|
|
Post by Phil-Cornwall on Feb 9, 2023 13:53:18 GMT -5
You apparently don't know Christopher.....
|
|
|
|
Post by Chris-Suffolk on Feb 9, 2023 15:24:16 GMT -5
Your 60 goal scorer is on pace for 74 goals this year if he stays healthy. If he only plays the 73 games he played last year he’d still be on pace for 66. Which is more than what he actually got. What are you complaining about again? Oh I get it, Glenn he wants the shooting %s lower! You're absolutely correct. How can I complain? Well, some people complained about In Game Shots for goaltenders from Zones. I asked for the same thing...EXCEPT, I asked for In Game Shot from Zones for the Shooters. It was proposed, that possibly, my shooters are taking shots from areas which are not typical areas to score from, ie: from just about anywhere. I'd love to see that data. This way, I too, can prepare my team or shift the sim to benefit my team. At the very least, Suffolk is still averaging 50+ shots a game, which is above the NHL avg for one team. But, those excess shots are not accounting for more goals. That is the object of my curiosity. You'd like to know why Shitsterkin isn't stopping the Zone 10/12 shots. I'd like to know why 11 shots on goal from Matthews aimed at M-A Fleury didn't hit the twine. Was it because he fired from the Neutral Zone? Was his 2 Positioning or 3 Quickness a factor? Was his 7 Reflex or 3 Big Save the reason? Maybe, maybeee, just mayyyyybe...I can get to the bottom of this whole simulation thing. Only way I can make a determination, is to see that data. Where are Matthews Shots coming from? Tkachuk? Hintz? Svechnikov? Burakovsky? Let me see the data please! |
|
|
|
Post by Jon-Seattle on Feb 9, 2023 17:12:38 GMT -5
Well with Glenn’s sliding scale you do have a much higher chance that of scoring. A one game issue is not an anomaly as at best his goal % per shot is probably around ~15-20% in Zone 14. Based on his shot numbers he might have taken 2-3 shots from there. That doesn’t mean he has a 60% chance of scoring a goal though. Now if we look over the course of the season and he has ~100 shots from that position then I’d expect he would probably get somewhere in the range of 20 goals (+/- 5-10 goals) at that 20% mark, as that’s how probability works. It doesn’t mean you score once every 5 shot from z14. And so far Matthew’s seems to be right on his scoring pace as he should be. Would it be beneficial to know his scoring patterns to see if there is something off, maybe. Particularly if he’s under/over achieving in certain zones compared to what he should be scoring at. Considering he’s scoring right where he should be though I don’t think we’d see much of an issue anywhere. Also he’s currently sitting at 11.25 shots/goal so even taking your thought methodology he needed another shot to be able to score in that game.
On the other hand if he was only sitting at say 30 goals then you might have an actual bone to pick as you’d be about on track with what’s going on with my goalie who over the course of the entire season seems to be allowing goals at a significantly higher rate then the 10-15% he is supposedly being slotted into by the sim, unless he’s actually been seeing 15 shots a night from Z14 like his numbers suggest.
|
|
|
|
Post by Chris-Suffolk on Feb 9, 2023 19:19:46 GMT -5
Well with Glenn’s sliding scale you do have a much higher chance that of scoring. A one game issue is not an anomaly as at best his goal % per shot is probably around ~15-20% in Zone 14. Based on his shot numbers he might have taken 2-3 shots from there. That doesn’t mean he has a 60% chance of scoring a goal though. Now if we look over the course of the season and he has ~100 shots from that position then I’d expect he would probably get somewhere in the range of 20 goals (+/- 5-10 goals) at that 20% mark, as that’s how probability works. It doesn’t mean you score once every 5 shot from z14. And so far Matthew’s seems to be right on his scoring pace as he should be. Would it be beneficial to know his scoring patterns to see if there is something off, maybe. Particularly if he’s under/over achieving in certain zones compared to what he should be scoring at. Considering he’s scoring right where he should be though I don’t think we’d see much of an issue anywhere. Also he’s currently sitting at 11.25 shots/goal so even taking your thought methodology he needed another shot to be able to score in that game. On the other hand if he was only sitting at say 30 goals then you might have an actual bone to pick as you’d be about on track with what’s going on with my goalie who over the course of the entire season seems to be allowing goals at a significantly higher rate then the 10-15% he is supposedly being slotted into by the sim, unless he’s actually been seeing 15 shots a night from Z14 like his numbers suggest. Yeah, I can speculate like that too. Let's us all see the data. Then we don't need to speculate. To be honest, over the past 10 Games, Matthews has 10 goals on 106 Shots. 6 of those games he's had 11 or more shots per game. But, I'd like to see the data of where Forwards, ALL forwards take their shots from during games. |
|
|
|
Post by Glenn-Philadelphia on Feb 10, 2023 7:59:30 GMT -5
The addition of collecting detailed shot information will allow for detailed review by skater and goalie. I have been reporting some general stats and haven't had the time to design/build out a page/tool that will allow everyone to dig into the numbers. If/when completed, this would give everyone the ability to mine the data to get some of the answers they are looking for.
|
|
|
|
Post by Jon-Seattle on Feb 11, 2023 16:40:01 GMT -5
|
|
|
|
Post by Scott-New York on Feb 11, 2023 18:36:17 GMT -5
While I am all for analytics, I am once again going to point out the obvious at the risk of being accused of being dismissive. Hockey can't be explained with numbers alone. You can argue numbers until you're blue in the face, if you wish, and hope the numbers prove your point of view. Yes, I realize that in our case, we use numbers and fancy calculations to determine the outcome of a play, and I also understand the law of averages. I also don't care! Trying to solve disappointing results by understanding the calculations will completely ruin your experience. You will not solve your misfortune using this mathematical approach. You will not be a better GM because you might be good with numbers. John Chayka got fired and was touted as the most knowledgeable analytics GM the NHL has ever seen. I'm sorry if you disagree with this but I'm sorrier for you if you continue to think understanding any of this will improve your experience or the league. Disclaimer, this is an opinion piece despite already being proven as true.
|
|
|
|
Post by Jon-Seattle on Feb 11, 2023 23:49:23 GMT -5
We use an actual dice rolling system (software coded which should make it even more impartial) and thus our results should absolutely look like what we would see in a binomial distribution because because it’s made to model exactly what we are doing.
If player A has a 15% chance to score a goal and shoots 100 times then using Glenn’s dice roll mechanic (D100) he should score between 11-20 goals with a 95%+ confidence interval. Anything outside of that is theoretically possible but increasingly less likely as the number of goals moves farther away from that 11-20 mark.
The more complicated part is that the actual probability for Shesterkins Z14 goals against is likely around 10-15% but using that high water mark I feel like we can have a pretty good idea if our dice rolling mechanism is working correctly.
But hey who cares about that…what’s more fun than having a good team and watching them lose right?
|
|
|
|
Post by Scott-New York on Feb 12, 2023 9:33:24 GMT -5
By that logic, you are only taking one side of the equation into account. I understand that you are trying to get to the bottom of why Sheshterkin is not living up to expectation so it makes sense to look at what his numbers are but the 'dice roll' takes way more than into account. As I pointed out much earlier, your team was right around average as far on shots taken vs shots given up, so that is part of the equation even though you are looking only at the results of what the outcome is when a pick reaches him from a certain area. I just wonder if that narrow view will solve the problem you are inquiring about without taking into account the outcome of each and every chance category that is determined before you get to that point. What is the chance that the shot is released vs it being blocked, deflected, or missing the net. Can we even tell if the puck is being deflected by a defenseman or our own forward and that deflection is eluding the goalie? It is just that only taking into account one part of an equation and leaving the hockey out of it completely may still not lead to answers, in my opinion. I'm more questioning the expectation of this as I don't see the result you may be looking for coming from a tweak in the algorithms. If it is off, yes, let's get it fixed, but I'm not sure that even then we can just expect GHL players to match what they did in the NHL just because of a math equation.
|
|
|
|
Post by Chris-Suffolk on Feb 13, 2023 16:13:56 GMT -5
By that logic, you are only taking one side of the equation into account. I understand that you are trying to get to the bottom of why Sheshterkin is not living up to expectation so it makes sense to look at what his numbers are but the 'dice roll' takes way more than into account. As I pointed out much earlier, your team was right around average as far on shots taken vs shots given up, so that is part of the equation even though you are looking only at the results of what the outcome is when a pick reaches him from a certain area. I just wonder if that narrow view will solve the problem you are inquiring about without taking into account the outcome of each and every chance category that is determined before you get to that point. What is the chance that the shot is released vs it being blocked, deflected, or missing the net. Can we even tell if the puck is being deflected by a defenseman or our own forward and that deflection is eluding the goalie? It is just that only taking into account one part of an equation and leaving the hockey out of it completely may still not lead to answers, in my opinion. I'm more questioning the expectation of this as I don't see the result you may be looking for coming from a tweak in the algorithms. If it is off, yes, let's get it fixed, but I'm not sure that even then we can just expect GHL players to match what they did in the NHL just because of a math equation. I know you don't really care what my opinion is, but I am in agreement more with Jon than this proposal of the "dice roll." IF the dice roll takes "way more" into account than the probability of "Shot Accuracy and Expected Save Pctg" from a particular zone...that in itself might be part of the problem. Are you speculating on this Scott? Or, do you have knowledge that this is in fact true? I'm not trying to be a smart ass, I'm simply asking the question. Please be specific in your answer for me. |
|
|
|
Post by Scott-New York on Feb 13, 2023 17:55:28 GMT -5
No, I'm saying there is a separate dice roll in every single play leading up to the dice roll that is the only one in question, but that the outcome of all the plays before that have an affect but are being ignored. I remain all for fixing something of its broken, but all I've seen so far is that one goaltender is underperforming, so there may or may not be a problem with something and it is ruining the sim. I'm just not buying in, and it could be said that I don't care because I'm winning. All I'm trying to say is that there are other reasons, above and beyond a problem that may or may not exist with one goal tenders proficiency to allow goals from a particular zone and whether or not that is mathematically unlikely but if you and others are convinced that is why a certain team isn't living up to expectations, then by all means, knock yourselves out.
|
|
|
|
Post by Chris-Suffolk on Feb 13, 2023 19:13:10 GMT -5
I think more than one goalie is underperforming, and some are over performing. Of course, those results can be the randomness of the sim itself...we'd all agree to that. But, I think a few of us would be happy to see if the randomness is purely randomness and also would be satisfied to learn that the sim is working correctly and as intended.
It bothers some more than others, when you have invested the time to put together a product with expectations to perform better than the products put together of other GM's, yet the sim just negates all the effort and continues to reward the lesser. I'm not saying the odd chance of losing to inferior products will never happen...we all know it does happen. That's not my contention.
I want to see the #'s relevant to where shots are being taken during regular matches. That really is intriguing me.
|
|
|
|
Post by Jon-Seattle on Feb 13, 2023 23:21:52 GMT -5
By that logic, you are only taking one side of the equation into account. I understand that you are trying to get to the bottom of why Sheshterkin is not living up to expectation so it makes sense to look at what his numbers are but the 'dice roll' takes way more than into account. As I pointed out much earlier, your team was right around average as far on shots taken vs shots given up, so that is part of the equation even though you are looking only at the results of what the outcome is when a pick reaches him from a certain area. I just wonder if that narrow view will solve the problem you are inquiring about without taking into account the outcome of each and every chance category that is determined before you get to that point. What is the chance that the shot is released vs it being blocked, deflected, or missing the net. Can we even tell if the puck is being deflected by a defenseman or our own forward and that deflection is eluding the goalie? It is just that only taking into account one part of an equation and leaving the hockey out of it completely may still not lead to answers, in my opinion. I'm more questioning the expectation of this as I don't see the result you may be looking for coming from a tweak in the algorithms. If it is off, yes, let's get it fixed, but I'm not sure that even then we can just expect GHL players to match what they did in the NHL just because of a math equation. I know you don't really care what my opinion is, but I am in agreement more with Jon than this proposal of the "dice roll." IF the dice roll takes "way more" into account than the probability of "Shot Accuracy and Expected Save Pctg" from a particular zone...that in itself might be part of the problem. Are you speculating on this Scott? Or, do you have knowledge that this is in fact true? I'm not trying to be a smart ass, I'm simply asking the question. Please be specific in your answer for me. ^This. This algorithm is only used after a location is chosen and after a shot block is factored in so there should be no other factors that are taken into account unless I am mistaken and Glenn can correct me. As far as Deflections I believe that would be calculated in at the beginning of the shot before the save % is determined but again Glenn can correct me if im wrong. Each additional rebound shot would be calculated separately and be given a likely higher % chance of scoring already without any need for a special algorithm as it would likely be closer into the net than the original shot and gets an entirely new shot attempt, and considering our scoring seems to be somewhat high already I dont see any need to add in additional code to increase the odd of scoring due to follow on shots/deflections if it isnt already there as it doesnt seem to be an issue, but again if I'm wrong Glenn can feel free to correct me. |
|
|
|
Post by Scott-New York on Feb 14, 2023 5:24:29 GMT -5
You misunderstood the entire point or ignored it. Either way, you are correct about the way that one minute part of the sim works so at least you can take solace in that.
|
|
|
|
Post by Glenn-Philadelphia on Feb 14, 2023 9:42:16 GMT -5
My responses in blue below:
This algorithm is only used after a location is chosen and after a shot block is factored in so there should be no other factors that are taken into account unless I am mistaken and Glenn can correct me.
The big factor you aren't taking into consideration above is the shooting players Shot% rating from his card. If a guy has a 60 rating this will be a max of 6 (depending on his wind status). If he has a 63 rating, then his value will be a max of 6 (70% of the time) and a max of 7 (30% of the time). This is compared to the goalies rating to determine which column to use as the base percent chance to score.
As far as Deflections I believe that would be calculated in at the beginning of the shot before the save % is determined but again Glenn can correct me if im wrong.
I can't recall at the moment if this impacts the column used or the actual percent chance to score. It is factored in somewhere for sure.
Each additional rebound shot would be calculated separately and be given a likely higher % chance of scoring already without any need for a special algorithm as it would likely be closer into the net than the original shot and gets an entirely new shot attempt, and considering our scoring seems to be somewhat high already I dont see any need to add in additional code to increase the odd of scoring due to follow on shots/deflections if it isnt already there as it doesnt seem to be an issue, but again if I'm wrong Glenn can feel free to correct me.
Rebounds won't always end up in a more advantageous scoring spot. Sometimes these are kicked out/into the corners. Goalies rebound control is used when trying to hold onto a puck especially when his teams skaters are very tired and need a faceoff
To the point Chris made above, there is no way to measure if the sim is working "correctly" as there is not one final calculation that can be made to verify that this is the case. In math its pretty easy to check that 4+5=9. There is no right or wrong answer about approximating a players defensive composite score. The only way we can determine how accurate it is is by showing how it was calculated and discussing ways to tighten this value (make it as realistic as possible given the metrics we have access to). I am always looking to improve player ratings on cards or look into a potential problem with the simulation.
All of that being said, I will offer once again to hold a Zoom meeting for anyone to join and have their questions answered. I thing the biggest detraction of using this board to try to explain a complicated simulation is the miscommunication that takes place that requires post after post to try to explain. Even with that, it seems like 50% there still is confusion on one or on multiple parties minds after its all said and done. I am all for a real time interaction where these can be cleared up and demonstrated in real time.
|
|
