|
Post by Oxford United on Jan 28, 2024 9:52:32 GMT -7
It's mid-season in the NHL, and good progress has been made getting CHL games played. I've completed changes to the engines of both the Card and Rating generators, and made several cosmetic changes to the cards. That work is ready to support any changes needed for next season.
It must be time, then, to solicit some feedback in advance of the end of the NHL season and the beginning of "Ratings Watch '24".
What feedback would you offer on this year's ratings and cards? What do you like and dislike. What changes would you like to see? Any additions? Subtractions? How do you feel about the current ratings, cards, and supporting rules?
I'd like to know how it's going and what I can do to make next year's ratings and cards the best yet. Be honest. You won't hurt my feelings, I promise.
Thanks in advance, Steve
|
|
|
Post by revelstoke on Jan 28, 2024 10:15:47 GMT -7
Things I like: Goalie cards having the same matrix. Having rebounders again
Things I don't like: Too many coincidental penalties (I think this was an issue we talked about already though) Finding I roll far too much on the broken stick. It just doesn't feel quite right for the 6-11 oddly enough. Still don't like that cards are based solely on even strength goals/points, though I understand the reasoning for it. Playoff goaltending stack. Too much likeliness of rolling in a backup goalie even though the starting goalie could play all but 1 game as an example.
General comment: I like to see fewer odd rules, more just playing with the rules as they were other than minor tweaks.
|
|
|
Post by Belfast on Jan 28, 2024 12:21:13 GMT -7
The ratings this year seem better, as they have been year over year to date. I think that the defense bell curve may have been tightened a tad too much, and we discussed this already, and I believe you already had taken notes and had a plan in place for next year. They're very strong though, I think you've dialed the layout, extra's and defense in pretty well.
I'd like to have some easily understood way to get an idea of a players likely rating during the season. I know it's your secret sauce, but since we're accepting that players can/will vary season to season (sometimes quite drastically) I'd like to at least be able to make a reasonable guess at a guys performance so I can make a semi-informed decision on trades and targets.
The penalties locations are just mystifying. Comparing 5 players withy similar minutes gives you 5 completely different X scenarios. Some will have 2 X on a 6 or 8 and then another wont have an X till you get to a 5 or 9. I understand it's a random generation, but they seem uneven to the eye test.
Coincidental penalties seem to be very high this year. Over my 14 games, Ive had 17 PP, 37 PK and 42 Coincidental instances. Over the 42 coincidental 28 have been initiated by the opposition and Ive generated 14. That seems like too many even up and not enough ADV/DIS situations.
I'm not fond of the matrix being 5v5 only and then having your PP range being the only factor for that advantage situation. We end up with excellent NHL players that are not really viable at 5v5 and can be overly charged for PP situations. Since we aren't the exact replica of NHL teams, we dont use some of these players in the same way that their NHL teams would. If you have too many players who score 30% or more, of their goals on the PP, you are anemic at 5v5, and since our PP's are only 5 ticks these players impact is severely limited. Look to an Eriksson-Ek or a Nichushkin for example. Big minute guys with 1-9 5v5 and 3-10+ on the PP. Pretty insignificant at 5v5 and outstanding on the PP. But since we utilize them differently, I think a lesser ratio of their PP contributions should be used to smooth out some of that discrepancy for our league play.
I like the Broken stick. Some people feel it occurs to frequently, but I seem to be averaging one per game which doesn't seem unbalanced to me. I've had 12 in 14 games and my opponents have generated 11 in 14 games. Seems pretty comparable to what you'd see in an NHL game.
The @ and + system still seems like it needs work. How many times do you get 2 passes and then a turnover, or successive passes and then to your schlub defenseman with a 1-5 range? I still think that NHLers with high assist ratings are not represented well in the game. I'd like to see the 2 passes = a shot implemented (I realize that's a rule proposal, just including that thought here as well). I'm not sure how we can make the excellent passing more impactful without unbalancing the game. Maybe we think of something like when a shot is generated from a successful pass the player adds the +numbers to their shot chance? OR maybe they use their PP range since that shot should be a quality chance after those successful passes? Maybe the successful passes choose who the pass goes to?
I think that this year's set is probably 85=90% of what we want to see, we're really just tweaking since you've done a fantastic job at getting the ratings et al up to top notch so quickly.
|
|
|
Post by alphalackey on Jan 28, 2024 16:54:13 GMT -7
The penalties locations are just mystifying. Comparing 5 players withy similar minutes gives you 5 completely different X scenarios. Some will have 2 X on a 6 or 8 and then another wont have an X till you get to a 5 or 9. I understand it's a random generation, but they seem uneven to the eye test. Definitely look not just at the minutes played but also the misconduct ratings. Like Montour had over 100 minutes, but with half his penalties being misconducts, I'd expect his matrix to be much less penalty-ridden than an equivalent top pair D with 107 PIM and a similar 'drag' percentage (37.5%, all even ups)
|
|
|
Post by Oxford United on Jan 28, 2024 21:41:38 GMT -7
Goalie cards have the same matrix. Too many coincidental penalties (I think this was an issue we talked about already though) Finding I roll far too much on the broken stick. It just doesn't feel quite right for the 6-11 oddly enough. Still don't like that cards are based solely on even strength goals/points, though I understand the reasoning for it. Reaction was good to this inadvertent change. I am quite tempted to make all goalie matrices identical going forward. There was a problem with the coincidental penalties formula which resulted in a higher probability of matching penalties than intended. This is fixed for 2024-25. The Defensive Breakdown (&) result has been moved from 6-11 to 6-12 for 2024-25, halving the chance of occuring. There's no way back from 5v5-based matrices, though fine tuning continues. Thanks!
|
|
|
Post by alphalackey on Jan 29, 2024 12:22:26 GMT -7
Tame flames if you must, I must only point out that I invoke both Barry's name and his landmark contribution to Faceoff theory with nothing but the deepest of respect, and at the very least, asking for clarification between "actual N/216 chance" versus "optics of placement" is reasonable as they have two different answers.
And besides, aren't the Flames tame enough this year?
PS when Lindholm to Jets?
|
|
|
Post by Oxford United on Jan 29, 2024 14:45:41 GMT -7
I think that the defense bell curve may have been tightened a tad too much, and we discussed this already, and I believe you already had taken notes and had a plan in place for next year. The penalties locations are just mystifying. Comparing 5 players withy similar minutes gives you 5 completely different X scenarios. Coincidental penalties seem to be very high this year. I'm not fond of the matrix being 5v5 only and then having your PP range being the only factor for that advantage situation. I still think that NHLers with high assist ratings are not represented well in the game. I'd like to see the 2 passes = a shot implemented (I realize that's a rule proposal, just including that thought here as well). I'm not sure how we can make the excellent passing more impactful without unbalancing the game. Maybe we think of something like when a shot is generated from a successful pass the player adds the +numbers to their shot chance? OR maybe they use their PP range since that shot should be a quality chance after those successful passes? Maybe the successful passes choose who the pass goes to? The 2023-24 ratings set 50% of players with average defence (3 for defence, 2 for forward). The 2024-25 ratings will have 45%. Penalty locations are random. The more penalties per minute a player has, the more likely some of those penalties land on the 6 and 8 row. There was a problem with the coincidental penalties formula which resulted in a higher probability of matching penalties than intended. This is fixed for 2024-25. There's no way back from 5v5-based matrices, though fine tuning continues. Assists (and goals) contribute to +PASS rating. Assists improve the quality of P results on the matrix. I think assists/60 are well represented on the cards. I'm not planning a change for 2024-25, though fine tuning continues. (I will be making a slight change to the transparancy of pass quality to make it easier to see who the really good passers are. More on that later.) Thanks!
|
|
|
Post by Oxford United on Jan 29, 2024 14:47:47 GMT -7
The penalties locations are just mystifying. Comparing 5 players withy similar minutes gives you 5 completely different X scenarios. Some will have 2 X on a 6 or 8 and then another wont have an X till you get to a 5 or 9. I understand it's a random generation, but they seem uneven to the eye test. Definitely look not just at the minutes played but also the misconduct ratings. Like Montour had over 100 minutes, but with half his penalties being misconducts, I'd expect his matrix to be much less penalty-ridden than an equivalent top pair D with 107 PIM and a similar 'drag' percentage (37.5%, all even ups) Misconduct penalties are not included in the volume of penalties represented on the matrix. Misconducts are set aside and assigned a range that represents their ratio to minor and major penalties incurred.
|
|