Ranking players in a multi-category league

[Pasha]

How do you rank players in a multi-category league?

In a points league, it's pretty straightforward. Whoever had the most points (or whoever is projected to get the most points) is the best.

But with categories, it's much trickier.

Full confession. For decades, I used to run a couple of leagues heavily weighted towards the offense. In one of them, G, A and P had the combined weight of 26, while +/- weighed 7, PIM 5 and SOG 3. In another league, all categories had the same weight, but most of them were offense-oriented (G, A, P, PPP, SHP, GWG), and the other two (+/- and PIM) didn't mean that much. So in both leagues, I simply ranked players by P, which was pretty easy and always gave you a good idea of who's better or worse than others.

Well, it changed two years ago. The first league got discontinued, and the second one changed. We got rid of P as a category, and we added Hits and Blks.

So now we have G, A, PPP, SHP, GWG, +/-, PIM, Hits and Blks. How do you rank that?

Suddenly, many offensive players became less valuable due to dismal PIM/Hits/Blks, and most of the best defensive players (PIM/Hits/Blks) aren't good at offense... And SHP has its own problems... namely those are quite rare... And don't even get me started on GWG, which is basically a random byproduct of G...

I've tried several ranking strategies, but I am still not sure which one is the best.

Strategy 1: In each category, you assign 100 points to the top player, and then you give everyone else the percentage of that. For example, if the best player has 80 assists, you give 50 points for 40 assists, 25 for 20 assists -- and so on.

However, if the maximum result exceeds the average one by a lot, it feels sort of skewed. For example, in PPP the top result is normally about 40, while the average one is about 10. Giving an average player 25 points just seems too low...

Strategy 2. In each category, you assign 50 points to the average player, and then you give everyone else a score proportional to that. If an average player has 25 assists, then a player with 50 assists will get 100 points, while a player with 75 assists will get 150.

But doesn't that give too much to the leaders? In the aforementioned PPP example, the top player would get 200 points. And when it comes to SHP, it's even worse. The average result is less than 1, so a player with 8-9 SHP might get 400-450 points, which will totally inflate his score...

Strategy 3. In each category, you assign 100 points to the top player and 50 points to the average player. Everything under average still counts as a percentage of average, and everything between top and average is counted in a slightly different way. For example, if the top player has 40 PPP (100 points), and an average one 10 PPP (50 points), then 5 PPP (half of 10) will get you 25 points (half of 50), but 25 PPP (midway between 10 and 40) will get you 75 points (midway between 50 and 100).

And that, of course, doesn't look perfect either. Indeed, 25 PPP is five times more than 5 PPP, so why does it give you only three times more points (75 vs 25)?

So what strategies do others use? Not just managers, but fantasy sites like Yahoo, Fantrax and others? They all rank players according to categories as well, but I never fully trusted them... mostly because of categories like +/-, SHP and GWG.

+/- is basically a team category, and most +/- predictions simply rely on team success, not individual performance.

SHP is something that happens rarely, and I can't possibly fathom giving a lot of points to top SHP-getters. In fact, I combed through all the weekly games in my league from last year and noticed tht 81 out of 154 of them had 0-0 in SHP. Should I, then, weight SHP points as .474 (73 out of 154)?

GWG is... something else. It's so random that basically, I almost feel like simply weighing G points as 2.00... but then again, different teams win different amounts of games... I suppose players who play for better teams should get better GWG/G rates. Oh, and 21 out of 154 games in my league ended with 0-0 in GWG... weigh this category as .864 (133 out of 154) then?

And that's just skaters...

[LuigiC173]

perhaps you are overthinking it and should not rattle the brain with prerankings that can just go to sh*t mid draft when players you want are scooped up before you can get them. With scoring like that, I would be heralding the likes of Landeskog as the prototype and trying to draft as many players like him as possible (Hertl, Gallagher, Schenn, etc.). Do not bother with plus minus it means nothing and you have no control over it, I routinely win leagues with a year end plus minus for the roster around the likes of minus 110-120. GWG is a random game changer but those types make fantasy fun, I have them as well as shutouts for goalies in the league I run, again pure luck of the draw. In the end you just grab as many bangers with scoring touch as you can. If for instance you had faceoffs like mine does on top of these cats then centers would dominate your roster if there is no position requirements.

[mounD]

If you set up an account at Fantasy Hockey Geek, you can set up your league categories and get rankings based on your league.

[Pasha]

If you set up an account at Fantasy Hockey Geek, you can set up your league categories and get rankings based on your league.

Yes. I do have an account there, and I do get the rankings... but how do they manage to create said rankings? I am afraid they put equal weight on all positions, even rare SHP and GWG.

[mounD]

Yes. I do have an account there, and I do get the rankings... but how do they manage to create said rankings? I am afraid they put equal weight on all positions, even rare SHP and GWG.

I think you can change the weighting if you want to diminish the importance of certain stats for ranking purposes. Or you could just leave stuff like GWG out entirely. I find that there are certain guys like Bergeron and Marchand who can pick up SHP as a repeatable skill.

[Pasha]

I think you can change the weighting if you want to diminish the importance of certain stats for ranking purposes. Or you could just leave stuff like GWG out entirely. I find that there are certain guys like Bergeron and Marchand who can pick up SHP as a repeatable skill.

Thank you, I'll try that. And yes, some players pick SHP more often than others... it's just that most of the time, they don't do it often enough to affect a fantasy weekly game. And yes, some people have higher GWG/G rate than others (Keith), while others have lower one (DeBrincat)...

[teepee1]

Use Fantasy Hockey Geek as a starting point. I urge you to look at both past 1-2 years stats as well as Dobbers projections. Dobber Guide is a great source for points, and also for other cats, but sometimes I find changes I disagree with when it comes to blocks, takeaways etc. So you need to apply your own logic to those stats.

FHG should help you find the underrated gems. By that I mean a guy who covers all categories, or someone who is underrated in their ability to block and hit. Take Boone Jenner as an example. You should get him late, but you'll be thankful you did. Most other managers wont be thinking of him.

Lastly, if other GMs over value goals and assists, let them, you can win the hits, blocks, PIMs categories and they are all worth the same in the end.

[stevegamer]

Also, if the league is on Fantrax, they have a "score" rating which is a metric, that indicates how well the player rated in your league. I found it useful last year, which was my first in a category multi-cat league.

I agree with ignoring +/-, as it's the most garbage stat out there.

[Pasha]

Thank you for all the responses. I have pretty much reverse-engineered FHG, and it does make sense to me. What I do NOT want, however, is to get a team that is uber-strong (head and shoulder above everyone else) in 1-2 categories yet middle of the pack in all the others. I am writing software that will hopefully enable me to draft a well-balanced team.

Also, in my case scheduling is mega-important. We have daily lineup changes (G, D, D, LW, C, RW) every day, so it’s important to cover as many player-games as possible. The software shall take the schedule into account as well.

[CMSpiced]

I have been in a multi-cat league for a while now.

Your system seems overly complicated, for me atleast. I do an excel dump, and do a formula for each category ranking each category in the whole group. For example, in Goals Ovechkin would get assigned 1 & Draisaitl would get 2, etc. Do that for each category and add up all the points and lowest is your relative best. I then sort highest to lowest based on points (hardest stats to trade for) and then manually adjust. Doing it this will will help identify strong players in certain cats.

PS. get ride of SHP and GWG, garbage categories.