Note: This article is using PPR data

While writing my Eliminate the Suck article, I started wondering aloud if I might be taking a wrong approach to drafting. What largely follows from those thoughts is a giant math exercise. My assumptions have not been fully back-tested nor do I know exactly how well to weight the three factors I will discuss below. But even despite those concerns, I feel compelled to state the things I am stating and show my work, because I think hidden in this exercise will likely yield some deeper meaning of some of the players we should build our teams around.

My assumptions (listed in my perceived order of importance):

1. The first five weeks are the most important of your fantasy football life. When your players do good in these weeks, you have the most options going forward with your team. These players can be traded high while once highly sought after players pre-draft might be acquired low. It also goes without saying that acquiring early wins gives you a lot better chance at (a) making the playoffs and (b) improving your roster via waivers later (as other owners lose hope of making the playoffs).


2. Winning your fantasy league involves winning the playoffs. It's not enough to just get there. You need to do well in weeks 14 through 16. In my early fantasy career, I would draft a large number of players from teams that I felt had the easiest fantasy playoff schedules. Those teams won their leagues a surprisingly high percentage of the time. And that makes sense. If you build your roster with the end in mind and manage to get that roster to those weeks in which it was created, you should be fielding a team very capable of winning it all. Conversely if you draft players that line up against the Jets, Ravens, etc in the final weeks, you may find your team struggling to land that trophy regardless of how well the squad played all year long. Since the payouts in most leagues are skewed heavily towards the top prize, building with the end in mind generally increases a roster's expected value.


3. Great offenses produce significantly more great values than average and below average teams. In fact, players from bad teams almost always disappoint. Who on the Oakland Raiders last year outplayed their ADP? Michael Bush likely did, but he was still a hard person to know when to play. Zach Miller might have outplayed his ADP, but with so few scores could you start him with any confidence? How about Detroit? Calvin Johnson started well and eventually disappointed. The RBs were all bad. The TEs and other WRs were far too inconsistent to ever matter. QB Matt Stafford threw way too many interceptions to be anything but a spot starter as well. You get the points. Bad offenses disappoint. They generally involve rotating QBs, juggling the offensive line, etc. All of these constant changes usually brings down everyone on the roster.

Starting with these three assumptions I set out to build a better set of rankings (Actually I set out to create multipliers that I would adjust my projections accordingly).

Tackling the first 5 weeks:

Clayton Gray is the expert on Strength of Schedule and he gets into deep number crunching on which positions will do good/bad against the various defenses. From experience though, I know his early positional SOS numbers have a high variance (this improves as we see more games played making his SOS extremely powerful by season's end). This early variance is likely caused by our overall failure to adequately predict how good defenses will play in the coming year. Are they better against the run or pass? History plays a part as does a team's scheme, but this does feel like a situation where it's not as easy to tell the above average over the below average defenses against RBs at the start of the year.

In an effort to simlify things, I looked to the FBG Consensus Expert Rankings. I divided the consensus list into 8 tiers. The top tier (toughest defenses) would count as -5 for the team playing them that week. Conversely, the bottom tier would count as +5 for the opponent. The eight tiers would yield (-5, -3, -1, 0, 0, +1, +3, +5) for every game played. Byes would be counted as zeros.

This approach yielded the follow (listed alphabetically by tier):

Tier 1 "Best" defenses (-5 points): Baltimore, Minnesota, New York Jets, Philadelphia

Tier 2 defenses (-3 points): Dallas, Green Bay, Pittsburgh, San Francisco

Tier 3 defenses (-1 point): Chicago, Cincinnati, New England, New York

Tier 4 and 5 defenses (0 points): Carolina, Denver, Houston, Indianapolis, Miami, New Orleans, San Diego, Tennessee

Tier 6 defenses (+1 point): Arizona, Atlanta, Buffalo, Washington

Tier 7 defenses (+3 points): Jacksonville, Oakland, Seattle, Tampa Bay

Tier 8 "Worst" defenses (+5 points): Cleveland, Detroit, Kansas City, St. Louis

Additionally to account for teams generally scoring more points (and having more success in home games), I gave +2 points for a home game and -2 for an away game. If a team had 3 home and 2 away games in the first five they would receive an additional 2 points.

Tackling the playoff weeks:

I took the exact same approach as the first 3 weeks with the same tiering scale.

Offensive Adjustment:

The approaches for the starting and playoff weeks yielded ranges of +15 to -11 and +11 to -15 respectively. I wanted to grade the offenses in a similar scale and opted to define them as +15 to -15 in groups of four. I totalled all PPR fantasy points the offenses would score using my projections. This approach yielded the follow (listed alphabetically by tier):

Tier 1 "Best" offenses (+15 points): Dallas, Green Bay, Indianapolis, New Orleans

Tier 2 offenses (+10 points): Houston, Minnesota, New England, San Diego

Tier 3 offenses (+5 points): Atlanta, New York Giants, Philadelphia, Pittsburgh

Tier 4 and 5 offenses (0 points): Arizona, Baltimore, Chicago, Denver, Detroit, Jacksonville, Miami, washington

Tier 6 offenses (-5 points): Cincinnati, New York Jets, Seattle, San Francisco

Tier 7 offenses (-10 points): Carolina, Kansas City, Tampa Bay, Tennessee

Tier 8 "Worst" offenses (-15 points): Buffalo, Cleveland, Oakland, St. Louis

Combining Everything:

I am not sure exactly how to weight these three factors correctly so I guessed at it. The rating of the offenses is already deeply imbedded into my projections so I feel it is of least importance. The playoff weeks only will matter if your team is good enough to get there. The first 5 weeks always matter. Based on these things, I opted to calculate my Combined Score as:

Combined Score = (3 x Early Schedule) + (2 x Late Schedule) + (Offensive Adjustment)

These scores are then sorted to apply the multiplier values.

I don't really have a good feel on how much to adjust values up or down here. I opted for a +15% to -15% range using the same 4 team separators between levels. This might be understating things, but I also did not want to just throw out my core work on the projections.

The table below summarizes all of the math discused above:

Applying the multipliers to my current projections yields these differences:

Quarterbacks

Running Backs

Tight Ends

Wide Receivers