Fake Teams Consensus Rankings - Projections Part I

Introduction:

Yesterday, you hopefully saw Ray's post announcing the schedule for the Fake Teams 2014 Consensus Position Rankings. Starting next week, you'll have the best morning, break, lunch, break, ride home and bathroom-time reads.

As part of the rankings, my 2014 Player Projections will be provided for reference. My rankings are based solely on their 5x5 output, but that's often not enough to evaluate a player's true fantasy impact. That's why the contributors here at Fake Teams will (consensus-ly) delve into each position to provide player profiles and prospect impact so that these rankings can also provide value for your different league types (keeper/dynasty and even AL/NL formats). Naturally we'll provide you with targets, players to keep at poles' length, but we'll also focus on ADP trends. Think of it as a weekly all-encompassing positional draft kit.

If you need further analysis, you can send requests to us at Rotobanter, where we'll either post content based on the questions or aggregate the questions for Fake Teams' podcasts. Comment away in each post to banter with the author and other Fake Teams contributors.

Onto the Projection Approach:

I'm not going to provide the ground-up process, but 3-year historical data is pulled from FanGraphs through a customizable player profile format that parallel's my projection tool. The data is outputted into 3 year averages as well as trended/weighted based on age and skill-sets (through certain ‘clusters' of peripherals). Most projection systems automatically trend certain skills and/or regress other skills back to their career/3-year average - including mine, but I keep the ability to go in and make manual adjustments if I catch some significant new approach. One example is Justin Masterson's 7% fastball-slider swap last year. This is something that I think is sustainable (since it's a repertoire change) so long as the additional breaking-ball usage doesn't cause him strain. I'll get more into Pitch F/X and the incorporation of that when I post the pitcher projection approach on Thursday. This post will focus on A) how I personally rank the players, B) adjust for position scarcity and C) associate auction values. The specific hitter projection approach will get posted Wednesday morning. The pitcher projection approach will get posted Thursday morning.

...Now on with the Borophyll...

Rankings, Positional Re-adjustments and Auction Values:

While the hitter and pitcher specific segments might provoke some banter, it's this section I can see stimulating zeal. I'll break it down as follows:

1) Projection Integrity: $ values are only as good as the projections associated.

2) Z-Scores vs. "SGP": There are other approaches to landing on fantasy and $ values, such as "SGP" or Standings Gain Points. Technically this approach is more parallel to Rotisserie Leagues in that you're trying to be in 1^st place in a specific category (or all categories ideally), but in my opinion this incorporates additional variables into your evaluation...not only do you have to put players on the same (categorically) and different (positional) scales, but now you have to make assumptions on your league type e.g. Last year you needed 175 Stolen Bases to win that category, but in 2012, you only needed 155. I prefer categorical (and then summed) z-scores (number of standard deviations that value is away from the categorical mean). So for homeruns, you take a player's HR total, subtract it by the average HR total for players with a significant enough amount of at-bats (400) and then divide that by the HR standard deviation: HRtotal - meanHR)/ sdHR.

If you want the z-scores/categorical values for all the 5x5 stats from my projections, go into the 2^nd tab here.

3) Z-sum/Fantasy Value: Sticking with the 5x5 stats, we find the z-score for each 5x5 category - R,HR,RBI,SB and AVG which has to be weighed by AB and then re-z-scored. The sum of all 5 z-scores is our z-sum or "Fantasy Value."

4) Positional Adjustments: ****It's important to z-sum before the positional adjustment because initially, we want to understand the actual categorical values in relation to the full player pool. We don't want to understand the categorical values within each position because that will heavily and overly weigh stolen bases for catchers and first baseman as an example. Some think adjusting for positional scarcity is unnecessary...you're simply trying to accumulate the most counting stats (and best average), but the argument and reality is, If you have two players with the same exact surface stats and one is a catcher and one is a first baseman, then a hundo% you want the catcher.

Another important clarification - I didn't make this stuff up. It's what Zach Sanders on FanGraphs calls Fantasy Value Above Replacement.

After you have the z-sum or "Fantasy Value," it's now time for the positional adjustment. In order to adjust by position, we need to understand the number of teams and roster requirements i.e. how many 1b you require so that we can define your "replacement player" in each position. It's not as easy as defining your 13^th 1B as the replacement player (in a 12 team league) because teams are also probably drafting 1B for the corner infielder and utility slots in their lineups.

***I will not only look at the roster requirements. I will also look at the fantasy value (z-sum) drop-off levels. Another possible option is finding the average replacement z-sum for each position (which Tanner Bell of Smart Fantasy Baseball (and I'm sure others) use).

Let's say 24 first baseman wind up being drafted to man the 1B, CI or U spots on your roster, that means the 25^th 1B is your "replacement level." This happens to be Mitch Moreland who has a -1.68 z-sum. This means that him and every other 1B (above and below his fantasy value) gets 1.68 added to their z-sum value. Freddie Freeman which was at a 5.83 is now at a 7.51. -1.68 is our 1b replacement value, which is highlighted in bold below. The fantasy value is the ‘zSum' column; The positional adjustment is the ‘Pos.Adj.' column and we'll get to the $ value in a bit.

Now let's look at catchers. Offensive values at the catching position is a shit-show. So let's say a 12 team league won't draft more than 12 catchers. Therefore, the 13^th catcher is our replacement value. According to my projections, this winds up being Yan Gomes who has a -3.33 z-sum. -3.33 is our catcher replacement value, which is highlighted below in bold. Every catcher gets 3.33 added to their fantasy value.

Now you can see how we came up with our positional scarcity adjustments. Based on the replacement player within each position, we took his z-sum/fantasy value and added that value to himself and everyone within the same position. The replacement player's new "Pos.Adj." value is 0.00 which will soon become $1.00 based on the $ value formula. For catchers, we added 3.33 to everyone but for first baseman (less scarce/more offensive value deeper within the position), we only added 1.68 to everyone.

As a reminder, the replacement value will change based on a) The Projections and IMO, the "drop-off value" b) the roster requirements c) quantity of teams.

5) $ Values: Again, right from Zach Sanders' series: [(260-(1*23))/23]*(PosAdj fantasy value / average PosAdj fantasy value for above-replacement players) + 1.

A) The "+1" is what gives us our $1.00 replacement player

B) "260" is our budget

C) "23" is our number of active roster spots

D) "PosAdj fantasy value" is the z-sum adjusted for positional scarcity or what we spoke about in step 4 above

E) And finally "average PosAdj value for above-replacement players" is basically the average of all the "PosAdj" values for the top 276 players (active hitters and pitchers on your team).

I'm not going to furnish a full $-value sample at this time, because I don't have my hitters and pitchers all integrated yet. But let's say the average positional adjustment fantasy value for above-replacement players was 3.00), the $ value formula would look like this for Freddie Freeman = ((260-(1*23))/23)*(7.51/3.00)+1 = $26.80. Again, "7.51" is FF's positionally adjusted fantasy value, which you can find in the first matrix under Step 4 above.

Look out for the hitter projection approach on Wednesday morning and the pitchers on Thursday morning.

If you're baffled or want to touch base on any of the steps/approaches above, you can follow Dan on twitter @Rotobanter or sign up for free on Rotobanter's Fantasy Baseball Forum.