DVHL preliminary re-rates

[PasiK]

DVHL preliminary re-rates:

http://www.dvhlsim.com/fhldvhl8/TRANSFER/

look towards the bottom, dmen and goalies

Click to read more...

 

[SPG]

DVHL preliminary re-rates:

http://www.dvhlsim.com/fhldvhl8/TRANSFER/

look towards the bottom, dmen and goalies

Thanks for posting. It sounds like we're going to be using DVHL as our ratings base again for 2007-08.

I do have to say that some of those ratings are just plain odd... Ari Ahonen a 66OV? The guy never even played an NHL game...

And ERIC FICHAUD is still being included in their ratings package... wow.

Things just haven't been the same since Zubial stopped producing ratings.

 

[SPG]

Dan McGillis is a 72OV??? He spent the entire 2006-07 season in the AHL.

These are looking worse than last years

 

[Ohio Jones]

The OV rating for defencemen is overly influenced by the ST stat, which is why we see a guy like McGillis with a higher OV than expected, although his individual ratings are not strong. Overall, though, I agree these are pretty brutal - once again, the IT and DF ratings are just plain stupid. Let's hope we can find some alternative leagues.

As far as the goalies go, I should have those ratings ready in a couple of weeks, based on teh calculation system I put in place last season.

 

[Toronto_AGM_Adil]

Perhaps we should come up with our own rating system for defencemen? From looking at the DVHL thread it appears they have a formula related to NHL individual/team stats and I'm sure we can use those same stats to come up with our own weights that seem more fair. All the relevant statistics are available on the NHL website... I don't mind mind putting some time into this (since it'll be a long summer anyways)

 

[Ohio Jones]

It'd be greatly apprecited, Adil. I do something similar with the goalies...

 

[Toronto_AGM_Adil]

It'd be greatly apprecited, Adil. I do something similar with the goalies...

I can put the statistics together but I fear the weights are out of my depth Doug. I'm not sure how the various ratings directly correspond to the FHL statistics, is there a general rule for this?

If I can get some help with this portion I have no problem crunching the numbers and doing the leg work...

 

[chump06]

I can put the statistics together but I fear the weights are out of my depth Doug. I'm not sure how the various ratings directly correspond to the FHL statistics, is there a general rule for this?

If I can get some help with this portion I have no problem crunching the numbers and doing the leg work...

I also am on holidays, perhaps we can work together? Josh said he'll give me a hand if he has time...

 

[Wildman]

I think it is a great idea for someone to create our numbers. I know last season i had lot of issue with the rating.

 

[Ohio Jones]

I can put the statistics together but I fear the weights are out of my depth Doug. I'm not sure how the various ratings directly correspond to the FHL statistics, is there a general rule for this?

If I can get some help with this portion I have no problem crunching the numbers and doing the leg work...

Do you mean in terms of "giving a player this SC rating should result in XX number of goals"? We don't know either. The important thing is for ratings to make sense relative to one another (a very subjective exercise in most cases, although ), and for the ratings assigned to reflect a bell curve in terms of distribution, and a "hockey stick" curve in terms of absolute ratings.

What do I mean? Most players' skill ratings should fall between 60 and 70 (maybe between 60 and 75 - Drew?), with decreasing numbers above and below that bracket - the "bell". And the ratings for stars should be demonstrably better than for most other players - the "hockey stick".

We try to work with a weighted three-year average when comparing stats - 50% for the most recent season, and 25% for each of the previous two seasons.

I actually come up with the ratings first to ensure I've got the right distribution and separation, then sort the players into them based on their weighted performance. Of course, that works for stat-driven ratings like SC, PA and the key goalie ratings (SP, PC), but many of the ratings are truly subjective, such as IT, DF, EX and LD. Even SP and SK (which shouldn't always be the same number!) are very subjective. There we just try to use common sense.

Rob and Nick have taken on the IT and DF ratings in the past, and I know they're near and dear to Drew's heart as well. They would likely have some suggestions to offer in those cases.

Radical sugggestion: Maybe we should come up with a "master distribution" for skaters - the standard range of ratings available for any given stat - and then have each person assigned to a stat go through and sort the players into the distribution for that stat. That way instead of having to come up with a number, you're only having to come up with a relative position - the number follows automatically.

Thoughts?

 

[Toronto_AGM_Adil]

Do you mean in terms of "giving a player this SC rating should result in XX number of goals"? We don't know either. The important thing is for ratings to make sense relative to one another (a very subjective exercise in most cases, although ), and for the ratings assigned to reflect a bell curve in terms of distribution, and a "hockey stick" curve in terms of absolute ratings.

What do I mean? Most players' skill ratings should fall between 60 and 70 (maybe between 60 and 75 - Drew?), with decreasing numbers above and below that bracket - the "bell". And the ratings for stars should be demonstrably better than for most other players - the "hockey stick".

We try to work with a weighted three-year average when comparing stats - 50% for the most recent season, and 25% for each of the previous two seasons.

This makes a lot of sense. I think we can come to an agreement for the number of NHL roster players for each position (12 forwards x 30 teams, 7 Dmen x 30 teams, 2 goalies x 30 teams) and then come up with a bell curve based on these numbers for each stat. For example 360 forwards would make up the top 4 lines of all teams combined, I'd say atleast 10-15% should be 80+, and maybe the bottom 10-15% should fall under the 60 mark.

I actually come up with the ratings first to ensure I've got the right distribution and separation, then sort the players into them based on their weighted performance. Of course, that works for stat-driven ratings like SC, PA and the key goalie ratings (SP, PC), but many of the ratings are truly subjective, such as IT, DF, EX and LD. Even SP and SK (which shouldn't always be the same number!) are very subjective. There we just try to use common sense.

Roughly, the stat categories I was thinking related to stats:
IT - PIM, Hits
ST - Weight/Height
DI - PIM (It would be nice to find something better then this...)
PA - Assists
PC - GvA?
DF - TkA, +/-, Block shots, SH TOI
SC - Goals
EX - GP, Age

If anyone has an idea of another stat to add to this list that would be great. I'm not sure right now where to put PP TOI, SH goals, GW goals, etc...

I'm not sure what to do with the subjective stats like LD, SK, SP... I guess take whatever was done last year modified with the DVHL stats? Also, the DVHL stats seem to be modified to take into account the teams overall stats such as team +/-, team scoring, etc... doing this for all of our stats may help average out the impact of a winning team on a particular players stats. I'm not saying we should completely normalize (since that would negate the fact that the winning team got there through that player) however a slight counter weight (maybe 25%-33%) might be realistic.

Rob and Nick have taken on the IT and DF ratings in the past, and I know they're near and dear to Drew's heart as well. They would likely have some suggestions to offer in those cases.

I would love their input on this...

Radical sugggestion: Maybe we should come up with a "master distribution" for skaters - the standard range of ratings available for any given stat - and then have each person assigned to a stat go through and sort the players into the distribution for that stat. That way instead of having to come up with a number, you're only having to come up with a relative position - the number follows automatically.

This would be a great help! If we could get a group of volunteers (including myself) to sort through and collect these stats that would be great. I think I have alot of the stats from the pervious NHL seasons, however the more statistics we can gather the more accurate our rating system will be.

 

[Ohio Jones]

Roughly, the stat categories I was thinking related to stats:
IT - PIM, Hits
ST - Weight/Height
DI - PIM (It would be nice to find something better then this...)
PA - Assists
PC - GvA?
DF - TkA, +/-, Block shots, SH TOI
SC - Goals
EX - GP, Age

If anyone has an idea of another stat to add to this list that would be great. I'm not sure right now where to put PP TOI, SH goals, GW goals, etc...

I'm not sure what to do with the subjective stats like LD, SK, SP... I guess take whatever was done last year modified with the DVHL stats? Also, the DVHL stats seem to be modified to take into account the teams overall stats such as team +/-, team scoring, etc... doing this for all of our stats may help average out the impact of a winning team on a particular players stats. I'm not saying we should completely normalize (since that would negate the fact that the winning team got there through that player) however a slight counter weight (maybe 25%-33%) might be realistic.

I've already got the goalie tracker set up, I just have to enter thsi past year's numbers, so don't worry about those stats.

Not all stats available relate directly to ratings. TOI relates to EN. Games played (as a percentage of games available in their league) speaks to DU - the more games played, the higher the DU. DI is straight PIM. IT tends to factor in Hits, PIM plus a subjective measure. Giveaways, takeaways and Faceoff win/loss are stats that might be best applied to PC for skaters. Historically that stat was just the average of a player's PA and SC ratings.

I'm not sure we need to factor in SHG, PPG, GWG, etc., as there's no rating that seems to reflect "clutch" play. (Although I suppose an argument could be made for LD, but it's a mostly subjective - and disputable - rating anyway...).

 

[Toronto_AGM_Adil]

I've already got the goalie tracker set up, I just have to enter thsi past year's numbers, so don't worry about those stats.

Not all stats available relate directly to ratings. TOI relates to EN. Games played (as a percentage of games available in their league) speaks to DU - the more games played, the higher the DU. DI is straight PIM. IT tends to factor in Hits, PIM plus a subjective measure. Giveaways, takeaways and Faceoff win/loss are stats that might be best applied to PC for skaters. Historically that stat was just the average of a player's PA and SC ratings.

I'm not sure we need to factor in SHG, PPG, GWG, etc., as there's no rating that seems to reflect "clutch" play. (Although I suppose an argument could be made for LD, but it's a mostly subjective - and disputable - rating anyway...).

What about shooting percentage? I'm not sure what effects the game in terms of number of shots taken... is it the SC rating or a combination of multiple ratings? I'll try for early next week to get the stats from the NHL page in order to start this process. BTW, does anyone know of a better source for stat tables then the NHL.com website?

I've heard very little comment on this from anyone else in the league besides a handful of GMs... is there a real interest in this or would everyone rather use the DVHL stats? I'd hate to put in the work and find this out later...

 

[Ohio Jones]

What about shooting percentage? I'm not sure what effects the game in terms of number of shots taken... is it the SC rating or a combination of multiple ratings?

I think there are a number of factors that contribute to shooting percentage, including ATOI, linemates, role on the team, etc. Players with limited opportunities to shoot often have a higher shooting % than players who play a regular shift, so I'm not convinced it's a good measuring stick. Perhaps it could be used to help rank players who are otherwise comparable in terms of goal production (although goals per game might be even more instrustive...)?

I've heard very little comment on this from anyone else in the league besides a handful of GMs... is there a real interest in this or would everyone rather use the DVHL stats? I'd hate to put in the work and find this out later...

There are always a few GMs who step forward more tha others for administrative duties like this, but it's also the height of summer vacation season, so that may be playing a role as well.

 

[Dr.Sens(e)]

It is much easier to use DVHL and perhaps adjust a few specific ratings like we have in past than to try and start this from scratch. The bottom line is, re-rating all of these players for each category is a job required of 4-6 people with a lot of time to apply to apply themselves. And even then, we'll likely end up with something like DVHL, but with a different set of complaints. I haven't looked at the new ratings yet.

 

[M3zin]

FYI, the Forward re-rates are up on the DVHL site. Is it just me, or are the defensive re-rates skewed terribly. It almost seems as if the re-rates are based on point production alone given the amount of high-calibre forwards, with few good-very good defensemen.

 

[Tampa GM]

Lets keep the DVHL ratings, if we do my entire first line with Sedin-Nylander-Hossa will be rated 80 or higher

 

[Ohio Jones]

Actually, the forward ratings don't look nearly as bad for some reason. Or is it just me? The defenceman ratings still need major work, though, especially (as usual) IT and DF...

 

[Dr.Sens(e)]

They are similar to last year, which makes sense given they seem to have used the same methodology. I say we use these with some adjustments made to IT and DF ratings, amongst others. I noticed some of the fringe players are probably rated a little higher than usual, but that might be ok, as fringe NHLers should at least be usable as call-ups (rated 65-69) versus the 55-60 ratings they used to get.

 

[Default101]

They are similar to last year, which makes sense given they seem to have used the same methodology. I say we use these with some adjustments made to IT and DF ratings, amongst others. I noticed some of the fringe players are probably rated a little higher than usual, but that might be ok, as fringe NHLers should at least be usable as call-ups (rated 65-69) versus the 55-60 ratings they used to get.

i totally agree with you on that one, the guys who are 6-8 defencemen should be at 65-69 instead of the 55-60 they had before, it just makes more sense

 

[Toronto_AGM_Adil]

Just a suggestion, but I was thinking maybe rather then re-rating some of DVHL why not leave the DVHL stats alone and increase the number of arguments again? This moves alot of the rerate work back onto the GM's themselves.

Again, I know I've been the one pushing for trying out our own rating system, however as a second option I'm thinking more arguments will be easier then sifting through the DVHL ratings and will lead to many fewer complaints...

 

[SPG]

Just a suggestion, but I was thinking maybe rather then re-rating some of DVHL why not leave the DVHL stats alone and increase the number of arguments again? This moves alot of the rerate work back onto the GM's themselves.

Again, I know I've been the one pushing for trying out our own rating system, however as a second option I'm thinking more arguments will be easier then sifting through the DVHL ratings and will lead to many fewer complaints...

Having our own rating system is the ideal way to go, but it just hasn't been practical in the last few years. Even with the way we have been doing it, we haven't been getting our seasons started until December/January. Maybe going with DVHL is the way to go, but with Douglas' goalie adjustments and more challenges. Its worth discussing anyway...

 

[Ohio Jones]

In case anyone missed it, I've converted the the DVHL skater ratings to an Excel file for ease of reference. They're in the Files, in the "2007 Ratings" folder. I'll add the modified goalie ratings when I get a chance to crunch those numbers.

 

[Toronto_AGM_Adil]

Having our own rating system is the ideal way to go, but it just hasn't been practical in the last few years. Even with the way we have been doing it, we haven't been getting our seasons started until December/January. Maybe going with DVHL is the way to go, but with Douglas' goalie adjustments and more challenges. Its worth discussing anyway...

Again, if we have a rating which we all question I can try and pull the NHL stats and re-create to whatever weight system we feel is fair. I havn't heard of anyone having any serious problems with the ratings except in reference to the overall defencemen stats.

I really like Ohio's earlier suggestion of the ranking players and applying a normal distribution... If you guys are worried about the amount of work or worried about a delay to the start of the season I don't mind putting in the effort on a couple of the more disputable stats, rank them based on NHL stats and apply the Ohio Distribution to it. DF looks like a ripe reason for this kind of stat ranking since there's so many factors which go into it and everyone hates the DVHL stat.

 

[Ohio Jones]

...rank them based on NHL stats and apply the Ohio Distribution (TM) to it.

Fixed.