Don't Be a Tool

Interesting stuff Lyle! I am always a fan of a bit of mathematics

This is a great applicaiton of the data. I've been fooling around with standard deviation in FF for years, but Lyle, I never had the insight to subtract it from the average. I always just kept two columns, and sorted by average, and eyeballed the STD. This goes hand-in-hand with what you and Kevin preach about the value of consistency, and why erratic scorers like Chad Johnsons and Michael Vicks are fantasy poison.

The only thing that I would add to your article would be this:

One Standard Deviation, by definition, encompasses 68% of occurrances. (34% above the average, and 34% below the average). Beyond one STD is 32% of occurrences: 16% above 1 STD and 16% below 1 STD. So you could "geek it up" a little bit more, but not scare people away, by saying that a player beat their MEV 84% of the time. And that's 11 games out of a 13 game season.

Obviously, there are other variables affecting his score, and a "lifetime" analysis would be insightful too.

THANK YOU for the formula.

Nice article, and nice Spreadsheet.

Somebody got bored with their TI-86 calculator...


jk, good article Lyle.

My head hurts.

What the hell does that, “mean”?

I continued reading anyway.

Thanks for the sexy sheet! It will be added to The File.

Hey Lyle - thanks for the article. Very well thought out. One stat that a friend of mine does is called "useable starts" were he takes a minimum number of points he would like for each position. For example 10 points per game from your WR.  Out of the 15 starts that you have for each player (don't count week 17) - how many times did that player score the minimum points you would like for that position. Last year Chad Johnson had 10 points in our ppr league 11 of 15 weeks while Brandon Marshall had the minimum of 10 points in 15 of 15 weeks.  So even though Chad scored 272 points to Brandon's 267 - Brandon was relatively more valuable due to consistency.  You may want to develop a spreadsheet with points per game multiplied by consistency factors to find the best players. I did that before before last year's draft and I ended up targeting T.J. Housh and Plax in my drafts.

Here is a link to Fantasy Football Today's Crank scores for RBs.  This is a nice way to see how consistent a player is without doing the math. The Huddle has something similar; but, it will cost you to see it.

http://fftoday.com/tools/crank.php?Seas … LeagueID=1

If you are interested in the points per game. Here is a link that will get you the info from PFW.

http://www.profootballweekly.com/PFW/Fa … alfpgp.htm

Oh yeah, consistency isn't the only thing to consider; cause, if it was - Robert Meachum from NO would be a first round draft pick. He scored exactly 0 points in every game last year.  Anyway, enough of that. Again, thanks for the article. I am going to reread it again and try to understand it. But, I think I am glad I have David Garrard and Tom Brady as my two QBs on my Dynasty team.

g00seFraba wrote:

I never understood statistics when trying to interpret data myself.  it wasn't until fantasy football (more specifically Lyle's formulas and explanations of stand. dev.) where I was actually able to comprehend it.  who knew a fantasy football guru would explain this much more clearly than my college professors.

many thanks again Lyle.

I actually took a statistics course in college 3 years ago. Prior to that, Lyles column would have looked like Chinese arithmetic. I actually found the course material exceptionally easy and got a perfect A+. I never understood how anyone could struggle with it. All your doing is plugging the numbers into either the formula or the calculator and translating the results from symbols and numbers to a statement or conclusion.

pjconley7 wrote:

g00seFraba wrote:

I never understood statistics when trying to interpret data myself.  it wasn't until fantasy football (more specifically Lyle's formulas and explanations of stand. dev.) where I was actually able to comprehend it.  who knew a fantasy football guru would explain this much more clearly than my college professors.

many thanks again Lyle.

I actually took a statistics course in college 3 years ago. Prior to that, Lyles column would have looked like Chinese arithmetic. I actually found the course material exceptionally easy and got a perfect A+. I never understood how anyone could struggle with it. All your doing is plugging the numbers into either the formula or the calculator and translating the results from symbols and numbers to a statement or conclusion.

thats what everyone has told me, but my professor spoke horrible English and never really taught statistics but rather he lectured us about his struggles of fulfilling his dream in America (which was very inspirational but not very educational for the purposes of statistics).  he never gave us a syllabus, jumped from chapter to chapter and testing us on things we havent learned.  granted, I pulled out a B no thanks to my professor, but still scratching my head at the end of the semester.

meathorse wrote:

Yeah, dumbasses.

my bad

WOW... First of all, I gotta say I never thought the feedback would be this positive. I have a hard time conveying this information without boring people to death or confusing the hell out of them. I guess you guys got it. I was convinced this thread would be filled with a bunch of question marks.

Thanks guys!

7grain wrote:

The only thing that I would add to your article would be this:

One Standard Deviation, by definition, encompasses 68% of occurrances. (34% above the average, and 34% below the average). Beyond one STD is 32% of occurrences: 16% above 1 STD and 16% below 1 STD. So you could "geek it up" a little bit more, but not scare people away, by saying that a player beat their MEV 84% of the time. And that's 11 games out of a 13 game season.

I was totally gonna add this, but refrained because I was convinced it was too confusing as it is. Maybe I'll go back in and add it, or write a follow up closer to draft time.

kilgoretrout wrote:

This is all really good stuff, but now how do you apply it? 

I think for drafting sake, StrategyMan has a perfect example of how to use it.  I've done this in the past:

Averaged out all our league's scores, so I know the average score I need beat each week.  Then divide that by how many player slots I can start.  That gives me my position average.  I then need to figure data to draft a team of players who will each score at or above their position average on a CONSISTENT basis.  The SD and MEV seem to offer such a way to measure consistency and point average.

Another thing I think would be very beneficial is to get a three year trend (or whatever range, but a way to see if a guy is going up/down over time).  The trends  could also be useful for taking a set of offensive guys from the same team and seeing if the team's offensive production, as a whole, has been depreciating, stable, or increasing.  Like for Arizona, they seem like a team who is on the rise, but are they?  Getting all this data would be able to answer this question better than just a "feeling". 

And if we all worked together, each doing different data tabulations and outcomes, we could develop lists with a much finer focus.  One's which could answer in-season questions, like who to start.  For example, we could get a player's SD and MEV for Home and Away.  Or how about grass vs. turf, good weather vs. bad weather, Top 10 D vs. Bottom 10 D, etc.  If each of us was given a task, we could pool them up and get a much broader context to make start/sit decisions.

A quest!  God be praised!


I think it is a good idea, but there are still going to be times when we are going to do sit/starts on other data as well.

kilgoretrout wrote:

Exactly, Bigdaddy.  But the more info the better......err.......more excruciating to make a decision as well......

And, upon crunching all this data, we may find some statistical anomalies not yet discovered.  L

Or maybe we'll find a nice shrubbery....Ni!

Behold! Complete weekly data for 2007 (no kicking stats though). The first 2 weeks are formatted nicely but the last 15 are split up into passing, rushing and receiving.

DOWNLOAD HERE

P.S. When I have time to crunch numbers I'll post that for you guys as well.

UPDATED: Check this thread.