Penalty Killers

Buchs · 03-06-2018, 07:02 PM

I'm not sure if something changed with this last update, but It seems that the AI just puts all your best offensive players on the PK now. can anybody tell me what attributes the game uses to determine who will be a good PK player? It certainly doesn't seem to be defensive ratings.

I'm also not sure if it is taking position or faceoff skill into the equation. For example my first PK pair is 2 centers and the other two lines are all players that have no experience at center and all their FO rating is between 11-13 among the worst FO guys on my team.

Buchs · 03-06-2018, 08:39 PM

I did a little more testing and I do see some defensive players getting used in a few tactics, but it seems like they are barely making it in on the last pairing. I still don't think it is pairing centers and a winger resulting in pairings that are terrible at face offs.

JeffR · 03-06-2018, 11:55 PM

We didn't change anything in the penalty-killing tactics or role requirements, other than the general changes that affected all of the tactical fit scores. It may be that the offensive players you're seeing may have the correct skills to fit the PK tactic in question.

And there is a routine to prioritize faceoff skill for one of the PK slots - if a good faceoff man doesn't get the first PK forward spot, it'll put more emphasis on that attribute when looking at the second one. Although that doesn't necessarily mean the player will be primarily a centre, just that he'll be decent on faceoffs.

Buchs · 03-08-2018, 10:56 AM

okay, so I didn't do a lot of testing with any other teams besides the Florida Panthers, so it is possible that this issue is limited to the Panthers, but seems more like a role evaluation issue to me.

I tried all PK tactics and it seems to be by far back checking forward is the primary role used on PK. I'm not entirely sure what goes into that rating but i'm not sure it is rating defensive players well. I think this is what is causing my defensive players to not get PK time while I have the AI putting players that are pretty bad defensively in there instead resulting in very bad PK performance.

I used the Florida Panthers for testing since I am most familiar with that team and that is who I was playing when I applied the latest update. before the update, my defensive players where getting regular PK time. The real panthers primary PK duo is MacKenzie and Sceviour. These guys usually got PK time in my game before the update. I think the ratings on those guys look about right to me. Sceviour especially is rated as a fairly good defensive player. Neither one of those guys make it into the AI evaluation for PK on any tactic when my roster is healthy now. consistently guys like huberdeau, Malgin, even Bjugstad get rated ahead of Sceviour although he has better defensive ratings and usually at least equal or close speed rating. I did some testing where I undressed guys ahead of them in the PK and still found that McCann, Brickley, and even Vrbata would get in ahead of Sceviour. I even tweeked his speed up a couple points to be better than a couple guys in front of him to see if that was the issue. I did the same by increasing his stick checking and position ratings by a point (which by the way was already equal or better than all the guys in front of him) and he was still behind these guys in AI evaluation.

I know its a small sample size, but letting the AI simulate 3 seasons Panthers were ranked 28th, 29th, and 22nd in PK. I did 2 more seasons with MacKenzie and Sceviour as my top PK pair (like real life) and finished 9th and 16th in PK. So it seems at least likely to me that poor defensive players getting on the PK is at least hurting some teams like the Panthers. I also think that bad FO players probably having an affect also. Like I said in my original post I had a guy with a 11 and a 13 in FO together on the 2nd PK line and it was like 2 of my worst 4 forwards for face offs.

Can you tell me what goes into the back checking forward rating?

on a side note Huberdeau is criminally under rated for offensive roles now. he is a 3/3 for all offensive roles. I seems like his ratings look okay, so i'm not sure why he can't even eke out a 4 in any offensive role.it does put him on my first line, but seeing a legit first line guy not even able to rate as a second liner (4/4) seems weird.

JeffR · 03-09-2018, 02:07 AM

I'll have a closer look at some full-season saves and try to see if I can find a pattern there, a couple of those things don't sound quite right.

Buchs · 03-09-2018, 09:14 AM

Quote:

Originally Posted by JeffR

I'll have a closer look at some full-season saves and try to see if I can find a pattern there, a couple of those things don't sound quite right.

I wish i could have tested more thoroughly, but I din't have enough time. I noticed it while trying to start a new season and ended up spending all evening on Wednesday night. I don;t doubt there may be some things there are probably things I missed.

I spot checked a couple other teams using the site left wing lock to check top penalty killers and compared to in game for one of my seasons and spotted some other PK guys not be used as well. This could simply be due to the chosen tactic or could be the same thing I'm seeing. unfortunately I can't remember which ones I spotted except Wayne Simmonds on Philly.

I just wanted to note that when I was testing to see who was ahead of Sceviour and what stats would affect the evaluation, It was limited to the role back checking forward. Pretty much all of these guys were 3/3 in that role.

Buchs · 03-22-2018, 08:42 AM

well, I may be the only one dealing with this, But it is driving me crazy, so I can;t leave it alone. Anyway, I figured out what the issue is, so I figured I would share . The issue in my case seems to be the puck handling stat. It seems to very heavily influence PK evaluation by AI far more than defensive ability. I don't know if that is intended or not.

I was comparing MacKenzie and Vrbata. In the player editor Mackenzie has about 100 more total on his defensive abilities than Vrbata. I raised MacKenzie's physical stats, and position stats to all be equal to or greater than Vrbata's. This meant MacKenzie was better than or equal to Vrbata in every category but offence and was significantly better defensively. However, Vrbata was still getting put in the penalty kill ahead of Mackenzie on every tactic. At that point I knew it was an offensive stat.

I figured it was probably puck handling, not only because it makes some sense, but there was a large gap between the two since MacKenzie only had a 12 in that stat. I raised it to 14 and it put him on my second PK line. I put all of his other stats back down to where they were and it kept him in the same spot. I played around with guys in the editor, removing one point of puck handling almost always dropped guys right off the bottom PK pairing. Bringing crappy guys up to 15 or 16 in puck handling almost always put them in on PK. No other stat I messed around with had such a large and obvious affect on the AI evaluating guys for PK.

In the case of Florida I don't know if the problem is the formula for PK is off or if the puck handling skill of our PK players is too low, but it is resulting in some of my worst defensive players getting PK time especially if Trocheck or Barkov is injured, and a consistently abysmal PK if I let the AI set lineups.

Gerry58 · 03-22-2018, 09:10 AM

Have you compared the ratings with the PK tactics available? They can change quite a bit depending on the tactics you are actually using

Buchs · 03-22-2018, 11:44 AM

Quote:

Originally Posted by Gerry58

Have you compared the ratings with the PK tactics available? They can change quite a bit depending on the tactics you are actually using

yeah the result was the same for every tactic. Pretty much every role uses defensive or two way role for forwards which is dominated by back checking forward. A few guys would get rated as up down wingers on the two way roles. There was one very aggressive PK tactic that gave slightly different results, because the tactic had a slightly more counterattack/office focused roles. I can't remember the name of the tactic right now, but I think it actually had the word aggressive in it.

Interestingly the puck handling stat doesn't seem to just effect the formula for back checking forward but has the same effect on the shadow role. In the case of MacKenzie, who has fairly good overall defensive ratings, just moving him up from 12 to 14 in puck handling increases both his back checking forward and shadow ratings from 3/3 to 4/4 and AI slots him in as shadow, probably because he is better defensively than most, if not all of the other guys on the PK.

maikgianino · 03-22-2018, 01:15 PM

In my case my assistant coach was using a RW with (8) Faceoff skill and that was one of the reasons why my PK was so weak. I'm slowly taking care of the full tactics and lineups due that.

Gerry58 · 03-22-2018, 03:51 PM

Quote:

Originally Posted by maikgianino

In my case my assistant coach was using a RW with (8) Faceoff skill and that was one of the reasons why my PK was so weak. I'm slowly taking care of the full tactics and lineups due that.

Problem is that you can't do that for every other team in the League. Hope Jeff can find the bug and fix it

03-06-2018, 07:02 PM	#1
Buchs Minors (Rookie Ball) Join Date: Apr 2013 Posts: 32	Penalty Killers I'm not sure if something changed with this last update, but It seems that the AI just puts all your best offensive players on the PK now. can anybody tell me what attributes the game uses to determine who will be a good PK player? It certainly doesn't seem to be defensive ratings. I'm also not sure if it is taking position or faceoff skill into the equation. For example my first PK pair is 2 centers and the other two lines are all players that have no experience at center and all their FO rating is between 11-13 among the worst FO guys on my team.

03-22-2018, 08:42 AM	#7
Buchs Minors (Rookie Ball) Join Date: Apr 2013 Posts: 32	well, I may be the only one dealing with this, But it is driving me crazy, so I can;t leave it alone. Anyway, I figured out what the issue is, so I figured I would share . The issue in my case seems to be the puck handling stat. It seems to very heavily influence PK evaluation by AI far more than defensive ability. I don't know if that is intended or not. I was comparing MacKenzie and Vrbata. In the player editor Mackenzie has about 100 more total on his defensive abilities than Vrbata. I raised MacKenzie's physical stats, and position stats to all be equal to or greater than Vrbata's. This meant MacKenzie was better than or equal to Vrbata in every category but offence and was significantly better defensively. However, Vrbata was still getting put in the penalty kill ahead of Mackenzie on every tactic. At that point I knew it was an offensive stat. I figured it was probably puck handling, not only because it makes some sense, but there was a large gap between the two since MacKenzie only had a 12 in that stat. I raised it to 14 and it put him on my second PK line. I put all of his other stats back down to where they were and it kept him in the same spot. I played around with guys in the editor, removing one point of puck handling almost always dropped guys right off the bottom PK pairing. Bringing crappy guys up to 15 or 16 in puck handling almost always put them in on PK. No other stat I messed around with had such a large and obvious affect on the AI evaluating guys for PK. In the case of Florida I don't know if the problem is the formula for PK is off or if the puck handling skill of our PK players is too low, but it is resulting in some of my worst defensive players getting PK time especially if Trocheck or Barkov is injured, and a consistently abysmal PK if I let the AI set lineups. Last edited by Buchs; 03-22-2018 at 08:56 AM.

03-06-2018, 08:39 PM	#2
Buchs Minors (Rookie Ball) Join Date: Apr 2013 Posts: 32	I did a little more testing and I do see some defensive players getting used in a few tactics, but it seems like they are barely making it in on the last pairing. I still don't think it is pairing centers and a winger resulting in pairings that are terrible at face offs.

03-06-2018, 11:55 PM	#3
JeffR FHM Producer Join Date: Apr 2002 Location: Kelowna, BC Posts: 16,623	We didn't change anything in the penalty-killing tactics or role requirements, other than the general changes that affected all of the tactical fit scores. It may be that the offensive players you're seeing may have the correct skills to fit the PK tactic in question. And there is a routine to prioritize faceoff skill for one of the PK slots - if a good faceoff man doesn't get the first PK forward spot, it'll put more emphasis on that attribute when looking at the second one. Although that doesn't necessarily mean the player will be primarily a centre, just that he'll be decent on faceoffs.

03-08-2018, 10:56 AM	#4
Buchs Minors (Rookie Ball) Join Date: Apr 2013 Posts: 32	okay, so I didn't do a lot of testing with any other teams besides the Florida Panthers, so it is possible that this issue is limited to the Panthers, but seems more like a role evaluation issue to me. I tried all PK tactics and it seems to be by far back checking forward is the primary role used on PK. I'm not entirely sure what goes into that rating but i'm not sure it is rating defensive players well. I think this is what is causing my defensive players to not get PK time while I have the AI putting players that are pretty bad defensively in there instead resulting in very bad PK performance. I used the Florida Panthers for testing since I am most familiar with that team and that is who I was playing when I applied the latest update. before the update, my defensive players where getting regular PK time. The real panthers primary PK duo is MacKenzie and Sceviour. These guys usually got PK time in my game before the update. I think the ratings on those guys look about right to me. Sceviour especially is rated as a fairly good defensive player. Neither one of those guys make it into the AI evaluation for PK on any tactic when my roster is healthy now. consistently guys like huberdeau, Malgin, even Bjugstad get rated ahead of Sceviour although he has better defensive ratings and usually at least equal or close speed rating. I did some testing where I undressed guys ahead of them in the PK and still found that McCann, Brickley, and even Vrbata would get in ahead of Sceviour. I even tweeked his speed up a couple points to be better than a couple guys in front of him to see if that was the issue. I did the same by increasing his stick checking and position ratings by a point (which by the way was already equal or better than all the guys in front of him) and he was still behind these guys in AI evaluation. I know its a small sample size, but letting the AI simulate 3 seasons Panthers were ranked 28th, 29th, and 22nd in PK. I did 2 more seasons with MacKenzie and Sceviour as my top PK pair (like real life) and finished 9th and 16th in PK. So it seems at least likely to me that poor defensive players getting on the PK is at least hurting some teams like the Panthers. I also think that bad FO players probably having an affect also. Like I said in my original post I had a guy with a 11 and a 13 in FO together on the 2nd PK line and it was like 2 of my worst 4 forwards for face offs. Can you tell me what goes into the back checking forward rating? on a side note Huberdeau is criminally under rated for offensive roles now. he is a 3/3 for all offensive roles. I seems like his ratings look okay, so i'm not sure why he can't even eke out a 4 in any offensive role.it does put him on my first line, but seeing a legit first line guy not even able to rate as a second liner (4/4) seems weird.

03-09-2018, 02:07 AM	#5
JeffR FHM Producer Join Date: Apr 2002 Location: Kelowna, BC Posts: 16,623	I'll have a closer look at some full-season saves and try to see if I can find a pattern there, a couple of those things don't sound quite right.

03-22-2018, 09:10 AM	#8
Gerry58 Minors (Triple A) Join Date: Mar 2006 Posts: 214	Have you compared the ratings with the PK tactics available? They can change quite a bit depending on the tactics you are actually using

03-22-2018, 01:15 PM	#10
maikgianino Minors (Double A) Join Date: Mar 2018 Posts: 147	In my case my assistant coach was using a RW with (8) Faceoff skill and that was one of the reasons why my PK was so weak. I'm slowly taking care of the full tactics and lineups due that.