DF

[Chris-Wild]

So i've been playing around with some ratings and something occured to me...we change catagorical ratings such as PA/PC/SC without blinking an eye...if my top winger has a bad season he goes from 82sc to 68sc, just like that.

But our DF remains constant year to year, the odd tweak, a challenge here, if a player has a bad season (high giveaways vs. takeaways) or a really bad +/- or he is aging and is getting less and less time on the PK, why shouldn't we reflect that in our ratings from season to season?

Could you imagine if we treated other ratings like we do DF...oh so and so is a 30 goal scorer, he can't be 68sc, he should stay @ 82...the reason this occured to me is that i am trying to find a sutible formula, but shitty seasons by historically great defense players go up and down based on what we judge all our ratings by, stats.

Why is DF so special, and why do we weigh it with so much more importance than other ratings...im just curious what other's take on this is...

[bluesgm]

I agree 100% this is soemthing that Should be looked at...

[John - Sharks]

DF does change somewhat over time. The RC knocks players down and I also knock players down every full season re-rate. Guys like Jay Pandolfo, Mark Eaton, John Madden, etc. have lost their ratings over time. These adjustments are also made for SP/SK.

[Nick - Philly]

I think it is hard to like DF ratings to stats, because we look at so many different stats to determine DF ratings. There is also an eye test involved.

That being said, it does bother me that some players have a DF rating that is way too high. I would rather do something where someone challenges DF ratings to drop too.

Filip Kuba, for example, should not have 72 DF. He was rotten in his last season, but he still has 72 DF from when he played big mins and lots of PK with the Sens. Even though everyone who watched Ottawa games knew he was riding the coat tails of better Dmen he was being paired with.

[Stefan - SabresGM]

It's just because its harder to calculate. It's not an open and shut case because there isnt a DF end all and be all stat. With scoring its easy. He scored 30 goals so he should have 78SC etc Similar with PA and assists. Whereas DF there are so many different things to take into account. Blocked shots is a good start, but some of the elite dmen in the league don't block shots that often.

DF needs to be a combination stat where you can find a fair balance. Where the elite will rise to the top, above the guys that just lay down on the ice all the time. It's a very difficult one to work out. I don't know how its calculated currently, but I think perhaps it should be worked out with LOADS of things incorporated.

If you work it out like my bonus is calculated at work for instance. It's done by 30% my performance, 20% how my department did, 20% how the site did and 30% how the whole company did etc etc. So if you treat it like that you might get a decent stat. It's just deciding what is worthy that' hard.

Just an example of a potential concept is you take blocked shots as 20% and then SHTOI/G as another 20% etc, but you have to think it out a bit more. For example...

Duncan Keith:

1. Blocked Shots Individual - 121Bks - 61st in the league out of 260 so pretty good - inside the top 23% of the league. So lets say that would give him 73DF, but that only counts toward 20% of his final DF.
2. Blocked Shots Team Performance - That 121BkS was 3rd on his team out of 9 dmen. So top 33%, slightly less good. 70DF but only 10% weighted.
3. Take-aways Individual - 48TkA good for 7th in the league out of 260. Top 3% in the league, so lets call that 76DF with 20% weighting.
4. Take-aways Team - 1st on team so top 11% comes to 76DF with 10% weighting.
5. SHTOI/G Individual - 2:17 is 78th in the league so top 30%. Lets give that 72DF with 20% weighting.
6. SHTOI/G Team - 2:17 that's 3rd on the team out of 9. Top 33% so lets say 70DF. 10% weighting.
7. +/- Individual - 23rd in league so top 9% which gives him 75DF at 5%.
8. +/- Team - +15 2nd on team so top 22% good for 72DF at 5%.

So now you do the awful calculation and come out with Duncan Keith as a 73DF.

Okay so it didn't come out very high considering who it is, but i forgot to pro-rate and its a really really rough concept idea as opposed to finalized numbers. SHTOI/G would probably have to be edited as some teams take a lot more penalties etc so there would be a lot of work to do on this, but if you get it right, using the right stats, you end the need for DF challenges.


EDIT: Forgot to mention, but doing it on percentages rather than final numbers you get a good balance throughout the league. That way you don't have a case where one year someone hits 200bks so ends up 76DF, even though the rest of the league got 200bks as well. If everyone got 200bks, i would suggest they should all be dropped to a lower DF as it was clearly easier to get them that year... kind of. With percentages it fluctuates with the entire league. Kind of like an inflation/deflation scenario.

[John - Sharks]

I think it is hard to like DF ratings to stats, because we look at so many different stats to determine DF ratings. There is also an eye test involved.

That being said, it does bother me that some players have a DF rating that is way too high. I would rather do something where someone challenges DF ratings to drop too.

Filip Kuba, for example, should not have 72 DF. He was rotten in his last season, but he still has 72 DF from when he played big mins and lots of PK with the Sens. Even though everyone who watched Ottawa games knew he was riding the coat tails of better Dmen he was being paired with.

Kuba is one rating that I've wanted to correct but have been handcuffed with only being able to adjust SP/SK/DF during the full season re-rate years. Maybe a rule change where we can change SP/SK/DF every season is in order. Sweeping through the rosters, I'm pretty comfortable with what we have. I don't think anything is terribly out of line.

Maybe we can also have an email address where GM's can anonymously notify the RC of anything that they feel is off in the ratings. That way, the major errors can be pointed out.

[Chris-Wild]

It's just because its harder to calculate. It's not an open and shut case because there isnt a DF end all and be all stat. With scoring its easy. He scored 30 goals so he should have 78SC etc Similar with PA and assists. Whereas DF there are so many different things to take into account. Blocked shots is a good start, but some of the elite dmen in the league don't block shots that often.

DF needs to be a combination stat where you can find a fair balance. Where the elite will rise to the top, above the guys that just lay down on the ice all the time. It's a very difficult one to work out. I don't know how its calculated currently, but I think perhaps it should be worked out with LOADS of things incorporated.

If you work it out like my bonus is calculated at work for instance. It's done by 30% my performance, 20% how my department did, 20% how the site did and 30% how the whole company did etc etc. So if you treat it like that you might get a decent stat. It's just deciding what is worthy that' hard.

Just an example of a potential concept is you take blocked shots as 20% and then SHTOI/G as another 20% etc, but you have to think it out a bit more. For example...

Duncan Keith:

1. Blocked Shots Individual - 121Bks - 61st in the league out of 260 so pretty good - inside the top 23% of the league. So lets say that would give him 73DF, but that only counts toward 20% of his final DF.
2. Blocked Shots Team Performance - That 121BkS was 3rd on his team out of 9 dmen. So top 33%, slightly less good. 70DF but only 10% weighted.
3. Take-aways Individual - 48TkA good for 7th in the league out of 260. Top 3% in the league, so lets call that 76DF with 20% weighting.
4. Take-aways Team - 1st on team so top 11% comes to 76DF with 10% weighting.
5. SHTOI/G Individual - 2:17 is 78th in the league so top 30%. Lets give that 72DF with 20% weighting.
6. SHTOI/G Team - 2:17 that's 3rd on the team out of 9. Top 33% so lets say 70DF. 10% weighting.
7. +/- Individual - 23rd in league so top 9% which gives him 75DF at 5%.
8. +/- Team - +15 2nd on team so top 22% good for 72DF at 5%.

So now you do the awful calculation and come out with Duncan Keith as a 73DF.

Okay so it didn't come out very high considering who it is, but i forgot to pro-rate and its a really really rough concept idea as opposed to finalized numbers. SHTOI/G would probably have to be edited as some teams take a lot more penalties etc so there would be a lot of work to do on this, but if you get it right, using the right stats, you end the need for DF challenges.


EDIT: Forgot to mention, but doing it on percentages rather than final numbers you get a good balance throughout the league. That way you don't have a case where one year someone hits 200bks so ends up 76DF, even though the rest of the league got 200bks as well. If everyone got 200bks, i would suggest they should all be dropped to a lower DF as it was clearly easier to get them that year... kind of. With percentages it fluctuates with the entire league. Kind of like an inflation/deflation scenario.

That is an excellent post, some concepts i never thought of...while on the surface it sounds like a lot to compile one rating catagory, i don't mind putting in the extra stuff to get the right result.

[Chris-Wild]

I see corsi stuff posted from time to time on challenges for DF...i don't really know much about them, is there a statistic from that site we could incorporate into a DF formula? If so which one?

[Nick - Philly]

I think it is hard to like DF ratings to stats, because we look at so many different stats to determine DF ratings. There is also an eye test involved.

That being said, it does bother me that some players have a DF rating that is way too high. I would rather do something where someone challenges DF ratings to drop too.

Filip Kuba, for example, should not have 72 DF. He was rotten in his last season, but he still has 72 DF from when he played big mins and lots of PK with the Sens. Even though everyone who watched Ottawa games knew he was riding the coat tails of better Dmen he was being paired with.

Kuba is one rating that I've wanted to correct but have been handcuffed with only being able to adjust SP/SK/DF during the full season re-rate years. Maybe a rule change where we can change SP/SK/DF every season is in order. Sweeping through the rosters, I'm pretty comfortable with what we have. I don't think anything is terribly out of line.

Maybe we can also have an email address where GM's can anonymously notify the RC of anything that they feel is off in the ratings. That way, the major errors can be pointed out.

Agreed. When I look as a whole the Df ratings are pretty solid. with just a few outliers.

Love the idea of anonymous email to challenge DF and SK/SP to drop. I'm fine with that hapening in only full rerates too. 1/2 seasons are too short to get a read on DF/SP/Sk. Even more so if injuries are involved.

[John - Sharks]

Also, when you base DF on stats, you'll have players that are known to be good defensively who are perennially underrated and undervalued. The top guys might be rated fine but you'll have a ton of players that'll be lost in the shuffle. I'm not a fan. It detracts from the realism more than adding to it. I have yet to see a league's Sk/SP/DF ratings that I like more than ours so I don't really see it as a problem.

[RichBasterd]

Also, when you base DF on stats, you'll have players that are known to be good defensively who are perennially underrated and undervalued. The top guys might be rated fine but you'll have a ton of players that'll be lost in the shuffle. I'm not a fan. It detracts from the realism more than adding to it. I have yet to see a league's Sk/SP/DF ratings that I like more than ours so I don't really see it as a problem.

Agreed and some stats that are used can be deceiving. For instance, I think takeaways as a stat can be very deceptive to be used towards defensive acumen. To give you an example, a forward for Team A knocks the puck off of a forward's stick for Team B and in the process, a d-man for Team A gets the puck, thus that d-man is credited with a takeaway. Did the d-man really generate the takeaway or was it really the forward?

[John - Sharks]

Also, when you base DF on stats, you'll have players that are known to be good defensively who are perennially underrated and undervalued. The top guys might be rated fine but you'll have a ton of players that'll be lost in the shuffle. I'm not a fan. It detracts from the realism more than adding to it. I have yet to see a league's Sk/SP/DF ratings that I like more than ours so I don't really see it as a problem.

Agreed and some stats that are used can be deceiving. For instance, I think takeaways as a stat can be very deceptive to be used towards defensive acumen. To give you an example, a forward for Team A knocks the puck off of a forward's stick for Team B and in the process, a d-man for Team A gets the puck, thus that d-man is credited with a takeaway. Did the d-man really generate the takeaway or was it really the forward?

Also, blocked shots/hits/takeaways/giveaways also vary wildly from team to team and is very much based on the home arena statskeeper/team's style of play/etc... You can only really use these stats to compare players on the same team.

[Ben-Sens]

I just think if you challenge unjustifiably there should be a chance to lose points
i.e. asking for +3 in one rating can lead to a drop of 3, asking for +11 can lead to -11

[Chris-Wild]

True that each arena scores differently but that also rings true for hits which affect IT.

Perhaps an improvement is a formula that captures blocked shots for example in reference to a teams total blocked shots like stefan mentioned...that way if the team is getting overblown, it won't affect the players ratings since they are only getting a percentage of those blocked shots...that could be used for all df inclusions that are based on homers...

In order to start running some examples, help me list some stats to be used to input into a df ratings:

SH TOI
BlkS (ratio of team total)
SH Goals (for forwards mostly)
+/-
Give aways/Take aways (ratio of team total?)

[anthgrt-la]

I don't like the idea one bit. Defense is way to subjective or a stat and something that is impossible to do purely by stats or any formula. Where goals/assists are same team wide, every statistician has different interpretations on BkS/TkA/GvA. In NEFHL someone once posted the bias from arena to arena and it was enormous from some teams to others. They used the bias to factor into ratings, if we used ratings would need to calculate or figure out those bias and include them too. I don't mind the idea of being able to e-mail a few names we find to be off. DF isn't a category that in any way or form can fairly be done via stats. Yes hit are valued different from arena to arena too (looking at NYR/MIN where you touch a guy and it is a hit), but don't think IT is nearly as important as DF.