) would still be the statistically worst forward on his roster if I just handed him over, despite having the 4th best F card. Second best goalie is in Colorado and no it’s not the guy with the 72-80 card. It’s the (still really good) 64-69 Pavel Francouz. He also owns a 5-5, fort-stamina but has only played 7 games. Once again all the forwards I have above are defensively more sound than all but Matt’s first line. When healthy his squad might have a slight edge when looking at defensive pairings but not by a significant margin. I’m trying to hopefully give you some data points to focus on but honestly feel free to ignore it Glenn if it isn’t helping. Im just a lot bitter right now. On the flip side the digging through stats is fun so at least I’ve got that going for me? |
|
|
|
Post by Glenn-Philadelphia on Nov 30, 2022 9:02:07 GMT -5
I did some further analysis over lunch. Part of the scoring increase may be because of teams PP efficiency as compared to the NHL. Last years NHL PP percentage average was 20.6%. Year to date in the GHL this year, teams are scoring on an average of 24.1% of their PP opportunities. As time permits I will review the code handling the situations to determine if the off season change had any additional unintended downstream impacts. That would def account for some of the extra scoring for sure. PPG roughly 25% of all goals scored. Was also thinking are there different boosts to the different scoring zones? Wondering if there is one zone (like 14) that goalies no matter the card are just bad at. If the average Sv Per in zone 14 is like .600 for example even with the best goalie would be a huge drag on his stats. Outlier like that could explain how the stats for goalies got more even. Hope that makes sense, theory is if Shersterkin sv per is .935 in all other zones but .800 in zone 14 could explain the stats. At that point it wouldn't be so much the card determining the outcome as # of shots from zone 14. Don't get me wrong it should be easier to score from zone 14 but if it's too easy I think it puts goalie stats more to random chance than card. Regarding zone bonuses. Each zone on the ice has it's own set of lookup tables that are referenced both when a player has an opportunity to shoot (does he get an opportunity to take a shot) and another table that shows the base percent chance to score. The "does he shoot" tables are weighted more heavily to taking a shot from the zones closer to the net. This lookup is also influenced by the shooters shoot rating and the defenders defensive, or neutral rating, selected on each potential shot attempt to try to keep the player from shooting. I believe this is further modified by a factor of how many players are on the ice for each team (PP/SH). Desperation shots can also be attempted from zones that generally have a zero % chance of taking a shot from at end of period/end of game situations. If a shot attempt is not made, the player will either skate with the puck or look to make a pass. If a shot attempt is made then a function is called to determine if the shot is blocked, wide, or on net. Blocked is a function of the players on the ice block ratings as well as the zone the shot is being taken from (less chance to block right in front of the net, higher chance at the point), on net is a function of the players shoot/shot rating. If the player does manage to shoot and the shot is on net, a check is made which includes his "shot" skill value, one of the goalies skill values ("glove", "anticipation", "footwork", etc.), the zone the shot is taken from. This can be modified by potential screens in front of the goalie. So in a nutshell better goalies will have a better chance of making a save from any spot on the ice because they in general have higher skill rating than their lesser skilled brethern. |
|
|
|
Post by Glenn-Philadelphia on Nov 30, 2022 9:39:25 GMT -5
I was assuming mid 40s defensive rating for a forward was pretty good but looking back it does seem to be more of an average rating, plus all had a positive +\- last year for what that’s worth, but I had been monitoring ice time heavily and all have played at least 2 minutes less than last years average TOI: Kase: GHL-11:40 NHL-13:59 Pearson GHL-9:36 NHL-16:03 Rodrigues GHL-13:15 NHL-15:50 Best pure forward defensive rating is your own Patrice Bergeron (92-73 card) and it falls off quickly after that with a 68, 64, 62 and mid 40s seems to be more average-above average for 3rd-4th liners which is what I meant to imply (and did so terribly) Last week I was looking at when goals were scored against me by period and there doesn’t seem to be any clear indicator that it was a fatigue issue, at least from 1st to third period, as it was a fairly flat rate. I didn’t have time to check early vs late in a period though. I do know that a significant portion of my goals against were/are from net front shots. I think my biggest point is that there’s no glaring explanation for why these guys are allowing so many goals against What do I fix to clean this up? Particularly when the GHLs leading goaltender (Jake Allen/NJ) has 8 games under his belt, a .929save%, 5-6 fort-stamina, and doesn’t have a single defenseman above an 80 defensive rating. Only his first line would beat out those three guys I have listed above. In fact Kase (and his 86-43 card) would still be the statistically worst forward on his roster if I just handed him over, despite having the 4th best F card. Second best goalie is in Colorado and no it’s not the guy with the 72-80 card. It’s the (still really good) 64-69 Pavel Francouz. He also owns a 5-5, fort-stamina but has only played 7 games. Once again all the forwards I have above are defensively more sound than all but Matt’s first line. When healthy his squad might have a slight edge when looking at defensive pairings but not by a significant margin. I’m trying to hopefully give you some data points to focus on but honestly feel free to ignore it Glenn if it isn’t helping. Im just a lot bitter right now. On the flip side the digging through stats is fun so at least I’ve got that going for me? I think there is always going to be room for improvement for the engine, the interfaces and the league processes we employ so I would encourage everyone to provide feedback and ask questions. The one thing I think we might want to look at is the teams that Allen and Francouz compiled their impressive stats against. Some owners are very hands-on and will make goalie decisions on a game by game basis. Others just use the general percentage breakdown for the most part to manage their goalies games played. If NY and COL handle their goalies game by game, I would imagine they would put their backups in against lesser quality opponents and save their number ones for the stiffer competition. This would have a direct bearing on stats they have compiled so far this year. I will continue my review as I look at the PP/SH portions of the code and as mentioned prior, am adding some extra debug data to assist in this and any further troubleshooting. |
|
|
|
Post by Chris-Suffolk on Nov 30, 2022 12:17:13 GMT -5
Very pleased to read this conversation. I recall a “young GM” brought this subject matter up back in 2013. I recall, he wasn’t very eloquent with wordsmithing or spreadsheets. I dare say, that young GM seems to have been “trying to say” very similar points as the GM’s from Seattle and Hamilton. They have a clearer way of presenting their points, which I applaud!
Just curious though, why Samsanov, my 3rd string goalie, was left out of the comps? ( totally kidding Jon! ) I agree, Shesterkin’s Card/stats should equate to better results this season. Just laugh! I’m just busting your chops on the point.
I had two separate seasons with teams, I felt, were dominant comparatively to the league/conference, but they didn’t do well at all. I tried to present my case each time, to no avail of the league. I think, my presentations actually led to the formation of the current Board…although it may have been my protesting actions ( selling off cheaply or my temper tantrums) that may have also contributed.
Either way, I respect and appreciate yours and Dane’s contributions to the league. Someday, and it does seem likely to be soon, something will be decided upon which is determined to be fair and a good representation in this subject matter.
|
|
|
|
Post by Chris-Suffolk on Nov 30, 2022 12:30:07 GMT -5
I question a few things about the sim, and I’m not trying to sound disrespectful. Not to toot my own horn, but Suffolk has been a front runner in the league this season for all but 3 or 4 days, yet only this week was able to crack the Top 5 in the power rankings.
That seems off a little to me. I do realize that the machine equates strength of opponent, but when your 12-4 and ranked out of Top 10 and the rest of the league is a game or two above .500 and Ranked in Top 5…something needs adjusting.
As well as the 3 Stars Awards, which isn’t affecting anything really.
Just saying, and not fretting.
|
|
|
|
Post by Glenn-Philadelphia on Nov 30, 2022 14:51:59 GMT -5
The power rankings are only updated on Saturday nights (with the caveat being there has to be at least one game played that night). The rankings only looks at the most recent 8 weeks of games and evaluates each teams points gained over the time period as a value of the points gained against your opponents over the same time period. So this is a rating of teams power when measured against the power of their opponents. ...and you were never young Chris... ..and for some reason you used the past tense when describing your elocution...  I keed, I keed |
|
|
|
Post by Dane-Hamilton on Nov 30, 2022 19:45:42 GMT -5
So the goalie stuff makes sense that better cards have a better chance of stopping the shot as it's all based on card stats referenced against others. Point I was trying to make if there is a zone that gives too much of an advantage to the shooters the goalie card, even if great, becomes average compared to other goalies. It's hard to put into words but example below (way oversimplified) Let's say shooters get a 50% boost from zone 14. Using my two goalies Sim determines a shot is coming and going glove side Anderson Glove = 9 Lehner Glove = 5 Shooter's Percentage is 60 - With bonus 90 Under this Anderson should be better and has roughly even dice roll chance with the shooter. Lehner doesn't and stats should be worse. My theory is that if bonus is tilted too much in the shooter's favor the gap between goalies disappears (becomes more about luck) If that shooter in the above example had shot percent of 90 with a potential bonus could effectively be 135. At that point goalie stats don't matter much as neither goalie are all that likely to stop that shot. As the shooter bonus rises the difference in stats between those goalies (in that zone) should shrink until both goalies are theoretically tied at 0% sv per. Just looking through Andersen's stats got me thinking about this. It looks like 25 of his 54 goals allowed came from zone 14, if zone 14 is essentially a toss up no matter the goalie it could explain how average goalies are leading league in sv per (comes down to which teams keep it out of that zone more). Also the 25 out of 54 in zone 14 isn't that far off though. www.icydata.hockey/vizzes/shots/9 website that has goal location data from 2019 and earlier. Just a quick look our zone 14 accounts for roughly 35% of all NHL goals. |
|
|
|
Post by Dane-Hamilton on Nov 30, 2022 20:39:12 GMT -5
Also just thinking, if that was the case raising shot bonus for the shooters in zone 14 while lowering the shooting percent bonus could keep goals in zone 14 roughly the same.
Total shots would likely go up but goalie cards become more "realistic" in that zone while overall goals staying the same.
Also if just lower sht per bonus could bring goals closer to NHL while also giving more importance to goalies.
|
|
