Forums/GM Game 2018-19

2018-19 GM Game - Official Scoring Thread

Aug 20, 2018 at 5:55
#1
Let'sGoBlues | WPGv4
Joined: Jun 2016
Posts: 4,277
Likes: 2,439
Edited May 16 at 2:42
Scoring Categories: data provided monthly by Corsica.Hockey

42%.....F...........Forward production, comprised of 62.5% EV and 37.5% PP
28%..D..........Defensemen production, comprised of 62.5% EV and 37.5% PP
7.5%.....CF%.......Corsi For Percentage, All Skaters, 5v5
7.5%.....xGF%.....Expected Goals For Percentage, All Skaters, 5v5
3.75%.......SV...........Total Saves
3.75%.......SV%........Save Percentage
3.75%....dSV%......Save Percentage Differential (versus Expected)
3.75%....GSAA......Goals Saved Above Average

To summarize, 70% production, 15% by shot/scoring chance contributions, and 15% goaltending.

Skater Stats: Only players with NHL stats in the month they are on your roster will count toward the scoring model.

EV Production: Even strength production will be calculated from each roster player’s goals (G), primary assists (A1), and secondary assists (A2) at even strength according to the formula below.
.....EV Prod = ((0.75 * G) + (0.7 * A1) + (0.55 * A2))

PP Production: Power play production will be calculated the same way as EV production, but using PP Points.

CF%: Corsi-For percentage is the percentage of shot attempts that belong to the player’s team while he is on the ice. It is calculated from Corsi-for (CF) and Corsi-Against (CA).
.....CF% = CF / (CF + CA)

xGF%: Expected Goals-For percentage is similar to CF% but instead of just counting shot attempts, it factors in shot quality to smooth some of the lucky/unlucky moments that tend to occur throughout the season.
.....xGF% = xGF/ (xGF + xGA)

Goalie Stats: Require a minimum of 3 games started per month, failure to meet this threshold will result in 50% of the season averages being applied for SV% and dSV% categories. The spirit of this rule is to force teams to ice NHL goalies.

SV: Total saves

SV%: Individual goalie save percentage

dSV%: Differential save percentage is the difference between a goalie’s actual SV% and the percentage of saves that would be expected given the shot quality that the goalie is faced with (xSV%).
.....dSV% = SV% - xSV%

GSAA: Goals Saved Above Average is quantifies how many more goals a team would have given up if they replaced their netminder with a “league average” goaltender.

Here is the full model, for your viewing pleasure.

https://docs.google.com/spreadsheets/d/1YpmoVUegSNXHvr66nD-XnvvFwENC9iyeRilhW_eva-M/edit?usp=sharing

Scoring Update:

Thanks to all of you who submitted your roster for preseason rankings. After counting up how much time it took me to compile the rosters that GMs sent to me, I have decided that the simpler approach for everyone will be this:

1. GMs will keep their team pages up to date with the roster that they want to be scored, especially around the 20th day of each month. You will be scored with all roster slots, including healthy scratches, for the entire month based on your roster on the 20th.

Here's an example: if you trade Martin Erat for Filip Forsberg on the 15th of the October (and he appears on your team page), you will get credit for Forsberg's stats for the entire month of October when we publish at the end of the month. However, if you make the trade on the 29th of October, since it is past the 20th it will not be included in that month's scoring for your team and you will remain with Martin Erat's stats for October.

2. A BOG member (usually me but we all have access) will enter the rosters to the scoring sheet directly from the team pages. If we have any questions or issues with your roster, we will notify you to get clarity. Otherwise, GMs can check the scoring sheet to make sure their rosters have been entered to their liking, and notify the BOG of any concerns.

Note: Just for fun, I may decide to upload the stats from Corsica more frequently than monthly, but the official rankings will come at the end of each month. I will publish an archived version of the stats in this thread each month that will serve as power rankings, and there will be a live sheet that I keep updated in between.

Live Sheet (https://docs.google.com/spreadsheets/d/11kF_rV7vT8q2KNcwX3ClKLzQaQ_Mz9igZUNyAhbkT4w/edit?usp=sharing)

Preseason Rankings (https://docs.google.com/spreadsheets/d/1LYHfK3K6ZozMg9nQwmNXoYIrQniGvkxu8YNIIm53nm0/edit?usp=sharing)

Direct any questions to the BOG thread.

Thanks,
@A_K, BOG
MrBooth and krakowitz liked this.
Aug 20, 2018 at 5:59
#2
Thread Starter
Let'sGoBlues | WPGv4
Joined: Jun 2016
Posts: 4,277
Likes: 2,439
Quoting: Max
So the only work I'd have to do is submit a roster at the end of each month?


Yes, that is correct. You may be inclined to review your team's score sheet after they are published each month to ensure there aren't any glaring errors, but that is up to you.
Pointerr21 liked this.
Aug 20, 2018 at 6:01
#3
Joined: Dec 2016
Posts: 3,111
Likes: 912
Quoting: A_K
Yes, that is correct. You may be inclined to review your team's score sheet after they are published each month to ensure there aren't any glaring errors, but that is up to you.


Im sold for the first part. As for the 2nd I'll pass since it sounds like too much work
A_K liked this.
Aug 20, 2018 at 7:16
#4
rangersandislesfan
Joined: Mar 2017
Posts: 26,878
Likes: 2,908
Are we still having a vote for the playoffs? If so, the GMs for the teams in each series cannot vote in their own series, can they?
Aug 20, 2018 at 9:16
#5
Thread Starter
Let'sGoBlues | WPGv4
Joined: Jun 2016
Posts: 4,277
Likes: 2,439
Quoting: rangersandislesfan
Are we still having a vote for the playoffs? If so, the GMs for the teams in each series cannot vote in their own series, can they?


We haven't decided yet on playoff win methods. But yes, if we have GMs vote we will set it up so you cant vote for yourself. Would you prefer using the scoring model or GM votes for playoff wins?
Aug 20, 2018 at 9:18
#6
GM - Avalanche
Joined: Dec 2016
Posts: 3,394
Likes: 677
Quoting: A_K
We haven't decided yet on playoff win methods. But yes, if we have GMs vote we will set it up so you cant vote for yourself. Would you prefer using the scoring model or GM votes for playoff wins?


GM Vote, assuming people aren't voting just to spite other GMs.
Aug 20, 2018 at 9:26
#7
Quebec GM v4
Joined: Mar 2017
Posts: 9,279
Likes: 2,205
Quoting: TMLSage
GM Vote, assuming people aren't voting just to spite other GMs.


I think that GMs are rational and responsible enough to not do this. If GMs do find that they want to vote to spite other GMs, then I would argue that they don't fit our culture.
Aug 20, 2018 at 10:23
#8
rangersandislesfan
Joined: Mar 2017
Posts: 26,878
Likes: 2,908
Quoting: A_K
We haven't decided yet on playoff win methods. But yes, if we have GMs vote we will set it up so you cant vote for yourself. Would you prefer using the scoring model or GM votes for playoff wins?


GM votes, but i hope people don't just vote by how teams are on paper.
Aug 20, 2018 at 10:40
#9
@Pointerr21
Joined: Mar 2017
Posts: 7,439
Likes: 2,528
Quoting: rangersandislesfan
GM votes, but i hope people don't just vote by how teams are on paper.


That's literally how voting works. Whoever is best on paper will get more votes.
HotsamBatcho liked this.
Aug 20, 2018 at 11:13
#10
Stars GM for V3
Joined: Mar 2017
Posts: 7,029
Likes: 1,443
Quoting: phillyjabroni
I think that GMs are rational and responsible enough to not do this. If GMs do find that they want to vote to spite other GMs, then I would argue that they don't fit our culture.


With 31/32 voters, it's too simple for only a few GMs to skew the voting. Get 5 GMs who are twitter buddies voting for each other, for example. A 6th of the league working together would tilt the top and bottom of the standings in someone's favor.
Aug 20, 2018 at 11:15
#11
Stars GM for V3
Joined: Mar 2017
Posts: 7,029
Likes: 1,443
Moving this over to the proper thread.

I'll take a closer look and get back to you. From what you guys are saying, it doesn't sound suitable for our purposes.
What would make sense, is a relative-to-team adjustment.
Predicting individual performance fluctuations, and whether a player is/could be up/down, doesn't have an impact team performance unless/until it happens. It only really serves to evaluate that player on their own.

Additionally. If I believe a player will produce better results, and I'm correct, I should be rewarded. The expected results might boost a GM's numbers if a player is performing poorly, but it would also lower the numbers of a GM who has accurately predicted that a player will outperform expectations.
The first scenario is a crutch. The latter is good player evaluation.

Now if we can modify a player's real world results to predict/reflect what they would produce with different linemates or on different teams, that would serve our purposes.
Expected values are useful for standard fantasy leagues, where scoring is based on an accumulation of individual results, but we are more concerned with a team picture than with a collection of individual snapshots.

(I have a new phone and its going to take some time for my keyboard to adjust, so if you spot a random word that seems out of place and is confusing, let me know. I try to catch them as I'm going, but inevitably end up having to cutest (see?) ... correct things later. Sticking Out Tongue)

Edit: Wondering if it might make more sense to incorporate WAR stats in place of xStats. It's probably one of the better ways to rate a player's impact on team performance.
What do you guys think?
Aug 20, 2018 at 11:18
#12
Quebec GM v4
Joined: Mar 2017
Posts: 9,279
Likes: 2,205
Quoting: ricochetii
With 31/32 voters, it's too simple for only a few GMs to skew the voting. Get 5 GMs who are twitter buddies voting for each other, for example. A 6th of the league working together would tilt the top and bottom of the standings in someone's favor.


We could also consider implementing some sort of safety net backed by the scoring system that we agree upon or a form of matchup simulation - perhaps a combination of both. If the voting is skewed in way that is too far off the model, then the voting is nullified or diluted, depending on the severity of the skew.

There are plenty of options that we could consider to keep a balance of the human element and the algorithmic element.

Regardless, it's really depends on what GMs are comfortable with. I can assure you that GMs are going to be willing to risk the opportunity of collusion over using only computer rankings.
Aug 20, 2018 at 11:33
#13
Quebec GM v4
Joined: Mar 2017
Posts: 9,279
Likes: 2,205
@ricochetii
RE: Expected Values
I'm not really sure that I understand your concern with implementing expected metrics in addition to actual results. The expected values act as a "shot quality" indicator, through our specific model that we are using (Corisca).
The xGF are one of the top predictor at an individual skater level relative to actual results (5v5 G/60, for example). The expected values will also serve the purpose of "grounding" the actual results.

RE: WAR
The WAR model that is produced by Corsica (Manny) incorporates xG metrics, shot rates, penalties taken/drawn, shooting, and zonal transitions. So the WAR would use these quality of shots, while also taking into account rates (P/60, P1/60, etc.) and other factors. Personally, it has it's upside and I derive value from using it to justify impact of a player (Oscar Klefbom), but it has it's downside. I don't particularly believe that it encompasses all data necessary to produce a metric that quantifies a player's overall contributions to a single number. Whether this is because there is data that is helpful that isn't accessible or not being tapped into, it uses other metrics that aren't weighed to my personal desire.

Again, it really comes down to the GMs and what they are comfortable with. If we were to consider it, I would standardize it the other data that is currently at 10% - this would come down to 6.66% for WAR, xGF%, and CF%. Other input is valuable at this time. I am not strongly for or against included this metric or my proposed values at this time.
Aug 20, 2018 at 11:43
#14
GM - Avalanche
Joined: Dec 2016
Posts: 3,394
Likes: 677
Quoting: phillyjabroni
@ricochetii
RE: Expected Values
I'm not really sure that I understand your concern with implementing expected metrics in addition to actual results. The expected values act as a "shot quality" indicator, through our specific model that we are using (Corisca).
The xGF are one of the top predictor at an individual skater level relative to actual results (5v5 G/60, for example). The expected values will also serve the purpose of "grounding" the actual results.

RE: WAR
The WAR model that is produced by Corsica (Manny) incorporates xG metrics, shot rates, penalties taken/drawn, shooting, and zonal transitions. So the WAR would use these quality of shots, while also taking into account rates (P/60, P1/60, etc.) and other factors. Personally, it has it's upside and I derive value from using it to justify impact of a player (Oscar Klefbom), but it has it's downside. I don't particularly believe that it encompasses all data necessary to produce a metric that quantifies a player's overall contributions to a single number. Whether this is because there is data that is helpful that isn't accessible or not being tapped into, it uses other metrics that aren't weighed to my personal desire.

Again, it really comes down to the GMs and what they are comfortable with. If we were to consider it, I would standardize it the other data that is currently at 10% - this would come down to 6.66% for WAR, xGF%, and CF%. Other input is valuable at this time. I am not strongly for or against included this metric or my proposed values at this time.


tl:dr
Why should we be grounding actual results?
Aug 21, 2018 at 12:12
#15
Thread Starter
Let'sGoBlues | WPGv4
Joined: Jun 2016
Posts: 4,277
Likes: 2,439
Quoting: ricochetii
Moving this over to the proper thread.

I'll take a closer look and get back to you. From what you guys are saying, it doesn't sound suitable for our purposes.
What would make sense, is a relative-to-team adjustment.
Predicting individual performance fluctuations, and whether a player is/could be up/down, doesn't have an impact team performance unless/until it happens. It only really serves to evaluate that player on their own.

Additionally. If I believe a player will produce better results, and I'm correct, I should be rewarded. The expected results might boost a GM's numbers if a player is performing poorly, but it would also lower the numbers of a GM who has accurately predicted that a player will outperform expectations.
The first scenario is a crutch. The latter is good player evaluation.

Now if we can modify a player's real world results to predict/reflect what they would produce with different linemates or on different teams, that would serve our purposes.
Expected values are useful for standard fantasy leagues, where scoring is based on an accumulation of individual results, but we are more concerned with a team picture than with a collection of individual snapshots.

(I have a new phone and its going to take some time for my keyboard to adjust, so if you spot a random word that seems out of place and is confusing, let me know. I try to catch them as I'm going, but inevitably end up having to cutest (see?) ... correct things later. Sticking Out Tongue)

Edit: Wondering if it might make more sense to incorporate WAR stats in place of xStats. It's probably one of the better ways to rate a player's impact on team performance.
What do you guys think?


I'm not sure if we're ready to go down the WAR path at this point, and also not sure if those data are readily available anywhere. There are skeptics of advanced stats in the GM game, and WAR is about as complex and convoluted as an advanced stat comes. Personally, I'm a fan and I'm learning more about how each model works, but for the game I'd vote that we avoid it for now.

Back to xGF%, I'm not sure if I totally understand your argument about rewarding/punishing GMs that are trying to predict a player's ups and downs. xGF% just gives us another quantitative assessment on a player's performance at the individual level (helpful when players are changing teams). A player can have any combination of good and bad real production and good and bad xGF%. I.e. we're not trying to punish overperformers and credit underperformers - the real production has nothing to do with xGF%. A player can have good xGF% and good or bad real GF% (or point totals in our model). Regardless, the points are going to be the driving force in the rankings, with 50% of the contribution coming from even strength and PP pts.
phillyjabroni liked this.
Aug 21, 2018 at 12:35
#16
Thread Starter
Let'sGoBlues | WPGv4
Joined: Jun 2016
Posts: 4,277
Likes: 2,439
Quoting: TMLSage
tl:dr
Why should we be grounding actual results?


It isnt really grounding the actuals. Look at William Karlsson. Even though he had a ridiculous GF% of 69% (nice) he wouldn't be punished by counting his xGF% because it was still a very favorable 54%.

xGF% is really just another stat to track to ensure we don't turn this into a fantasy league a la points are everything. Points are already going to be 50% of the scoring contribution, not sure if it makes sense to give them even more weight.
phillyjabroni liked this.
Aug 21, 2018 at 12:44
#17
rangersandislesfan
Joined: Mar 2017
Posts: 26,878
Likes: 2,908
Quoting: BoltsPoint21
That's literally how voting works. Whoever is best on paper will get more votes.


So that's a bit of a problem ...
Aug 21, 2018 at 1:03
#18
V3 Canucks GM, BOG
Joined: May 2017
Posts: 1,932
Likes: 652
I'm slightly annoyed that we're just finding out about this now, because it has a significant effect on player values.

For example I definitely wouldn't have traded Craig Smith for Charlie Coyle had I known how important advanced stats would be towards the standings in the game.

I also think that it's a bit unfair since Bolts probably had a pretty good idea about what the new scoring rules would be when we made the trade, while I had no clue.

Other than that I really like the idea, but I kind of feel like I should be allowed to reverse the Smith/Coyle trade.
TMLSage liked this.
Aug 21, 2018 at 1:03
#19
@Pointerr21
Joined: Mar 2017
Posts: 7,439
Likes: 2,528
Quoting: rangersandislesfan
So that's a bit of a problem ...


What's your solution then?
Aug 21, 2018 at 1:05
#20
@Pointerr21
Joined: Mar 2017
Posts: 7,439
Likes: 2,528
Quoting: rangersandislesfan
So that's a bit of a problem ...


And for starters, what does this even mean?
Aug 21, 2018 at 1:11
#21
Quebec GM v4
Joined: Mar 2017
Posts: 9,279
Likes: 2,205
Quoting: TonyStrecher
I'm slightly annoyed that we're just finding out about this now, because it has a significant effect on player values.

For example I definitely wouldn't have traded Craig Smith for Charlie Coyle had I known how important advanced stats would be towards the standings in the game.

I also think that it's a bit unfair since Bolts probably had a pretty good idea about what the new scoring rules would be when we made the trade, while I had no clue.

Other than that I really like the idea, but I kind of feel like I should be allowed to reverse the Smith/Coyle trade.


The basic breakdown has been agreed upon since we kicked off v3: 50% points (25 - EV, 15 - PP, 10 - SH); 10% - CF%; 10% xGF%, 5% - Hits; 5% - iPenalty Differential

The only proposed changes are making the Points more production based and combining the special teams into one category - reasoning: the amount of SH points in one month increments are going to be too slim to have anything more than a marginal difference in the standings. By combination, it creates a more impactful metric.

The production aspect of this formula was only brought up within the last 24 hours of discussion.

edit: this only represents the changes for skater stats. no goaltender stats changes are reflected in my above comment.
Aug 21, 2018 at 1:18
#22
Quebec GM v4
Joined: Mar 2017
Posts: 9,279
Likes: 2,205
Quoting: rangersandislesfan
So that's a bit of a problem ...


Any form of subjective rankings will be based on how well GMs perceive that team to be. Based on the rules doc, the way we are intended to vote on the teams is using in-real-life factors (fatigue, etc.).

If you'd rather use objective rankings (this model), that's perfectly fine. If you'd like to use subjective rankings (your personal evaluation of how well the team performs in the playoffs), then you're going to have to base your evaluation on whatever you use to evaluate them (eye test, stats, etc.)
rangersandislesfan liked this.
Aug 21, 2018 at 1:41
#23
Stars GM for V3
Joined: Mar 2017
Posts: 7,029
Likes: 1,443
Quoting: phillyjabroni
@ricochetii
RE: Expected Values
I'm not really sure that I understand your concern with implementing expected metrics in addition to actual results. The expected values act as a "shot quality" indicator, through our specific model that we are using (Corisca).
The xGF are one of the top predictor at an individual skater level relative to actual results (5v5 G/60, for example). The expected values will also serve the purpose of "grounding" the actual results.

RE: WAR
The WAR model that is produced by Corsica (Manny) incorporates xG metrics, shot rates, penalties taken/drawn, shooting, and zonal transitions. So the WAR would use these quality of shots, while also taking into account rates (P/60, P1/60, etc.) and other factors. Personally, it has it's upside and I derive value from using it to justify impact of a player (Oscar Klefbom), but it has it's downside. I don't particularly believe that it encompasses all data necessary to produce a metric that quantifies a player's overall contributions to a single number. Whether this is because there is data that is helpful that isn't accessible or not being tapped into, it uses other metrics that aren't weighed to my personal desire.

Again, it really comes down to the GMs and what they are comfortable with. If we were to consider it, I would standardize it the other data that is currently at 10% - this would come down to 6.66% for WAR, xGF%, and CF%. Other input is valuable at this time. I am not strongly for or against included this metric or my proposed values at this time.


RE: Expected Values @A_K

Grounding the actual results is one issue I have with them. I realize the statistics don't correlate directly as proposed, but the influence on the overall result is still the same, it only differs in degree. To keep the explanation simple, I will be using them as if they correlate directly in my examples.

A: B. Gallagher: Actual - 41, Expected - 50
B: W. Karlsson: Actual - 66, Expected - 48

A: My team should get credit for the 41. If we factor in the 50 expected, I end up gaining additional credit beyond what the player actually contributed. Why?
What's the reasoning behind adding additional credit for a player who did not produce as expected. We aren't trying to judge his future performance. W aren't incorporating his past results in order to better evaluate him.
All we want to know, is what impact does he have on our team's performance right now. Otherwise, why bother updating stats at all during the season? If we are going to use predictive metrics, we can use past results, determine the final standings now, and move right into dealing with playoffs.

B: My team should get credit for the 66. If we factor in the 48 expected, I end up losing credit because a player performed better than he should have. This is "grounding" the results. Why?
Players having career or breakout seasons, or even just a hot streak, should reward their GM for having the foresight to recognize that player's higher capacity for output and making the effort to have him on their team.

We aren't trying to predict the player's production in this case. We already know the exact numbers they have produced. Now if you wanted to make an argument to use this model for the playoffs, that would make sense. We won't have actual results for Gallagher if the real life Canadiens don't make the playoffs. I might even argue that we should not use real life results for our own playoffs, in order to keep the playing field even.

RE: WAR

This was just a thought/suggestion. There's no way to know how well Kopitar and Gallagher would perform if they played on the same line for a season. Ignoring the underlying statistics, WAR is intended as a predictor of a player's performance in any situation with any team. It's an attempt to boil down all of the player's relevant individual statistics, into a singular determination of their impact on a team's ability to win, if that player were to replace one of the weaker players on a team. It's the best statistic we have available to make an educated guess of a player's potential current impact on a team other than the one he is playing for IRL.
We can ignore it if we choose, and just rely on real life results, but realistically the chances are that Kopitar playing in place of one of Montreal's worst current centers, is going to have a positive impact on team performance. It doesn't even have to be a large factor in the overall scheme, just enough to tell the story that this player is probably going to make a difference if you put him on this team.

-------

As for the playoff voting, it doesn't have to be statistics, but I've seen too much juvenile and sketchy behavior over the course of the 3 iterations of the GM game to agree with the statement that our GM's are all rational and responsible enough to be tasked with making a fair evaluation of other teams. A selection panel combined with some other checks/balances might work, but open voting by all of the GMs just sounds like it will be a bad idea. To this day we have GM's in arguments and petty feuds over things of far less significance.
Aug 21, 2018 at 2:00
#24
GM - Avalanche
Joined: Dec 2016
Posts: 3,394
Likes: 677
Quoting: ricochetii
RE: Expected Values @A_K

Grounding the actual results is one issue I have with them. I realize the statistics don't correlate directly as proposed, but the influence on the overall result is still the same, it only differs in degree. To keep the explanation simple, I will be using them as if they correlate directly in my examples.

A: B. Gallagher: Actual - 41, Expected - 50
B: W. Karlsson: Actual - 66, Expected - 48

A: My team should get credit for the 41. If we factor in the 50 expected, I end up gaining additional credit beyond what the player actually contributed. Why?
What's the reasoning behind adding additional credit for a player who did not produce as expected. We aren't trying to judge his future performance. W aren't incorporating his past results in order to better evaluate him.
All we want to know, is what impact does he have on our team's performance right now. Otherwise, why bother updating stats at all during the season? If we are going to use predictive metrics, we can use past results, determine the final standings now, and move right into dealing with playoffs.

B: My team should get credit for the 66. If we factor in the 48 expected, I end up losing credit because a player performed better than he should have. This is "grounding" the results. Why?
Players having career or breakout seasons, or even just a hot streak, should reward their GM for having the foresight to recognize that player's higher capacity for output and making the effort to have him on their team.

We aren't trying to predict the player's production in this case. We already know the exact numbers they have produced. Now if you wanted to make an argument to use this model for the playoffs, that would make sense. We won't have actual results for Gallagher if the real life Canadiens don't make the playoffs. I might even argue that we should not use real life results for our own playoffs, in order to keep the playing field even.

RE: WAR

This was just a thought/suggestion. There's no way to know how well Kopitar and Gallagher would perform if they played on the same line for a season. Ignoring the underlying statistics, WAR is intended as a predictor of a player's performance in any situation with any team. It's an attempt to boil down all of the player's relevant individual statistics, into a singular determination of their impact on a team's ability to win, if that player were to replace one of the weaker players on a team. It's the best statistic we have available to make an educated guess of a player's potential current impact on a team other than the one he is playing for IRL.
We can ignore it if we choose, and just rely on real life results, but realistically the chances are that Kopitar playing in place of one of Montreal's worst current centers, is going to have a positive impact on team performance. It doesn't even have to be a large factor in the overall scheme, just enough to tell the story that this player is probably going to make a difference if you put him on this team.

-------

As for the playoff voting, it doesn't have to be statistics, but I've seen too much juvenile and sketchy behavior over the course of the 3 iterations of the GM game to agree with the statement that our GM's are all rational and responsible enough to be tasked with making a fair evaluation of other teams. A selection panel combined with some other checks/balances might work, but open voting by all of the GMs just sounds like it will be a bad idea. To this day we have GM's in arguments and petty feuds over things of far less significance.


I agree with Ricochetti
Aug 21, 2018 at 8:55
#25
Thread Starter
Let'sGoBlues | WPGv4
Joined: Jun 2016
Posts: 4,277
Likes: 2,439
Quoting: ricochetii
RE: Expected Values @A_K

We aren't incorporating his past results in order to better evaluate him.
All we want to know, is what impact does he have on our team's performance right now.
.
We aren't trying to predict the player's production in this case. We already know the exact numbers they have produced.


Don't let the word 'expected' skew your interpretation of xGF. xGF% does measure the impact a player has right now. It uses real, measured shot impacts and says whether that player is producing quality chances at a good or bad ratio to what they're giving up. It can be used as a predictive stat but also as a descriptive one of past play. If we eliminate xGF% from the standings, we will lose that piece of the evaluation... pts (or +/-) won't show the entire impact that player is making on the game. That is why we proposed also tracking CF and xGF.

And if you're worried about losing credit due to differences in point production vs xGF%, keep in mind that the real production will be worth 3x more than the xGF%. It is proposed to be a small player in the grand scheme of things.
NateElder12 liked this.
 
Reply
To create a post please Login or Register
Question:
Options:
Add Option
Remove Option
Submit Poll