Three year weighted averages

Theories, Concepts and Analytical Discussion (draft strategies, valuation, inflation, scarcity, etc.)
Post Reply
Message
Author
da_big_kid_94
Hall of Famer
Posts: 1574
Joined: January 3rd, 2009, 12:09 am

Three year weighted averages

#1 Post by da_big_kid_94 »

I was glancing at, of all things, a blurb about Knowshon Moreno and a thought started to come to the forefront. If you've followed him, you know he gets hurt a lot - but that's not what got me thinking. He's not exactly the poster boy for the three year weighted average because of these injuries but he spurs a bit of analysis. Many projections are based on three year weighted averages - but then many of us seem to do something strange with them. Assume these fictional histories are for players who have played a full year for each of the prior three years that will make up the average;

Option 1:

Year 1 - Bad
Year 2 - good
Year 3 - good
Three year weighted average and year 4 projection? Probably good - based on reasoning 1st year was learning.


Option 2:

Year 1 - good
Year 2 - bad
Year 3 - good
Three year weighted average and year 4 projection? Probably good - based on reasoning 2nd year was sophomore slump.

Option 3:

Year 1 - good
Year 2 - good
Year 3 - bad

Now, here's where it gets interesting. Options 2 and 3 should produce pretty close to the same three year weighted average - but many of us will turn on Option 3 because of WHEN the bad year took place and downgrade this player - where as the player in Option 2 will be more than likely given a pass while producing very similar numbers to the three year weighted average in Option 3.

So ...in evaluating the three year weighted average, in cases where 2 out of 3 seasons were of a certain type (and given the averages are based on numbers - and numbers are fact and really can't be disputed) - should there be consideration used in judging the player's 4th year as to WHEN in the three year weighted average analysis the anomaly season occurred? And if so ... what shape might it take?

I thought this because theoretically, one can't dispute the numbers ...yet, seemingly, a lot of us do just that ....and I'm not totally sure it's wrong. Nor am I saying that projections should be treated as absolute here. I'm just throwing out an idea that maybe there are other insights out there on these evaluations that I have not considered.
These are my views based on my own opinions and observations - your mileage may vary.
"KNOW THY LEAGUE" - the Forum Funklord - 4/13/2009
Fantasy is managing stats ... roto is managing teams

User avatar
Todd Zola
Hall of Famer
Posts: 8260
Joined: December 25th, 2008, 12:45 pm

Re: Three year weighted averages

#2 Post by Todd Zola »

I don't know that the three-year average applies to football projections - at least not with the same level of research that shows how well they correlate in baseball. Also keep in mind that it's more than a three-year average, it's adjusted via an aging curve (at least in baseball).

In baseball, the players are more in control of their own skills, their own usage. The sample size is large enough to flesh out most of the biases relating to quality of opponent, field, weather, etc.

This isn't the case for football (and something I have been thinking a lot about as I'm getting more into the one-week football contests). The unbalanced schedule, not playing every team, not playing those you do play home and away, the weather, injuries, then add in new coaches, new schemes, new players -- all this really skews the sample.

if I did football projections. they would be much more subjective, done by hand and not mostly generated via a spreadsheet where all the factors are coded into the engine like baseball.

My gut tells me someone that does 3-years in football does them more because they think they should as opposed to knowing that they should. I haven't seen any research that shows they're a reliable foundation in football (reliable for ALL positions would have to be a part as well).
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord

I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord

Always remember, never forget, never say always or never. - The Forum Funklord

You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord

da_big_kid_94
Hall of Famer
Posts: 1574
Joined: January 3rd, 2009, 12:09 am

Re: Three year weighted averages

#3 Post by da_big_kid_94 »

I was afraid of this ... when I cited Moreno, I felt it might cloud this issue ... my little homily was not meant to be intended towards football ....rather, it was intended to be used in a baseball frame of reference. Moreno was just an abstract trigger that made me want to look at this further but from a baseball perspective.
These are my views based on my own opinions and observations - your mileage may vary.
"KNOW THY LEAGUE" - the Forum Funklord - 4/13/2009
Fantasy is managing stats ... roto is managing teams

User avatar
Todd Zola
Hall of Famer
Posts: 8260
Joined: December 25th, 2008, 12:45 pm

Re: Three year weighted averages

#4 Post by Todd Zola »

I explain this in the Platinum sections, but my process involves a three-year weighted average but it's more than that.

Each year is prorated to the number of PA from that season. The first determination is a skills average that gets converted to a per PA number, then I project PA for the upcoming season and multiply

Depending on the circumstances for a season, I may tweak the weighted average. For example, there's a very good chance Dustin Pedroia played the majority of this season hurt so I may tone down the weight for 2014 if the reports are he'll be fully healthy in 2015.

Or if a pitcher spent a good part of the season getting the rust off, I may tone down that season (can't come up with an example off the top of my head).

This off-season, I plan on coming up with a more refined team component of runs and RBI which helps account for personnel changes.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord

I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord

Always remember, never forget, never say always or never. - The Forum Funklord

You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord

da_big_kid_94
Hall of Famer
Posts: 1574
Joined: January 3rd, 2009, 12:09 am

Re: Three year weighted averages

#5 Post by da_big_kid_94 »

See, Todd ... I understand what you are saying ...but you are looking at it from the role of the projection creator; I am looking at it from the perspective of the end user. Your numbers are your numbers and your algorithms are your own. I am looking at it from the consumer who has the numbers in front of them and still manages to question their voracity based on WHEN those anomaly seasons occurred (i.e, the "career year") during the period weighted - and why the end users would question SOME projections and not others (Option 2 vs. Option 3 in the O.P.) - even thought they both may be 100% correct.
These are my views based on my own opinions and observations - your mileage may vary.
"KNOW THY LEAGUE" - the Forum Funklord - 4/13/2009
Fantasy is managing stats ... roto is managing teams

User avatar
Todd Zola
Hall of Famer
Posts: 8260
Joined: December 25th, 2008, 12:45 pm

Re: Three year weighted averages

#6 Post by Todd Zola »

Couple comments ...

1) You need to really know how your source comes up with their projections (or have a vehicle to ask specific questions, like this forum) so you don't double-dip --- account for the same thing twice.

2) You need to choose a source where you share some commonality with respect to philosophy

3) You need to keep in perspective what projections are and what their intent is -- a guideline, a weighted average of all outcomes, not the definitive outcome. Some care more about the variance associated with a projection than the projection, which is a measurement I'm not sure how to generate but attempt to discuss, when apropos, in my profiles.

4) if you opt to use some subjective bias, be fair and apply it across the board to everyone or have what you believe is a reason it is unique to that player.

It's #4 that is most troubling to me and what I sense is the crux of this discussion.

I suppose it's why we play the game.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord

I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord

Always remember, never forget, never say always or never. - The Forum Funklord

You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord

Trav The Ump

Re: Three year weighted averages

#7 Post by Trav The Ump »

I actually like #3 as a talking point, I'm just currently doing up my own personal hockey projections as my draft is on the 5th and I gather resources and projections and than weigh them with my own interpretation/ instinct. However that variance plays a big part in my draft strategy. Is a guy a 45 point replacement player who could go to 50 points or is he a 30 point bum who could goto 70 if everything broke right?

User avatar
Todd Zola
Hall of Famer
Posts: 8260
Joined: December 25th, 2008, 12:45 pm

Re: Three year weighted averages

#8 Post by Todd Zola »

A lot of baseball variance revolves around the luck-related metrics like BABIP, HR/FB etc for hitters, BABIP, HR/FB, SwK%, LOB% for pitchers. Certain traits lend themselves to having luck play a larger or smaller role.

I don't know what the analogous consideration would be in hockey, other than time spent on the power play/shorthanded situations or perhaps playing time.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord

I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord

Always remember, never forget, never say always or never. - The Forum Funklord

You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord

Trav The Ump

Re: Three year weighted averages

#9 Post by Trav The Ump »

People much smarter than me are just starting to try and define those stats. I've always found the easiest way to find a player on the rise is shot totals. If he's shooting more, he's scoring more, he's on the ice more, etc etc etc. Now of course to determine WHO will shoot more this season...that is the question.

Taking it back to baseball, isn't the dream essentially to come to a point where babip isn't luck but expected outcomes dependent on defense positioning, batted ball speed and type of hit?

A slow roller into a shift is an out 98.7% of the time, a hard roller is 82.4%, a chopping ball is 67%. Something like that?

User avatar
Todd Zola
Hall of Famer
Posts: 8260
Joined: December 25th, 2008, 12:45 pm

Re: Three year weighted averages

#10 Post by Todd Zola »

There will always be luck associated with BABIP -- as batted ball data gets parsed into smaller, more refined samples some of the luck will be removed, but there will always be variance within each subset.
Catchers are like prostate exams -- comes a time where you can't put if off any longer, so you may as well get it over with and take it up the butt - The Forum Funklord

I'd rather be wrong for the right reasons than right for the wrong reasons - The Forum Funklord

Always remember, never forget, never say always or never. - The Forum Funklord

You know you have to seek therapy when you see one of your pitchers had a bad night and it takes you 15 minutes to find the team you have him on. - The Forum Funklord

Post Reply