Very good discussion.

FHG, one thing I am curious about is, as mentioned by Pengwin above, that some categories are more easily found later in the draft, is this incorporated in the FHG values? That is, are the more difficult, offensive categories given more 'weight' than peripheral categories such as PIM/Hits?

I haven't checked recently but I believe guys like Ott/Clarkson have quite high (relative) FHG due to extreme numbers in peripherals, and perhaps that is given too much importance (my opinion) compared to more offensive categories. Any thoughts on this? Perhaps a comparison with more offensive/less gritty players could be worthwhile?

__________________
All I've got...all I've got is insane!

If certain categories are available later it's either one-dimensional players (goons, stay-at-home dmen) where those players are poor overall and deserve to be drafted late, or a case where GMs are incorrectly drafting their teams leaving you with better talent later.

If someone plans to draft a goon late for hits and PIM, that's fine but what about the other categories for that player? My personal philosophy is "everyone does everything" so that I am not stuck with one-dimensional players late. It allows me to go with the best player available each round. If I need to punt a category then I can make that decision according to what I draft.

__________________
UHL Calgary Flames
24-team H2H (roto G, A, GWG, SHP, PPP, PIM, +/, SOG, Hits, Blks, FOW, W, SV, SV%, GAA, SO)

Capped article archive

Montreal Canadiens coverage

Twitter:@DH_EricDaoust.

The thing about PIM/Hits/BkS is that they are newer categories in the history of NHL statistics.
There are still many older hockey fans that, if you asked them, probably couldn't name the top PIM/Hits/BkS players in the league.
But every hockey fan can name the top scorers.

We all have to realize that so many pools are just regular hockey fans (not statistical number-crunchers). For example, last year myself & horrorfan played together in a H2H pool with G,A,+-,PIM,PPP,SOG and also FOW,Hits,BkS. Myself & hf both finished top 3 and both of us dominated the competition in FOW/Hits/BkS. From what I remember, the only time I lost those three categories was in H2H match-ups against horrorfan.

Where am I going... OK... what this means is that in fantasy leagues with non-scoring peripherals (PIM, FOW, Hits, BkS) the average hockey fan does not understand the value of these categories. They draft scorers... so it waters down the overall statistics to make the league a scorer-heavy league. This phenomenon actually drives UP the value of peripheral players.

The EFFECT of this cause is that we have all gotten used to drafting our PIM/Hits/BkS players later... because we can.
We simply need to get on the boat before it sails - and we downgrade these players draft positions because we know they'll be there. But their actual value (Clarkson... or Ott) IS, indeed, very high. Computers only know a player's ACTUAL value... it does not reflect where you should be able to get them based on an assumption of GM-drafting-strategy (offensive-stat-based).

Here's an example, let's say you are in two leagues (both 12 teams, 9F):

a) In league A, the entire league drafts 4800 PIM worth of players. (approx. 400 PIM/team.)
b) In league B, the entire league drafts 1200 PIM worth of players. (approx. 100 PIM/team.)

Q: So... in which league does a 200 PIM player (Dorsett) have more value?
A: League B. Dorsett represents 1/6th of all league PIM. Therefore, Dorsett alone also represents 1/6th of the ENTIRE category value for the ENTIRE fantasy league.

Now... other things can be considered too... like you don't want to over-invest in any category.
You only need your team's statistics to be slightly better than others... enough to win H2H category... or enough to be top 3 in ROTO.

With something like FHG (or any mathematical model), there is NO WAY for the computer to forecast draft-trending in your fantasy league.
It's like... you can't ask the computer to understand what's in the head of all of the other GMs in your league. (LOL! That would be great though!)

Examples in Draft-Trending that computer can't see:
i) If everybody in your fantasy leagues is going on a goalie-run, you have to either choose to follow-suit or punt the category.
ii) If nobody in your fantasy league is rostering a PIM goon, you may only need to draft PIM through other players.
iii) If everybody in your fantasy league is rostering ONE PIM-goon, you may want to roster TWO.

One of the best tools I've heard of is FHG's draft tool where it follows along with the draft. (I haven't used this, just heard of it.)
It's a brilliant idea because it's the only true way to modify player value based on your co-fantasy-GM drafting trends.

Summary:
The actual value of peripheral players like Ott/Clarkson IS quite high in PIM/Hits leagues.
If ALL of us numbers-guys were in a league together... the draft would be very, very, very different from a draft with typical-hockey-fans.
But there is no way for a computer to know this.

For now... most hockey pools have typical fans that draft scoring. Fantasy leagues are scoring-heavy.
The effect is that a player with solid peripherals (Ott/Clarkson) can now dominate more of a peripheral category... and this makes his actual value... quite high.
Slowly, everybody will understand that a category-is-a-category and more people will understand they need to draft more PIM/Hits.
The more PIM/Hits are drafted into a pool... the lower the value of those players will actually become.

I kind of go in the opposite direction early in drafts. Before I draft, I create auction-style dollar values for every player. Since I play roto, I need to have some balance on my team so I can't simply go for the highest rated player every round or else I won't have any scorers. If I know that most people will undervalue hits and blocks and PIM, then I let those guys slide for a round or two. I know the elite scorers will be gone quickly and in roto, you need to be competitive in every category. If I want to compete in G/A/PPP then I need to get in on the ground floor. And early in the draft, the gap between the best scorer and the best overall value is smaller than it is later in the draft. It might mean passing up on Callahan/Backes/Brown early on in favour of a more scoring-heavy forward that is rated slightly lower according to my calculations. But I know I can probably get those guys a little bit later.

And there's also the fallacy that you can always find PIM/hit/blocks on the waiver wire. That may be the case, but those guys are usually absolute sinkholes in the other categories, otherwise they wouldn't be on the waiver wire (I know I'm preaching to the choir, here). You can't find a guy like Steve Ott on waivers. From my experience, the best waiver wire players are usually not heavily biased towards any certain categories. Everybody knows they need the peripherals, but they want until the last few rounds to take them, because they are so focused on drafting scorers. For this reason I usually try to finish middle of the pack in Goals, and then just dominate all of the other categories, which is surprisingly easy even in an active and somewhat knowledgeable league.

I think it'd be fun to have a league of similar-minded stat geeks/numbers guys, to kind of mix it up and see how our strategies work when there aren't a lot of biases to exploit and we're all using pretty much the same method of drafting. Of course we'd need the season to start eventually for that to happen but maybe by 2013-14 they'll have a deal in place.

So true.

I actually do this with assists. As strange as it seems, assists is the category I actually find I ignore.
Last year, I won my ROTO league by finishing top half in every category... except assists.

Brian Campbell is probably the most overrated player in fantasy hockey and Yahoo!s "Rank" calculation on him shows it.
He finished 2nd in defenseman scoring last year... but his 6-cat statistics actually placed him as only the 138th best fantasy player (18th D).
And in Hits league's he'd be even worse (with only 46hits).

Surprisingly, Yahoo! O-Rank-ed him this year at #62. (Talk about overranked!)

Ott is reasonably projected to be 5th in PIM's and Hits and 25th in FOW - along with 40 points and 10 ppp. You don't simply replace that concentration of value later in the draft with a player or two.

Offensive production may be rarer to come by, but getting the load Ott brings is equally as rare if not moreso.

Let's not confuse draft strategy with systematic player evaluation.

__________________
Yahoo weekly starts 12 team league, keep 17 plus farm

GS, W, L, SV, GAA, SV%, SO G, A, P, +/-, PIM, SOG, GWG, PPP, SHP, HITS, FOW

3C: H. Sedin,Stepan, Stastny, R. Johansen, Hanzal, Anisimov,
3LW: Burrows, Vanek, Ott, Lupul, Dubinsky
3RW: Doan, Pavelski, Brown, Okposo,
6D: Franson, Doughty, Voynov, Subban, Markov, Gonchar, Karlsson
1Util:
2G: Price, Emery


Farm: (< 100 games): S. Elliott, J. Bernier, Galchenyuk, Etem, Eakin, Atkinson

Offensive stats are more rare, but in roto I want Plekanec over Loui Eriksson.

__________________
UHL Calgary Flames
24-team H2H (roto G, A, GWG, SHP, PPP, PIM, +/, SOG, Hits, Blks, FOW, W, SV, SV%, GAA, SO)

Capped article archive

Montreal Canadiens coverage

Twitter:@DH_EricDaoust.

Shoeless has the answer exactly right:

The only reason those stats are more easily found later in the draft is because you're competing against GMs who don't understand player value and hold the inherent bias that winning the G category is more important to than winning the PIM category.

Valuation shouldn't look at the 'ease' of finding a given stat; it should look at how much a player contributes across the stats tracked in a league.

Let's not confuse draft strategy (grab those stats late because the other GMs make the mistake of leaving them there) with valuation (Ott is damn valuable.)

__________________

Fair enough. Perhaps I am linking it with draft strategy but I feel they are connected so I consider it when I value players. It doesn't mean one is right and another is wrong, it's just different. Though I disagree on the assumption that people don't understand player evaluation if some of these categories are available late. League type and size dependent, I could counter that by saying that these players may recognise the ease of finding these stats later on (late or post draft) and focus on more offense first. They can be just as successful as those who choose the multiple category route. It's not necessarily a lack of understanding, it's having a different valuation on specific players. Balance across all categories is not always 'better' than more offensive players early and peripheral focus late - I feel it's all connected to draft and team management. That being said, I've gone both routes and been just as successful.

But I'll leave it there for now. I respect everyone's reasoning and analysis though, so if you disagree that's fine.

__________________
All I've got...all I've got is insane!

It's kind of a chicken and egg situation. Although I really do believe that most people simply don't recognize the value in players that aren't big point scorers. There's so much value to be had that it makes no sense they're avoiding them on purpose.

It's funny when I run the numbers after an auction/draft and look at where each team stands in each category. So many teams are focused on only a few categories. They'll have like 20% more goals and SOG than any other team and be dead last in 4 other categories. In roto leagues, diminishing returns need to be addressed. Same applies to teams that focus way too much attention on hits or blocks or PIMs. They're rarer, but still exist. I have to make a conscious effort to NOT be that guy, haha