|
|
Post by Lance-Pittsburgh on Dec 24, 2018 11:01:07 GMT -5
Merry Christmas (Eve), everyone! (Note: I purposely did not post this on Festivus because I did not want anyone to think I was airing grievances.) Recently the card generation process has been the subject of many questions (goalie cards and nerfing, stats used for card generation, lack of clarity in how particular stats influence a player's card). In other simulation leagues that I'm in, the formulae for card generation is public knowledge ( www.simhockeyfans.com/eat.sleep.sim-ratings.pdf, as an example). These other leagues also have a publicly available card calculator ( www.shl.simhockeyfans.com/skater-ratings-calculator/ and www.shl.simhockeyfans.com/GoalieRatingsCalculatorv2/GoalieRatingsCalculatorv2.htm) so that GMs can accurately project future ratings. Would this: a) be something GMs would want?, and b) be something Glenn would do? It may go a long way to making clear the process around player card generation. Personally, I'd be in favor of such things, if only so that I can make more accurate and informed decisions on my franchise's future. |
|
|
|
Post by Scott-New York on Dec 24, 2018 12:56:55 GMT -5
Would this not open the crystal ball effect even wider than it already is?
|
|
|
|
Post by Glenn-Philadelphia on Dec 24, 2018 13:35:58 GMT -5
I know Daner already has a pretty accurate OFF model that be built based on the requirements that I published detailing what goes into those.
Also, it seems like its almost a 2nd Christmas once the cards get published in early July. It would be kind of like seeing everything your parents got you unwrapped a few days prior to Christmas.
|
|
|
|
Post by Lance-Pittsburgh on Dec 24, 2018 13:51:12 GMT -5
Would this not open the crystal ball effect even wider than it already is? Without a doubt. But when a simulation league re-rates based NHL/AHL performances, I'm not sure it matters much. While future player attributes would be clearly defined, a season still needs to be played. The sim engine seems to have a history of not caring what a team looks like on paper. |
|
|
|
Post by Lance-Pittsburgh on Dec 24, 2018 13:53:11 GMT -5
I know Daner already has a pretty accurate OFF model that be built based on the requirements that I published detailing what goes into those. Also, it seems like its almost a 2nd Christmas once the cards get published in early July. It would be kind of like seeing everything your parents got you unwrapped a few days prior to Christmas. I don't know much about Dane's model, but I agree on your second point. But I also liked sneaking around the house and determining what my parents got me for Christmas well before the big day. It was not my intention to open a can of worms with the original post -- I just wanted to throw out an idea that I've seen elsewhere. |
|
|
|
Post by Phil-Cornwall on Dec 24, 2018 14:04:33 GMT -5
It's a shit ton more maths. The big man likes maths. Let's use more maths.
|
|
|
|
Post by Jon-Seattle on Dec 30, 2022 15:46:38 GMT -5
I'd still like something clear on this. I've moved on from players knowing they are great players in the NHL because I thought I understood what was translating to the GHL and what wasn't. Then I see a guy like Ekblad, who I realized would never be a powerhouse defenseman in the GHL because he's more of an offensive defenseman, suddenly has a top 4 defensive rating.
Ekblad:
Average TOI = 24:55
Total TOI = 1520
Blocks = 69
Blocks/60 = 2.7
Hits = 62
Hits/60 = 2.5
Takeaway/60 = 1.7
Giveaway/60 = 2.9
Average SHTOI = 2:47 (1st on team by a single second)
Player X
Average TOI = 20:24
Total TOI = 1611
Blocks = 144
Blocks/60 = 5.4
Hits = 227
Hits/60 = 8.5
Takeaway/60 = 1
Giveaway/60 = 2.9
Average SHTOI = 2:34 (14 seconds behind second and nearly 40 seconds behind first because someone in Montreal has a hard on for Ben Chariot)
The only stat in which Ekblad seems to eclipse the other player is average SHTOI (by a whole 13 seconds) and takeaways. Have you guessed player X yet? It's Alexander Romanov who has a defensive rating of 80 compared to Ekblads suddenly dominant 97. Tell me what I am missing? Ya'll keep telling me I'm missing something so please, tell me what I'm missing here. I correctly moved on from a player who I know doesnt fit in with the GHL model and suddenly despite his stats not being something that jump out at you as dominant (because they're not) suddenly he's the 4th best defenseman in the league? This is ridiculous.
|
|
|
|
Post by Chris-Suffolk on Dec 30, 2022 15:54:20 GMT -5
Totally agree! I watch Adam Fox every time the NY Rangers play! He is 100% NOT A DEFENSIVE DEFENSEMAN! He literally couldn't push shaved ice out of the crease. He may block a few passes and he might even have the IQ to be a defenseman...BUT, HE AIN"T GOOD AT BEING ONE! He has a 79-84 Card and the Ratings in his Defense Column are indicative of some of the Best D Men in the NHL. He SUCKS at Defense!!! He is a 4th Forward!!! He's Slower than shit!!! His card shows him as a 9 SPEED!!! LOL!!! ghl.greytowergames.com/PlayerCards2.aspx?PlayerID=16228 Check Out his card for yourself |
|
|
|
Post by Glenn-Philadelphia on Dec 30, 2022 16:12:20 GMT -5
Another part of the defensive calculation is the players +/- rating as a function of his team's +/- rating are also, you need to factor that in as well. I don't believe those were included in you stats so I took a quick look.
Ekblad was a +38, Romanov was a -9.
Now, these are modified to account for playing on a team with a good +/- rating versus a poor one however the guys with there is a net difference of 47 between these two guys which will also account for some of the difference.
Short handed playing time is factored in as well.
I have always said that calculating defensive ratings was the most difficult part of card creation because the stats aren't as defined as their offensive counterparts. I am always looking to make the cards as accurate as possible and any constructive criticism is more than welcomed.
|
|
|
|
Post by Dane-Hamilton on Dec 30, 2022 20:37:50 GMT -5
Also if I recall SAT % has a decent factor on D rating. Glenn can correct me if I'm wrong but I believe every % point above or below 50% raises/lowers D rating by a point.
For Romanov 46.9% so 3 off his D rating right there
For Ekblad 57% so 7 points to his D rating
Without the SAT % adjustment
Romanov - 83
Ekblad - 90
Which is much more reasonable looking at stats. I believe my argument back in the day was it hurts players on bad teams too much and gives too much of a boost to good teams. Pretty sure that's what happened here.
Also don't get me wrong I like the SAT% as a factor it's another good measure. You get shot on more than you shoot you probably deserve to get a hit to the D rating.
|
|
|
|
Post by Jon-Seattle on Dec 30, 2022 20:46:03 GMT -5
Totally agree! I watch Adam Fox every time the NY Rangers play! He is 100% NOT A DEFENSIVE DEFENSEMAN! He literally couldn't push shaved ice out of the crease. He may block a few passes and he might even have the IQ to be a defenseman...BUT, HE AIN"T GOOD AT BEING ONE! He has a 79-84 Card and the Ratings in his Defense Column are indicative of some of the Best D Men in the NHL. He SUCKS at Defense!!! He is a 4th Forward!!! He's Slower than shit!!! His card shows him as a 9 SPEED!!! LOL!!! ghl.greytowergames.com/PlayerCards2.aspx?PlayerID=16228 Check Out his card for yourself According to my research, as per the GHL card rating, he's the best defenseman the NYR's have to offer LOL That said I do understand the randomness of substats. It'd be nice to see more correlations between NHL stats and card stats. Not sure if it's possible without going absolutely insane. Maybe using NHL draft combine stats to fill in Neutral (?) stats such as speed with their on ice split times (I think there's a red line to red line time collected but I honestly don't know.) In addition to Glenns point I find it really weird that a single stat seems to overshadow nearly everything else. Though I will add that after writing my last entry regarding Ekblad v Romanov that ekblad has a + CF and Romanov has a - CF, though I was sure that last time I had brought it up you pointed out it is a very minor part of the calculation. I will also point out that they both played nearly the same total amount of ice time due to injuries. Takeaways seem to be an odd stat as the numbers seem really low to me and somewhat subject team to team (though I know we can partly say the same about hits) but maybe thats just me. I just want to take a quick look at what Glenn says is used to calculate defensive cards real quick on here and for once I havent looked ahead before writing this seconds played between blocked shots - Romanov (2:1) seconds played between takeaways - Ekblad (2:1) seconds played between giveaways - Even percentage of time played shorthanded - Evenish (PP opportunities are slightly less for Ekblads Panthers but he missed significantly more games and SHTOI per game are within 13 seconds of each other) Corsi Rating - Ekblad (57.8:41.6) player plus minus compared to teams plus minus - Ekblad (+38:-9, didnt compare it against team average but a quick look shows Florida was a +95 and Montreal a -101 so may be closer than what is originally perceived) So I will say that Ekblad has a dominant Corsi and great +/-. I always forget hits arent included in defensive rating which always seems odd as it is something I personally associate with playing defense. Romanov is better than 3:1 in hits for what thats worth. Its just really odd to see +/- being such a key stat that puts Ekblad up by 20 points in his defensive rating (Corsi is strongly related to +/- if I remember correctly). Also Ekblads 20-21 stats dont look much different to his 21-22 stats with biggest difference is that he's only a +5 over half the number of games (similar high Corsi). Also, takeaways & giveaways are about half what his per game stats are in 21-22 which seems like would cancel each other out. Despite all that ekblads 21-22 card is 82 defense while this year he's a 97. looks like the card increased by one for every additional +. Without knowing the formula though it's difficult to understand |
|
|
|
Post by Brenden-Oregon on Feb 15, 2023 11:32:51 GMT -5
If a player gets traded during the season, is his +/- rating compared to his new teammates only? Or will the +/- for each team be weighted?
Example: A player is on a bad team, but has a great +/- compared to his teammates. Player gets traded to a good team and now has one of the worst +/- on his new team.
|
|
|
|
Post by Glenn-Philadelphia on Feb 15, 2023 13:58:59 GMT -5
I believe the generator will use the new teams +/- rating to determine the +/- portion of the players defensive composite score. I don't think I put anything in place to apportion his old versus new teams stats but I would need to double check to be sure.
|
|
