Frank DuPont

Draft Analysis

print article archives RSS

Value Based Drafting Revisited

Monday, August 13, 2012

The following article is a guest post by Frank DuPont, author of Game Plan: A Radical Approach to Decision Making in the NFL. You can also follow Frank on Twitter.

(For background on Value Based Drafting, see Part I and Part II of this series.)


Earlier in the summer I wrote a two part series on Value Based Drafting (VBD) that offered a new spin on the fantasy football drafting method that's been around for a long time.  VBD was pioneered by Joe Bryant, who offered an idea which is now almost universal throughout fantasy football thought.  The idea is that a player's value is determined by how many points he can score relative to the other players at his position, rather than how many points he scores in total. 


But in order to know how many points a player is worth relative to others at his position, you need a baseline player for comparison purposes and that's the thing that tends to give people fits.  My primary contribution to the discussion was to offer a way of thinking about the baseline that focuses on the supply and demand for players at a position based on how many games you need out of each position.  However, all I really did was to formalize something that everybody already thinks about when they draft their teams.  Everybody already does the mental calculation where they think about the number of players they start at each position and potential for injuries that might make it important to have good backups.  So my approach to the problem was really just a way of re-stating what everybody was already thinking.


I wanted to come back to the topic today because I've been working on it a little bit over the past two months.  I've been listening to comments on my approach, making slight tweaks to the methodology, and also fine tuning my thinking on the logic behind it.  I'm hoping that this article will clarify things a little bit as well.


The Supply and Demand View

In my earlier articles I proposed that we think about the baseline in terms of the supply and demand for players at each position.  I used a method that someone a lot smarter than me has since called the "Man Games" approach.  If we play in a 12 team league, we know that we need a starting quarterback for every team in our league for every week of the NFL season (or all but week 17 anyway).  So how many QBs does it take to actually fill out that need?  That's where I propose that we look at historical data to determine how many QBs are actually required to fill that need.


I want this to be clear, so I'm going to go through my process here step by step.


Step One.  Calculate the needed "Man Games" for each position.  If you have 12 teams and you start 1 QB, that's 12 TMs x 1 Starter x 16 weeks = 192 required Man Games.  That formula actually is a slight tweak to my prior work as I previously used 17 weeks.  Since most fantasy seasons wrap up in week 16, I am now throwing out week 17 completely.  Let's use a few more examples to be sure you understand calculating the required Man Games.  For a 10 team league that starts two running backs, it would be 10 TMs x 2 Starters x 16 Weeks = 320 Man Games required.  For a 14 team league that starts 3 wide receivers it would be 14 TMs x 3 Starters x 16 Weeks = 672 Man Games.


Step Two.  Calculate how many games are actually played by each position starting with the top ranked player at the position from the previous year.  This part is actually simpler than it sounds.  All I'm going to do is look at every RB that finished the season as the top ranked RB one year, and then see how many games in the following year they were able to score at least one fantasy point.  Then I'll do the same thing with the 2nd ranked RB and so on.  That will allow me to figure out how many games I'm going to get from each position rank and I can create a graph that looks like this:



There are some important points to make here on methodology.  First, I'm looking to the previous year to provide the position ranks because fantasy drafts the following year will generally track the prior year's fantasy finishes.  The top WR from the previous year will generally be drafted first among WRs in the next year.  So when I want to figure out how many games I should pencil in for each position rank, I use the rankings from the prior year's fantasy finish.  In my previous articles I used ADP instead of the prior year fantasy finish, so this is a change.  But it shouldn't be a huge difference and the reason I've made this change is so that I could include more seasons of data (11 seasons of data compared to 5 seasons of data in my previous articles). 


Also, since my first articles on the topic I've formalized my definition of what counts as a game.  I'm now using the very simple definition of a game as one in which the player scored at least one fantasy point.  I'm doing this because it's a simple test and is equally fair to all positions.  So in the graph above, when it says that RB1 contributes a little less than 13 games, that means 13 games where they scored at least one fantasy point.


The following graph shows the relevant game expectations for each of the fantasy skill positions.  This graph is based on 11 years of data which has been smoothed.  Also keep in mind that week 17 has been excluded entirely, so the maximum that you would ever get from a single player is 15 games.  Then when you average out 11 years of data and include players that either got hurt (Tom Brady 2008) or were eventually benched (lots of Mike Shanahan coached players), you get lower game expectation numbers that top out at about 13 games and go down from there.



Step Three.  The next step is to take the Man Games demand from Step One and then fill them in with the Man Games supply that we get from Step Two.  I've done that for each position, for a number of different league sizes and roster compositions.  The tables that follow show those baselines excluding flex plays, which we will deal with later.


Quarterback Baselines

Starting QBs10 Teams12 Teams14 Teams
1 QB14 QB16 QB19
2 QB31 QB42 QB55


Running Back Baselines

Starting RBs10 Teams12 Teams14 Teams
1 RB13 RB16 RB19
2 RB27 RB34 RB40
3 RB44 RB55 RB68

Wide Receiver Baselines

Starting WRs10 Teams12 Teams14 Teams
1 WR13 WR15 WR18
2 WR26 WR31 WR37
3 WR40 WR49 WR58


Tight End Baselines

Starting TEs10 Teams12 Teams14 Teams
1 TE14 TE18 TE21
2 TE34 TE47 TE64


These baselines are roughly consistent with the baselines I came up with in my previous articles, although there are some small differences that can be attributed to the slight tweaks in my methodology.  Remember that these baselines exclude the flex position.  I didn't include flex in the above baselines because I didn't want to confuse the issue… yet.


Calculating the new baselines for RB and WR when you include the flex position is complicated because it depends on a few things.  First, it depends on your scoring format.  If you play in a PPR league you're going to have more WRs in your flex than RBs.  But also, because there are so many league roster formats, it's almost impossible for me to address all of the potential scenarios that exist.  However, the following tables are an attempt to address a number of the possibilities that might exist.  The tables show the new WR and RB baselines for a few roster formats once flex is included.


Standard Scoring

Roster Make-up/Starters10 Teams12 Teams14 Teams
2RB/2WR/Flex RB33/WR35 RB40/WR44 RB44/WR52
1RB/3WR/Flex RB28/WR40 RB34/WR49 RB38/WR61
2RB/3WR/Flex RB40/WR43 RB48/WR56 RB58/WR66


PPR Scoring

Roster Make-Up/Starters10 Teams12 Teams14 Teams
2RB/2WR/Flex RB27/WR40 RB34/WR49 RB42/WR57
1RB/3WR/Flex RB26/WR42 RB34/WR49 RB38/WR61
2RB/3WR/Flex RB34/WR49 RB43/WR60 RB54/WR70


I didn't include TE in the flex calculation because in general they don't score enough to be included.  That doesn't mean that every once in a while you won't have the opportunity to start a TE in the flex spot.  It just means that as a group, including them in the flex calculation doesn't change the TE, RB or WR baselines.  To put it another way, the value of the TE position isn't impacted in most league formats by their ability to fill the flex spot.


Now that you have the baselines, you can take them and look at positional value using the projections from the Rotoworld Draft Guide


My goal in going through all of this again was primarily to get you baselines for roster formats and scoring rules that will be close to the ones in your league.  I had to go through some of the methodology again in order to do that, so if you're about ready to jump out of your window, I apologize.  Thanks for hanging in there and I hope at a minimum I've given you some food for thought on how you value positions.

Email :Frank DuPont

Highest Searched Players over the last 7 days

Video Center

    Player News: Week 17

    Player News: Week 17
    Matchups: Gordon, Jeffery

    Matchups: Gordon, Jeffery
    DFS Analysis: Williams/Coleman

    DFS Analysis: Williams/Coleman
    Dose: Cam

    Dose: Cam's Season Ends
    DFS Analysis: GB

    DFS Analysis: GB's Williams
    Dose: Cam Shut Down

    Dose: Cam Shut Down
    Dose: Lindsay Makes Pro Bowl

    Dose: Lindsay Makes Pro Bowl
    Dose: Gordon Eyes Wk 16 Return

    Dose: Gordon Eyes Wk 16 Return