How does Yahoo (or any system) use stats to create their rankings? I'm trying to come up with an equation to compare players at different position using my on projected stats and I'm having trouble refining the equation.

I have the final stats for all of the team's in my league from last year. I am comparing my projections to last years stats to understand each players impact on that category (Pitcher A will have 15 wins, and it will take 100 wins to lead the category, therefore he contributes ~15%).

Hmmm. I'm somewhat interested in this as well. I've been playing around with excel spreadsheets and have been hitting walls as to what I should do to get it just right. I took a stats class last year, but my brain has already forgotten most of the advanced stuff. One thing I tried is taking all the stats of players at each position who got at least X number of ABs (I used 400). Then I found the mean and standard deviation in each category and applied it to my projected numbers for this year.

The problem with this is that some categories are freakishly skewed, especially stolen bases. Players like Jose Reyes and Carl Crawford are literally 4-6 standard deviations from the mean, which creates an insane disparity, as most other stats dont have anywhere near these kinds of outliers.

What I have envisioned in my head that would seem to be effective, but that I have yet to figure out how to exactly calculate is to assign an equal max value (say, 1) to each of the five offensive categories and assign a decimal to each player based on projections. Then just sum them all up. So a player who is projected to hit .370 60 HR 150 RBI 150 Runs 75 SB is a perfect 5.0, with 0 SB but all the rest is a 4.0, etc.

I'm interested to hear about what other people are trying. I like the mean and standard deviation, but I see that pertaining more to a pre-draft prep and we have already had our draft.

Here's a little more on what I am trying to do. I'm taking the stats from category winners from last year in my league (its a keeper league with all the same managers). I then figured out what percentage each player will contribute to each category. I then weight each category so that I don't have any problems with individual stats skewing the results (like Reyes stealing 70+ bases).

For the pitchers, the process is a little more complicated because the max innings is for the whole staff, not individual positions. Obviously, a pitcher who wins 20 games is great, but if he throws 400 IP he's no better in wins than a 200 IP, 10 win pitcher (unrealistic, I know, but it makes my point). For wins and Ks, I have to add an additional term to the equation that weights it in favor of people who throw less innings. For ERA and WHIP, I have to weight it in favor of those who throw a lot of innings. (Batting average also has an additional term to favor players with more at-bats).

by Fantasy Sports Genie » Thu Mar 13, 2008 9:50 am

BitterDodgerFan wrote:your best bet is to ask the genie.

Alas, this is one of those areas where I'm not really sure how much I'm allowed to say, and I'd rather play it safe then make someone angry and have them tell me to stop posting here. How about if I say some of Brooklyn's ideas are a good start, and leave it at that.

Mr. Genie, is there any more info you can divulge? Or can you direct us to any articles or blogs that might be of use?

BrooklynBums wrote:One thing I tried is taking all the stats of players at each position who got at least X number of ABs (I used 400). Then I found the mean and standard deviation in each category and applied it to my projected numbers for this year. What I have envisioned in my head that would seem to be effective, but that I have yet to figure out how to exactly calculate is to assign an equal max value (say, 1) to each of the five offensive categories and assign a decimal to each player based on projections. Then just sum them all up. So a player who is projected to hit .370 60 HR 150 RBI 150 Runs 75 SB is a perfect 5.0, with 0 SB but all the rest is a 4.0, etc.

Brooklyn Bums, have you thought about including negative numbers into the mix? Basically, having stats above the mean would provide a positive impact and stats below the mean would have a negative impact on the overall sum rating. A player who is projected to hit .370 60 HR 150 RBI 150 Runs 75 SB is a perfect 5.0, but the same player with 0 SB would then be less than a 4.0.

smoothcats13 wrote:Brooklyn Bums, have you thought about including negative numbers into the mix? Basically, having stats above the mean would provide a positive impact and stats below the mean would have a negative impact on the overall sum rating. A player who is projected to hit .370 60 HR 150 RBI 150 Runs 75 SB is a perfect 5.0, but the same player with 0 SB would then be less than a 4.0.

Yeah, if you use the mean and standard deviation, that would definitely have a lot of negative numbers, especially in average and SB. But again, I think using mean for SB is pretty crazy, maybe median would be a better number to go by, but then Reyes and Figgins would be even more gaudy.

So...what's this genie thingie? Care to give any helpful hints so I can find him? You can be ambiguous, make me think a little bit

in teh basketball area they have posts that link to excel spreedsheets that calculate ranks. it might have to be tweaked bc of hitters/pitchers....but taht might be a good place to start. ill see if i can find its link