Valuing categories in roto leagues

Naslund's great thread on draft lists got me thinking about this. It's projection season, and I'm trying to pull things together for my roto league. We all know that goals are harder to come by than assists, but by how much? What about PIM, how much should they weigh? Etc etc.

Ideally we want to assign a relative weight to each stat category, right? Based on Dobber's projections, the top 45 centres (the ones that were on peoples' rosters) will put up an average of 74 points, with 27 goals. So goals are 1.7 times more difficult to come by than assists (47 assists / 27 goals). But what about other categories? What makes sense in terms of ranking their relative value?

(I'm also looking at doing this using the "above the bar" technique for each position, but I'm still looking to better understand how to weigh the values of each stat)

Edit: Come on, I know there are stats nerds out there...

Nobody has thoughts on the comparative value of different categories in roto leagues? I would have thought that this is a pretty key issue to valuing players...

To dangle a carrot, I'm happy to share my the file with those who give thoughtful answers!

i'm no stats major, but here's what i did:

in each category, divided each player's total (either last season or projected, whatever) by the highest value in that category. you end up with a percentage of how valuable that player was in that category compared to everyone else.

then added each percentage together.

very rudimentary, and i'm eager to learn more. for instance pooladdict, i've been giving some thought to how i too can accurately weight categories for importance.

at the very least, since for roto leagues you get points for how well you stack up relative to every other team, my crappy formula at least gives you a relative measure of a player's worth.

Is this input for creating a new league and therefore trying to set up the different categories with an appropriate weight? Or is this an effort to rank players?

If it's an effort to rank players then I question the value of knowing assists are 6/9 as valuable as goals etc..

I think my problem with this type of ranking is that it attempts to value a stat in a vacuum. For example, let's say you're trying to get B. Boyes' relative value. With his high goals totals, he may look like a stud. Factor that, though, with his relative low shot total and he retracts in overall value. In a roto league you will always be trying to build the most well rounded team. In that light, all stats count as exactly the same weight [You get 10 points for winning the goals category. You get 10 point for winning the PIMs].

I think of more value is to try to find the correlation (actuarial sciences anyone?) between the various categories... for example, what is the average correlation between shots and goals? goals and GWG? assists and PPP?

If you can decipher these you can get a better idea on how a player will contribute across the categories - and therefore be a more valuable player.

Example: If Boyes scores 30 goals on 200 shots, and constitutes 20% to your team's overall G's and 15% of team's overall shots, he is less valuable than B. Rolston who scores 25 goals on 300 shots, contributing 18% of team's overall goals and 25% team's overall shots. Math is off here, but point is succinct.

Once again, FantaSteve finds X.

Boyes is the perfect example. Our roto legaue is simple Goals and Assists; your total ranked against everyone else's and scored accordingly. Boyes helped me kill people in RWG last year, but i got dusted at RWA. Would you dress Dan Carcillo to win PIM, if it's going to cost you at G,A,+/-, and PPP. Probably not. It's all about finding players who mix well together and give you a balanced attack.

Good comments, guys. Steve, you're right that stats earned count for exactly the same weight, but what I'm trying to do is somewhere in between what you two are suggesting: I'm trying to determine how easy a certain stat category is to fill relative to what is available, and use that "rareness" number against player stats to assess just how valuable they are.

To take it back to your example, FantaSteve: comparing Brad Boyes scoring 30 goals and 60 points (25 PP) on 200 shots and Brian Rolston scoring 25 goals and 55 points (25 PP) on 300 shots. (Just throwing numbers out there) roto-friendly. Intuitively, points are easy to get, whereas shots, goals, and powerplay points are harder. But which is harder to find, shots or goals?

What I'm looking to do is quantify this. Make sense?

Figuring out what's left is easy enough, but where I'm having trouble is sorting out how valuable/rare certain stats are when compared to the other ones. One could use this to assess how well-rounded a player is.

But I appreciate you questioning whether this is useful or not - is this just a silly thought exercise or could it be a useful tool? Is this something that common sense at the draft table would simply make redundant? I guess in the end what I'm looking to do is develop a tool that assesses a player's value in a roto league.

fantasteve..

i *think* that is essentially what my method has done, alhtough in pretty brute fashion.

i'll try to explain via example what i'm not confident about:

lets compare iginla (50-48-98) vs thornton (29-67-96). iggy scored ~ 20 more goals, but jumbo joe had ~ 20 more assists.

the question becomes, if i have one of those guys, how easy is it for me to make up the lack of goal/assist production?

the team with iggy needs another player to make up 20 assists, whereas the thornton team needs to make up 20 goals. let's look at guys who were in the 55 point range:

saku koivu (16-40-56) vs corey perry (29-25-54)

if the iggy team also has koivu, then he gains 15 assists vs the thornton/perry team gaining 13 goals.

hmm... not a very good example!

let me think more about this at lunch.

I see where you are coming from.... a 70 point center that scores 35 goals and 40 assists is more valuable than a 20 goal scoring one(at least in my league he sure is).

What i do to prepare for is this. Let's say that 100 roto points will win my pool, or at least come close to it. I take projected scoring of my protected players, and see what i will still need to complement them to get that 100 points. Now, if your league has multi-scoring stats, like +/-, PIM, SOG and the like it becomes harder to evaluate. But the bottom line is, all you are looking for is what you need to be competitive with what you already have...it doesn't need to become rocket science

here's another thing...does the league break down into individual forward position like LWG, LWA, CG, etc....or is just straight g,a,+/- regardless as to forward or defence. in the former, two players that are goal heavy/assist weak at that position will hurt you. In the latter you can afford to have a player who dominates one category and not much else, as hopefully the rest of your team will balance other categories based on their strengths.

In this league, we have player positions but stats are tracked across the board so it doesn't matter.

It's one thing to assess what your team needs to put it over the top in a given year, but I'm looking to really understand player value in a roto league. This is obviously not an easy subject, looking at the number of big words and awkward sentences that are in this thread so far...

I guess in the end this could be a measure of a player's "roto-studliness", which is a very useful thing especially when evaluating trades.