The Standard Deviation as an Important FF Statistic?

[The_Dude]

Hey Cafe'ers,

Duder here. So some of you may or may not know that I am a political scientist - and most may or may not know what it is that I do, outside of teaching. Most of my research uses applied statistical methodology to look at questions in survey research. Basically, this preface is meant to say that I'm a stats guy, although for some odd reason, I've never really mixed my work with my hobby and my fantasy football strategies, etc., have rarely employed much science (if we can call it that).

However, I've been thinking a lot about a stats 101 concept and if/how/when it has some applicability to fantasy football statistics analysis - the standard deviation.

What is a standard deviation? The standard deviation is essentially the average absolute value deviation from the average. Mathematically, you would calculate the average of a series of numbers, take each number and substract the average from it, square that term, and add all of those up. Divide them by the number of observations in your sequence and take the square root (or with a row or column of data in excel, you can type stdev(a1:a20) or stdev(a1:a1).

Is there any value in it for fantasy football? Well, in a sense, standard deviation is a measure of consistency - how much a player fluctuates from week to week.

Just as a matter to see if what others think - if there is any value in this (because its a pain in the arse to calculate), I've calculated the mean and standard deviations for the top 10 RBs in my league last year (based on yardage of the top 10 fantasy point scoring backs).

It looks like this...

Player Avg yds stdev
S. Alexander 122.38 40.55
L. Johnson 130.81 66.80 (39.93 after starter)
L. Tomlinson 114.50 57.19
T. Barber 149.38 60.97
E. James 122.87 35.61
C. Portis 108.25 30.87
L. Jordan 113.43 34.21
R. Johnson 102.07 29.92
T. Jones 98.53 33.92
M. Anderson 81.73 42.96


So, for those (like me) who have been saying for years that Rudi Johnson is the most consistent fantasy back out there, you now have some data to back up your claim. As far as total yardage go, Rudi has the lowest standard deviation of any other top 10 back.


Thoughts? Comments? Concerns? Queries? Suggestions? Would you like to see more?


Dude

[Mercer Boy]

Well to me, winning in fantasy football is all about consistency from week to week. If you win 120 to 40, it counts just the same as you winning 87 to 86. I always try to get at least 90 points each week because most times that will be enough to beat teams unless they have a week where they get like 10 TD's randomly.

Unless you get bonuses for having the top score every week, it doesn't matter as long as you outscore the opponent. In order to do that on a week to week basis, I try to find the players I know will get consistent yardage (providing it's a 10 yards per point league) and hopefully a TD (you can never count on those...as KG once said, they are luck for the most part unless you're one of the top RB's).

Therefore Dude, I am right there with you. Rudi is a guy I would like to have on my teams this year as it is just because he's always putting up good yardage and a decent amount of TD's. Showing that he's steady from week to week just makes me want to draft him even more.

Could you run the numbers for all the other positions too?

[onnestabe]

I definitely think that stsnadard deviation is an important measure with regard to fantasy football. I think it is a big mistake to look solely at yearly totals when comparing players.

That being said, I would also like to point out the fact that a player's performance in 2006 is independent of what they have done up until that point. What this implies (I am sure you know this Duder, just pointing it out for the sake of completeness) is that past performance alone cannot be used to judge what is expected in the future. So, while we can compute standard deviations till we are blue in the face, it is much more productive to use those stats and measures to figure out why they had success (or did not have success) in 2005, and then apply what you have learned to the upcoming season.

Step 1. Perform statistical analysis on previous seasons (find the effect in 2005)
Step 2. Analyze data to find out why the 2005 stats came out the way they did (find the cause in 2005)
Step 3. Use what you learned from your analysis to make new predictions for 2006. (assume that if you can find those same causes before they happen in 2006, and then base your rankings on that)

Too often I see people assume that statistical trends will continue, without considering what caused the trends in the first place.

Anyway, I've got to go, but I look forward to seeing how this thread develops.

[The_Dude]

Well to me, winning in fantasy football is all about consistency from week to week. If you win 120 to 40, it counts just the same as you winning 87 to 86. I always try to get at least 90 points each week because most times that will be enough to beat teams unless they have a week where they get like 10 TD's randomly.

Unless you get bonuses for having the top score every week, it doesn't matter as long as you outscore the opponent. In order to do that on a week to week basis, I try to find the players I know will get consistent yardage (providing it's a 10 yards per point league) and hopefully a TD (you can never count on those...as KG once said, they are luck for the most part unless you're one of the top RB's).

Therefore Dude, I am right there with you. Rudi is a guy I would like to have on my teams this year as it is just because he's always putting up good yardage and a decent amount of TD's. Showing that he's steady from week to week just makes me want to draft him even more.

Could you run the numbers for all the other positions too? :-D

Yup, I'd like to get to it within the week and post the top 25 for each position - just wanted to post a little teaser and see how much interest there was in consuming this information.

This, by the way, would be made considerably easier if someone had weekly data by player from 2005, by position, entered into a spreadsheet that they would share with me. (PM me if you are such a person).

[The_Dude]

I definitely think that stsnadard deviation is an important measure with regard to fantasy football. I think it is a big mistake to look solely at yearly totals when comparing players.

That being said, I would also like to point out the fact that a player's performance in 2006 is independent of what they have done up until that point. What this implies (I am sure you know this Duder, just pointing it out for the sake of completeness) is that past performance alone cannot be used to judge what is expected in the future. So, while we can compute standard deviations till we are blue in the face, it is much more productive to use those stats and measures to figure out why they had success (or did not have success) in 2005, and then apply what you have learned to the upcoming season.

Step 1. Perform statistical analysis on previous seasons (find the effect in 2005)
Step 2. Analyze data to find out why the 2005 stats came out the way they did (find the cause in 2005)
Step 3. Use what you learned from your analysis to make new predictions for 2006. (assume that if you can find those same causes before they happen in 2006, and then base your rankings on that)
Too often I see people assume that statistical trends will continue, without considering what caused the trends in the first place.

Anyway, I've got to go, but I look forward to seeing how this thread develops.

This would absolutely be a more rigorous way of doing this, although this is a simple statistic that we can and probably should be looking at to inform our decisions on players.

I wouldn't be opposed to building a big omnibous regression model and seeing how it works out. The major obstacle there is still building the dataset.

[Infidel]

Hate to rain on your parade, but I think it has almost no value. It would cause you you to not draft players like L Jordan or S Smith last year at all.

Now, if you look at it as just another arrow in the quiver (and only in the very early rounds), then maybe it's something that would be another weapon.

But, FF does not work that way. Too many changes from year to year. New coaches, new players, new systems, injuries, returns form injuries........it's way more about those type of issues.

As a former math, econ major, I applaud your effort, but if the premise is False, the conclusion is immaterial, the statement is False. And in this case, the premise is false. JMHO

[biju]

Hate to rain on your parade, but I think it has almost no value. It would cause you you to not draft players like L Jordan or S Smith last year at all.

I see what you're getting at here and I really like the fact that you've stood up to make that point.

I still see the value simply in the fact that knowing you have a "risky" player (meaning someone who has shown a high amount of fluctuation) and can somewhat perpetuate this notion by picking another similar player or help mitigate this by taking someone who is more consistent.

For example, taking Rudi Johnson in the first round allows you to take a WR in the second round more easily because even though the WR numbers tend to fluctuate you'll get steady production from Rudi. Alternatively taking LaMont Jordan in the first round might make you opt to take another RB to help counter those fluctuations.

I most certainly wouldn't bank on this by any stretch as being *the* reason to pick anyone but it could possibly affect players I take in the latter rounds a little.

As for the dataset these are usually the easiest to accumulate over the years instead of trying to datamine them. I think that would be really tough to do and take a long ass time.



Duder, I would love to see the other positions and if you feel like farming out the work a little to help I could probably be of assistance. The only problem is all of my leagues have special scoring systems so you'd just have to give me the list as opposed to saying "the top 10 in yahoo scoring".

-M

[Mercer Boy]

I'd still rather have a RB that gets me 10-15 points each week than have one that gets 30 one week and 5 the next. Same thing goes for QB's and WR's with different numbers thrown in (especially WR because they are as inconsistent as hell).

You still may not win if your guy gets 30 if the other guy has a randomly awesome week...whereas you should win all the "close ones" if all of your players score well every week...I have at least 3-4 (if not more) games a year decided by 5 points or less.

Those wins/losses add up; I know...they killed me in two leagues last year. I counted at least 7 times where I was either the second highest scoring player going against the top scorer for the week or where I had the highest score of all the losing teams. Scheduling is luck of the draw, I know.

(Note that I'm not saying I wouldn't take a superstar RB with a top pick...they bust from time to time, but they help your team win more often than not. But, when you get into the "grey area" of players with similar ability, I want the consistent guy. )

[Crimedogg32]

looks like you did just yards, wouldnt it be better to find the variation in fantasy points since yards is only half the fantasy football game. I know on football guys you can get the total points each player scored for every game last year so shouldnt make it to difficult.

[DelhommesFan1]

Hate to rain on your parade, but I think it has almost no value. It would cause you you to not draft players like L Jordan or S Smith last year at all.

Now, if you look at it as just another arrow in the quiver (and only in the very early rounds), then maybe it's something that would be another weapon.

But, FF does not work that way. Too many changes from year to year. New coaches, new players, new systems, injuries, returns form injuries........it's way more about those type of issues.

As a former math, econ major, I applaud your effort, but if the premise is False, the conclusion is immaterial, the statement is False. And in this case, the premise is false. JMHO

Chill out man. Do you ever grab a fantasy magazine in may/june and follow it exactly? No, its a guideline. There is no exactly list to follow on who to draft. Ofcourse there is going to be changes from year to year....anybody knows that. I think what dude is developing is something that can "help" you draft. Should you follow it like concrete...not necessarily.

By the time my fantasy drafts roll around...i have typically read all kinds of things like dude is making ranking players. Its helpful to have things like that

Thanks dude