Thursday, October 4, 2012

Week 4 Results Episode

Introduction

I've had this one in the works since week one and I can't begin to express how excited I am about this. My friend and I have finally designed a quick and easy way of assessing player rankings from different sources against eachother. With this first post I'm going to outline what the percentages mean and how you get them. What I want from my community of readers is feedback. Take into consideration that changes to the formula are not easy or taken lightly, but that the goal of this is to create an accurate representation of individual player rankings from a fantasy standpoint; I am looking for feedback about why this does or does not achieve this goal. Don't hold back and don't be nice; I want feedback. In addition to feedback, I would love to have some of the smaller websites or individuals submit their rankings. Fantasy football is about compilations.

Methodology

These ratings work on effectively three different levels for each position. These can be simply broken down into quarterbacks and tight ends, and running backs and wide receivers.

Ratings will be based first on the top players in each expert's rankings. For QBs and TEs, this is the top-12 ranked players from those positions; for RBs and WRs, this is the top-24 ranked players from those positions. This difference is meant to represent the typical fantasy starting lineups of 1QB, 2RBs, 2WRs, and 1TE.  These top-tier rankings are rated on two different qualities; head-to-head percentage and inclusion. The same two ratings are then applied to the expert's full list of players.

In the head-to-head rating, there is a simulated matchup between every two players on an expert's list. For example, if my #1 RB was Arian Foster, the script would simulate a matchup between every pair of players with the expected outcome of Arian Foster scoring more points. If Arian Foster did score more points, the script credits me with a win, and if Foster scored fewer points I am not credited with a win. The final ranking is the number of wins divided by the total number of matchups recorded. This gives an effective correctness percentage; If you followed the advice of this expert exactly, what percentage of decisions would be correct? This tells a very important part of the story, but absolutely does not tell the whole story. Suppose for an instant that I predicted Arian Foster 1st and then a whole bunch of injured and backup running backs. I would get a perfect head-to-head rating, but it wouldn't do you, the fantasy readers, a damn bit of good. To counteract this problem, there is the inclusion rating.

The inclusion rating measures how many of the players within a given range the expert puts in that given range. In the Arian Foster example above, I would only get a 1/24 (approximately 4%) inclusion rating. This is an equally important parameter when considering fantasy advice. If an expert consistently fails to include the top-ranked players each week they aren't doing you a service, even if they are giving you appropriate advice when looking at the players they do list. This is how we get our inclusion percentage; If you followed this advice exactly, how often would you play a top-tier player at each position over lesser players? This parameter was a late addition after I thought about the ramifications of using only head-to-head rankings. Before I was including all players into rankings and attributing phantom losses, but this got extremely complex from a computing standpoint. Inclusion percentages are markedly easier to execute and understand.

Finally, you might be wonder why the lists are broken into both truncated and full forms. The abridged lists are extremely important because they give you a snapshot of whether experts are truly giving you good advice. When considering quarterback ratings, there are only about 12 quarterbacks who are truly worth starting each week. It doesn't do the average fantasy football player any good for the expert to be perfect at ranking quarterbacks 15 through 24, because odds are you won't be playing those guys anyways. There should be a considerable amount of attention given to only the upper portion of an expert's list, so the head-to-head and inclusion ratings are given to the "starters" section of each expert's list in addition to just the full list.

Results

The unfortunate truth is that most fantasy football sites write over their weekly rankings every week. I grabbed the NFL.com rankings before they disappeared this week, but those are the only comparisons I have this week. In the future I will try to get more experts and individuals for comparison. It is a considerable amount of data input, so finding the proper formatting is extremely important. If you are interested in having your rankings calculated, send me some sort of message on reddit, email, or here on the blog so that I can let you know the proper format.



Quarterbacks:

Expert Name: [Top-12 H2H] - [Top-12 Inclusion] - [Full list H2H]

How2WinFantasy: [72.7%] - [91.7%] - [76.5%]

Michael Fabiano: [66.7%] - [83.3%] - [74.9%]

Dave Dameshek: [65.2%] - [83.3%] - [71.1%]

Matt Smith: [63.6%] - [83.3%] - [76.4%]

Adam Rank: [63.6%] - [75.0%] - [71.6%]

Jason Smith: [71.2%] - [75.0%] - [71.4%]



Running Backs:

Expert Name: [Top-24 H2H] - [Top-24 Inclusion] - [Full list H2H]

How2WinFantasy: [54.3%] - [62.5%] - [44.7%]

Michael Fabiano: [45.3%] - [66.7%] - [55.0%]

Dave Dameshek: [51.1%] - [58.3%] - [45.4%]

Matt Smith: [54.7%] - [62.5%] - [52.8%]

Adam Rank: [56.9%] - [62.5%] - [51.9%]

Jason Smith: [55.8%] - [66.7%] - [54.0%]



Wide Receivers:

Expert Name: [Top-24 H2H] - [Top-24 Inclusion] - [Full list H2H]

How2WinFantasy: [45.3%] - [54.2%] - [44.5%]

Michael Fabiano: [50.0%] - [54.2%] - [50.1%]

Dave Dameshek: [40.6%] - [58.3%] - [47.5%]

Matt Smith: [NA%] - [NA%] - [NA%] (He gave no WR or TE rankings)

Adam Rank: [48.9%] - [50.0%] - [50.0%]

Jason Smith: [46.0%] - [50.0%] - [47.6%]



Tight Ends:

Expert Name: [Top-12 H2H] - [Top-12 Inclusion] - [Full list H2H]

How2WinFantasy: [48.5%] - [33.3%] - [25.8%]

Michael Fabiano: [47.0%] - [33.3%] - [42.5%]

Dave Dameshek: [53.0%] - [33.3%] - [45.7%]

Matt Smith: [NA%] - [NA%] - [NA%] (He gave no WR or TE rankings)

Adam Rank: [57.6%] - [33.3%] - [39.6%]

Jason Smith: [50.0%] - [25.0%] - [40.3%]

No comments:

Post a Comment