Hey all,
As Doug and I are finishing up the edits to the Covenant to reflect the recent votes, I noticed that our definition of the Omnivore "baseline" is really not clearly defined in the Rules.
Can we agree without too much effort on a suitable definition to include in the final document?
I believe there was general consensus that the highest pointing players at all 3 forward positions, 6 defense, and one goalie were to count.
There was the suggestion that neither TG/PIM or plus-minus be considered for the baseline, as this reflects coaching decisions through the year (much like line-ups). Thus doing a good job throught the year with judicious use of TG and plus-minus improves your Omnivore score.
Does this sound acceptable? Other thoughts?
Thanks!
Dan
23 comments:
Hey Dan,
I think you have it right - the only thing I'd comment on is whether we should be using the draft order to determine who gets counted, rather than best points of the players available.
By using draft order GMs would get full credit for having decent back-ups who could dress, and it would be a market determined order. By using the best points available we in effect discount the advantages a team gets by drafting well.
Thoughts?
Actually, I disagree. Drafting well can also include scooping a guy in round 20 who may in fact be your 3rd line guy over the season. You may draft him, being savvy, at a later round because he's undervalued, not because you think he's worse than the guys you drafted ahead of him.
Thus using the best points methodwill CREDIT a team for drafting well, not discount it
By your own example, the smartness of holding back and drafting a guy late isn't recognized by best lineup system - i.e. he makes the active roster on merit of his points, and not based on what the market said he was worth when he was picked.
I argue that by using the draft as the baseline, we can reward both intelligent prospect additions (they are not part of the baseline), and solid drafting.
The other value of using the draft as the baseline is that it isn't arguable whether a player should be included. He's either part of the active lineup by virtue of when he was picked, or he isn't.
The first likely objection is that a player who gets hurt (Nylander) or even a player who is drafted hurt (Stillman), will count as a zero on the baseline, when they shouldn't.
I'd counter argue that the Omnivore is meant to represent the greatest team improvement from the draft, including prospects, tg decisions, trades, lineup adjustments, etc.
So in the case of Nylander (injury prone), the fact that he is injury prone and that the GW's recover from this (at whatever level they are able to through Nylander's replacement in the event of injury) is recognized in my suggestion, but not in the current system.
In the case of Stillman, the objection is strange. If Stillman comes back from his injury early and performs well, he would still be on the baseline by virtue of where he was picked (and as such the Omnivore score isn't improved). If he never recovers from his injury than we have a setback that you must recover from - and to the extent that you do recover from it or not, the baseline would reflect that.
The flipside to this arrangement would be the 'perfect' baseline. Something we toyed with in the past. You take the best of the teams assets retroactively, and you include prospects and tgs (at the perfect level). The perfect baseline then represents the apex of what you might have accomplished with what you started with - and the calculation is then between how you finished and this baseline - and the difference recognized is that made almost exclusively by trades, coaching decisions (but with the exception of +/- guys, only as a negative) and WD roster moves.
Currently we use a blend of perfect baseline (in that we take the best players available) and what I argue for (where the market determines baseline).
The result is that we don't have a coherent philosophy behind the Omnivore. It seems to reflect improvement, and certainly seems to reflect those GMs who made the most difference in terms of trades, but I think there is a general uneasiness about what the Omnivore means.
Personally, I think we need to sit back and think about what we want it to do.
Therein lies the rub, to be sure. As I recall, the concept of the Omnivore was to reward the GM who improved his team the most over the course of a season.
To my mind, that means the team that improves the most from what they have at the end of the Entry Draft.
The idea as I recall was to compare to the standard pick em and watch em office pool, and how well a team could do on autopilot.
Actually, as I write this, I recall other issues, such as trying to use first week lineup (flawed for players drafted injured) and so forth.
How about the following:
After the ED, each team submits a line-up that is their Omnivore baseline (ie the lineup they think will garner the most points over the season on autopilot). This can include TG and +/- if desired, but should only use the roster as of the end of the ED (no prospects promoted etc).
If we can trust that all GMs will submit a lineup they genuinely believe to be the best, it should work. One way to help ensure this would be to recognize the highest pointing baseline lineup submitted (ie the winner of the Autopilot pool) with something - an extra ballot in the Slot Selection Lottery perhaps?
I like it...
Yeah, the Omnivore debate has had a wide variety of permutations over its convoluted history. In all cases the issue breaks down to;
'What do we want to recognize as 'improvement'?'
It's a tougher question than it first looks.
Here are some of the baseline options we have or could consider;
- 'Perfect'
This means the Omnivore rewards the GM who makes the biggest difference between a perfectly coached team and the reality of the league with trades/WD's etc.
- Draft order (or 'Market')
This means the Omnivore reflects changes made including lineups, prospects, WD's, TG's, etc. between reality and what you drafted.
- GM selects
I think no matter how this is constructed placing the lineup into GM hands to determine is fraught with complications, not the least of which is that there is an incentive to select the worst possible lineup for ones Omnivore.
- 1st lineup
This tends to include high-end prospects (who are promoted in week one), TG's and some +/- players, so the Omnivore would reflect the difference between wk 1s lineup for the whole year, and reality.
Each methodology creates an Omnivore award that charts some sort of progress between reality and where we started off, but it isn't clear to me that any of the methods is obviously the right way to go, and I'd suggest we consider each of them in their rights as possibilities until we reach some sort of conclusion as to which is the best for what we want the Omnivore to do.
Yes, there are a multitude of flawed options. I would like to think that there is sufficient honour among the pool GMs that no-one would deliberately submit a bad baseline lineup just to win. However, if we have a sufficient motivator to teams to aim high (eg highest pointing submitted baseline receives 2-3 (however many is sufficient to get interest) extra ballots in the slot selection lottery in addition to their allotted number based on finishing order.
I think this would reflect how a GM does with trades, lineups, prospects etc over what they would have scored using a roster for a pick 'em and leave 'em pool on autopilot.
Don't you think that is sufficient reason for valid baseline submissions, without being arbitrary?
If not, then I propose the following unusual, but well intended option:
Take rosters at the end of the ED. Leave out TG and +/- as they would count as good coaching, and should improve your Omnivore score. At the end of the season (or any point along the way), perform the following calculation:
Each forward position - Add the points of all four players and multiply by 0.75
Defense - Add all 8 players and multiply by 0.75
Goal - Add both players and multiply by 0.5
This gives figures in range with what a dressed roster looks like, while giving credit for well drafted 4th liners
Thoughts?
OK the more I think about it the more I like it. Using rosters as of the end of the ED:
Omnivore baseline = (All LW points x 0.75) + (All C points x 0.75) + (All RW points x 0.75) + (All D points x 0.75)+ (All goalie points x 0.5)
Simple. Equitable. Unambiguous. TG and +/- count towards improving your score above baseline thus rewarding good coaching.
All in favour?
Yes Doug, the official time for rule changes has passed. However we can make a change if everyone agrees to it. I think it is worth a shot...
Damn Dan is fast,
I deleted my rushed post almost as soon as I put it up there because I wanted to clarify that I think this is a very valuable contribution to the whole Omnivore debate, I am however concerned that we may be opening up a hornet's nest three weeks before the ED - something that I think we would all wish to avoid.
I think your proposed solution (a modified tally of all players drafted) is really clever and I would heartily support it. If you can get unanimity, by all means have at 'er.
If not, lets remember this discussion next summer as I think its time to get this major award's governing principles settled.
Thanks for joining the fray Rob - lest we keep talking to our own echos.
Brian will need to be consulted here but I had thought that we had been using the first week's line-up for the last few years as the baseline. An imperfect solution to be sure but I think it is generally accepted that GMs are unlikely to sabotage their chance at the Predator Cup for a better chance at the Omnivore.
As to the broader question of the purpose of the Omnivore, their are two basic ways of looking at "improvement" in my opinion. Improving your team and improving your franchise. They are not the same thing.
Your team, as set out in your active line-up (which I agree should include TGs and maybe even +/-, though this is fraught with some dangers) is how we measure pts in the Predator Cup. It is not an arbitrary arrangement. Fourth liners are not irrelevant but if you are playing then you are very lucky (its good luck if they have a breakout and its bad luck if one of your top-liners had a bad break). Certainly good/fortunate line-juggling can eke out additional pts from the same set of players by benching players at the right time and catching a few weeks of a back-ups' scoring streak, but essentially the real improvements are at the GM level - trades, waiver draft/wire acquisitions and developing prospects.
It's at this point that we need to distinguish between improving one's franchise and improving one's team. Drafting Nicklas Backstrom in this year's prospect draft is a clear improvement to that GMs' franchise but, given that he will play in Europe this year, does very little to improve the drafting GMs' team this year.
The Omnivore can not capture the value of the improved franchise, only the value of the improved team so the real question is how to set the baseline.
I like Dan's suggestion of doing it in such a way as to capture all four lines. I think that there is also some merit in using the week one line-up as well. There is some symetry in measuring the improvemnt from the "opening night" line-up (though clearly one misses out on the injured/locked out superstar who returns to the team two weeks later). Perhaps the two could be combined? 75% of all skaters (active and benched) of the week-one line-up and 50% of the goalies.
Doing this sets a "natural" starting point to begin to measure from - the same starting point as the Predator Cup uses - catches the injured/locked-out stars and will also catch the obvious prospects who are no longer truly developing but have clearly already become valuable contributors to a GM's team while at the same time allowing an astute (read fortunate) GM to get the Omnivore "credit" for a prospect who exceeds expectations and, through his play, demands a place on the roster part-way through the season.
Thoughts?
WW: The baseline cannot be a submitted lineup as it is to subjective which is something we are trying to avoid.
CH: Agreed.
WW: To me it should be your best total point line up of your drafted players at the end of the entry draft.
CH: Damn, it was almost four things we agreed on in a row! I ALMOST agree to this, my problem is that I'm not convinced it should be the BEST lineup, and couldn't better be handled with a straight forward lineup derived from the draft. I like the idea that the market determines the baseline (homo economicus that I am) and I further like the idea that some improvement by prospects be considered for the award, more on that in a bit.
WW: Now I think as we currently always have done we should include to TG minutes as this is a requirement of the lineup.
CH: Agreed.
WW: Now I also think that you should also take into account Propects and plus minus in the base line because these are parts available to you with out going outside your team.
CH: Again, I half agree. Yes to prospects, no to +/-, but I think on the whole Rob and I are pretty much on exactly the same page.
WW: Staying with current system of best lineup plus 2tg out of drafted roster should not be changed until we have a concensus on the meaning of the award and how to change its calculation.
CH: Agreed. With apologies to Dan, it's impossible for us to fairly blow up the Omnivore and force-feed a vote on it prior to the draft. I think we can muddle through with a blend of common-sense and precedents recalled to handle this draft, and if the blog continues to rock as it has so far, I wouldn't be surprised if we don't have it worked out prior to the draft at least some sort of coherent theory as to how to move forward. Heck, I'll settle for knowing the right questions to ask everyone to get us pointed in the right direction. There is a lot to consider with regards the Omnivore, and I want to chew each piece carefully before I decide to swallow anything (even/especially my own cooking).
Damnit! It chopped off the first half of my post! AAARRRGH
No worries Cam, I only wish we'd thought about it sooner. To be honest, using week one lineup is inane, and if that is how it has been done, I think the meaningfulness of the Omnivore these last few years is faint (apologies to those who won it).
Obviously it is very unclear what the definition of the award is construed to mean, so we are trying to define an answer when we don't even know the question.
All I can say to that is "42".
Obviously we won't be able to change for this year, so we'll have to do what we've done before (whatever THAT might be - personally I thought it was best possible total from post-ED roster) and the award will remain a misguided and ill-defined achievement.
Put it down now for post-season debate...
I find myself agreeing with Rob.
Using the snapshot we take at the end of the draft (including prospects) and comparing that to the end of the year total is both simplest and fairest way to determine a baseline active roster. (It's a 'stupid baseline' as no input goes into creating it).
Having that in agreement, I think we also agree that TG's should be assigned (and I'd argue for the moment that these should be from those already on the active roster baseline by virtue of their points at the end of the season).
However, I disagree that we should find a way to project +/- into the Omnivore, we don't require anyone dress a player that way, so no need to force the issue in the Omnivore should exist. It's the purest coaching decision that exists, in that it is utterly voluntary, and bears a high risk, and I'd prefer to reward Gms who make the right call.
OK, alreadythere is a lack of clarity:
1) end of the ED, including prospects - At the end of the ED, no prospects have been promoted. Only some P4's have been drafted as RFAs in the late rounds. Prospect promotion comes after. So it can't be post-ED including prospects - it can be post ED including drafted RFAs who were once prospects, or it can be more like a first line-up post prospect promotion. Not both.
2) those on the active roster baseline based on points at the end of the season - Which is it? There is the active roster baseline snapshot post ED, OR there is the end of season retrospective on best performing players. It makes little sense to mix apples and oranges and do half of one and half of the other. Besides, does that mean you count the two players with the most PIM that make the roster by being the higher pointing players over the season? What if counting PIM meant that the lowest pointing player (eg pure goon) actually becomes an "active" player IF you count his PIM?
3) TG should be assigned - by whom?
I think we need to back up the bus, try not to run over one of Cam's right wingers who always stand too close, and first figure out what we want the award to reflect.
THEN we can get about designing a system that best reflects that goal.
My mistake, the last few seasons we have been using the best line-up from a team's drafted players. TGs were included but prospects were not.
I share Rob's view that certain prospects who are clearly part of a GM's planned line-up should be included in the baseline, but I also feel that "surprise" prospects shouldn't be.
Crosby and Staal, the top two prospects last season, are good examples of this difference. There was no question that Crosby was going to be dressed by me and probably would be my number 3 or number 2 center. If I didn't have Thornton as an FP, Crosby would have been my number 1 center. Properly he ought to have been on my baseline.
Staal, on the other hand, was a big surprise to Brian. It took weeks before Brian finally convinced himself that Staal was for real and promoted him. This was a player who had developed faster than expected and falls into the catagory of prospect that I think a GM ought to be able to benefit from in the Omnivore.
The question is how do you distinguish between the two different types of prospects - assuming that you see them as different - I would say that prospects promoted in the first week are clearly "part of the plan" and should be included. Prospects promoted much later are, IMO, not.
Dan said:
1) end of the ED, including prospects - At the end of the ED, no prospects have been promoted. Only some P4's have been drafted as RFAs in the late rounds. Prospect promotion comes after. So it can't be post-ED including prospects - it can be post ED including drafted RFAs who were once prospects, or it can be more like a first line-up post prospect promotion. Not both.
CH: Agreed - however, Rob correctly restates the position I intended to take, that the baseline is created after the end of the prospect draft (which I more confusingly called 'the end of the draft'). This ensures that high end prospects like Crosby, etc. are counted (and counted based purely on their performance).
Dan then said:
There is the active roster baseline snapshot post ED, OR there is the end of season retrospective on best performing players. It makes little sense to mix apples and oranges and do half of one and half of the other.
CH: As I see it, the baseline is a retrospective assessment done at the end of the year. The active roster baseline would be the best players at their positions over the course of the season. So the players are determined by the snapshot at the end of the draft, but the actual roster isn't finalized until we see how the season plays out.
GW: Besides, does that mean you count the two players with the most PIM that make the roster by being the higher pointing players over the season? What if counting PIM meant that the lowest pointing player (eg pure goon) actually becomes an "active" player IF you count his PIM?
CH: A fair point. It certainly is possible to assess the two best TG's irrelevant to actual active roster points and then determine who should play based on the adjusted stats. If the two best goons wouldn't dress based on points, but would based on TG adjustments, then it makes sense to use the adjusted TG stats to determine baseline.
GW: TG should be assigned - by whom?
CH: Our statistician based on performance. The downside is that some players become 'unexpected' goons for the Omnivore (i.e. Crosby, Kubina, etc.), but if the facts say they were the best TG calls, well who are we to argue with the facts?
Then Doug chimed in and said:
DM: I share Rob's view that certain prospects who are clearly part of a GM's planned line-up should be included in the baseline, but I also feel that "surprise" prospects shouldn't be.
CH: I think this enters us back into the realm of subjectivity we would prefer to vacate. Who decides that Staal (a former 1st overall pick) is a 'surprise'? To who? Do we vote on who is a surprise and who isn't?
"Sorry Doug but even though none of us had ever heard of this guy you drafted who went on to score 150pts, he's still no surprise, even if you think he is".
Or the reverse;
"Congrats Brian, none us saw coming that 100 pt season from the 1st overall selection in the NHL ED. Total shock to us all, here's your Omnivore!"
If the stats say he would dress based on his stats over someone Brian drafted, then I am comfortable with that. The fact some of us didn't see his greatness coming isn't relevant.
All in all, adopting this method would reward GMs for aquiring better players than they started with, coaching moves, etc. The downside is that using a perfect baseline will result in lower overall Omnivore scores. I can live with that.
Lastly, I haven't commented directly on Dan's statistical blending proposal, in part because it has math involved and I did a humanities degree precisely to avoid that subject.
Cam,
It sounds like the consensus we are coming to is that we pick the best possible line-up from all thirty post-draft (last pick in PrD) rosters. We include TGs letting the chips fall where they may (most possible points is still the most possible points, even if they come from an unlikely source). We do not include +/-. Essentially what we have been doing for the last few years but now including prospects.
Am I correct that this is the emerging consensus?
This will be a baseline line-up that will be fluid (as injured players come back to play, prospects are brought up from the AHL and players go on unexpected scoring streaks). I am fine with this but it means that we are accepting that improving one's team, as reflected by the Omnivore, will NOT include the drafting of OR trading for prospects (at least trades that occur prior to the end of that year's PrD). I am surprised by Cam's concession on this point as it was only a year ago that he was ferverently espousing that drafting and trading for quality prospects was one of the halmarks of a good GM.
The trade of a star player for a potential superstar prospect will be recorded by the Omnivore as a "loss" in the year of the trade and will not be recorded as a "win" when the next E.Malkin arrives to ply his trade the following season.
It does seem to be an imperfect solution.
Doug said:
"The trade of a star player for a potential superstar prospect will be recorded by the Omnivore as a "loss" in the year of the trade and will not be recorded as a "win" when the next E.Malkin arrives to ply his trade the following season.
It does seem to be an imperfect solution."
- Totally Agree. The caveat is that what the Omnivore represents is NOT the best GM, but is instead the greatest improvement. If Brian starts the year with Staal he can hardly be said to have improved his team by dressing him for the year.
The fact is we have a Best season award (Predator), Worst (Herbivore), Head to Head (Challenger) and most improved (Omnivore). If the latter is to mean what we say it means it should actually measure improvement.
That said, if we want a Top Gm award, I think we could find a way to calculate that through some formulae or other.
OK, I get what you are aiming at, but there is one quality about the Omnivore I have always liked. The players counted align with an active roster, so the total represents a total you could actually have accummulated had you been prescient enough to pick the right ones. You could look at your final total, and baseline, and see if you made more or less points than your team had potential to garner.
I am gathering that the desire is to capture and count (in some manner) all of a team's player assets after the prospect draft.
There is a way to do both, but with apologies to Cam, it does require math.
Here's how it would work:
Take my LW for example:
I will draft Nash, Frolov, and for the sake of argument, Roberts and Gelinas. I also have LW prospects in Jokinen (who serves the example as a prospect likely to be promoted), and Semin.
At any point in the season, we take the four best players of all my LW, FA/RFA/Prospect. Lets say that is Nash, Frolov, Jokinen and Semin. These four would represent my best lineup roster (active and benched) from all my post-drafts assets. These four would be added together.
Now to make it consistent with how players are counted in our pool, and to allow direct comparison to the points actually garnered, we count 3/4 of that total, which represents an active roster worth of points derived from the best full roster (since in the Predator race, only 3 of 4 LWers count for any given week)
So to summarize, we take all players a team has, prospect or draftee, and identify the best 4, total them, and multiply by 0.75.
So for LW, C and RW:
(Best 4 from entire roster summed) x 0.75
For defense:
(Best 8 from entire roster summed) x 0.75
For goalies:
(Best 2 from entire roster summed) x 0.5
Now all we have to do is decide how to count TG (should actually be not too hard - Which gives the better total?
1)counting a pure goon among the top 4 C/RW/LW and top 8 D based on PIM points, or
2)counting the top point getters, and the best PIMs from among those players
That way we are counting all players a team has (FP, FA, RFA, prospect) and extracting the best points a promoted roster could achieve, then counting the equivalent of an active roster only.
It sounds complex, but actually it is very simple...and I think it satisfies everyone's concerns.
Thoughts?
Dan said:
"OK, I get what you are aiming at, but there is one quality about the Omnivore I have always liked. The players counted align with an active roster, so the total represents a total you could actually have accummulated had you been prescient enough to pick the right ones. You could look at your final total, and baseline, and see if you made more or less points than your team had potential to garner."
CH: You can do exactly this with what Rob and I are proposing. The snapshot we take contains all the assets from immediately after the last pick of the prospect draft. As such, I can track their current totals to see what my running omnivore total looks like.
Further consider that in your averaging scheme, you get rewarded on your omnivore for having a player get zero in the 4th line role.
i.e. - assume your LW is as follows;
Nash = 80
Frolov=60
Semin =40
Jokinen=0 (assorted massive injuries)
Under Rob/my suggestion, we would count 80+60+40=180 pts of LW scoring for the Omnivore. Simple, straight forward.
In your scheme we take 80+60+40+0=180 x.75 = 135! Despite the fact you would have played the above left wing for 180!
Because you are averaging in the crappy 4th liner, you are artificially deflating the baseline, and thus inflating the potential Omnivore score.
How does that make sense over simply taking the best three at each forward spot, etc?
Fair enough - Guess I'm still thinking in terms of the 22 man post ED roster.
So let's say we take the best possible active lineup from all assets the team has after the PrD. Plus minus would be excluded, and during the season we can simply count the two highest PIM among the 15 active skaters.
At the end, we'd have to check to ensure the team wouldn't be ahead by playing a pure goon, but otherwise should work.
I'm all for trying to get a vote now, if no one objects...
Post a Comment