The road to creating fantasy baseball auction dollar projections is paved with many decisions. Some are plagued with methodological potholes but, if done right, the resulting $ projections should basically get to the same place. (The player projections are the primary driver in differences between competent systems)
Since this is the time of the year where this topic matters, I thought I would shine a light on some of the less visible decisions that impact fantasy baseball auction dollar estimates. I will do my best to keep this from devolving into a mathematical or methodological exercise.
These musings will bleed into at least a second post. This first post is going to focus solely on Position Adjustments.
I have been spending a lot of time thinking about position adjustments this preseason. A significant portion of the fantasy baseball world seems to unconditionally believe that ‘position scarcity’ exists. EVERY draft has a number of C/2B/SS that would be drafted later if they were 1B/3B/OF.
Here are four theories on position adjustments – ranging from smallest to largest:
- Meritocracy Theory – Do no position adjustments.
- Apples-to-Apples Theory – Compare hitters against players at their position vs all hitters.
- Replacement Theory – Project with no adjustment then add/subtract $ to positions so that the last rosterable hitter at the position is at $1 (and one could subtract if the last rosterable hitter above $1).
- Communist Theory – Adjust so that the dollars allotted to each position are equal – e.g., in a 1 catcher league, the catchers get 1/13th of the total $ allotted to hitters.
The Meritocracy Theory is what I have adopted this year.
The Apples-to-Apples Theory is the one I followed prior to this year – though I hedged it a bit by comparing players 75% in their position and 25% against the average hitter (for AL/NL-only leagues, I did 25/75). Mike Podhorzer of FanGraphs/RotoGraphs has told me that the model he learned from Todd Zola follows this theory.
I think Replacement Theory is the prevalent one in the fantasy baseball world. I cannot speak for every $ generator but, for , this appears to be the underlying adjustment behind the FanGraphs Auction Calculation. It was the approach favored by expert Larry Schecter in his Winning Fantasy Baseball book. I think most people who manually rank or determine player dollar values conciously or subconciously follow this theory and boost up C/2B/SS (code words: ‘scarcity’, ‘shallow’).
I have never heard someone promote the Communist Theory but it is the logical extreme of position adjustments.
I think mathematically proving the correct theory with league data would be impossible as a good or bad year at a certain position might skew the data. Picking Tulo at the end of the 1st round was not brilliant in 2015 before his injury and dumb after his injury. Maybe if I review multiple years of RCL there might be a little signal within the noise. I tried with RCL 2013-2014 data and middle infielder draft investment was negatively correlated with a team’s Hitter Points in RCL 2014 (-7%) and positively correlated in RCL 2013 (+6%).
If I had to guess on the impact of position adjustments, I would say they have very close to zero impact (unless someone screws up the math or goes way off the reservation). While drafts have a significant impact on final standings, the impact of a couple picks is very small. I finished 3rd out of 1,008 teams in the Razzball Commenter Leagues and drafted Justin Verlander/Allen Craig in the 4th/5th rounds. I drafted Jose Altuve in the 6th round of the KFFL BAD draft and finished 14th out of 15. Even if the ‘best’ strategy was to not make position adjustments and you jumped three rounds for a SS you loved, it is hard for me to see this having anything but negligible impact.
Rather than focus on the ‘right’ way for position adjustments, this post is about understanding the impact of these various methodologies and helping you determine the best one for your draft preferences. I will also try to point out the possible unintended consequences of position adjustments.
Below is a distribution of auction dollars per hitter position for 12-team mixed (C/1B/2B/SS/3B/5 OF/MI/CI/UTIL) using the results from my 2015 Player Rater, the 2014 Razzball Commenter League ADP (converted to $), and the FanGraphs Auction Calculator (Note: I did not adjust hit/pitch splits so focus on the percentages/indexes versus totals).
I cut this two ways – the top portion ignores CI/MI and I just counted the top 3 OF. For the second, I incorporated CI/MI (crediting 1.5 for 1B/2B/SS/3B) and all 5 OF. I just included all the catchers in each and ignored DHs. The two indexes compare each source’s distribution against the ADP (a gauge at what ‘real’ drafts look like) and the second shows the index for each source per player (e.g., The Razzball Pre-Season’s 150 for 1B means that 1B dollars are 50% higher than the average auction dollars per hitter).
|Total $||% of $||Index vs ADP||Index per average player|
|2015 Razz Pre||RCL ADP 2014||2015 FG Pre||2015 Razz Pre||RCL ADP 2014||2015 FG Pre||2015 Razz Pre||RCL ADP 2014||2015 FG Pre||2015 Razz Pre||RCL ADP 2014||2015 FG Pre|
- For 12-team mixed, the Razzball Player Rater and FanGraphs calculator are similar in their $ allotments per hitting category with 1B/2B/3B having the biggest differences. This appears driven by FanGraphs position adjustments which give 3B the smallest position adjustment (behind 1B/OF which is clearly stronger based on my $ allocation and past experience). So while my Player Rater suggests spending the average dollar amount on 3Bs (index 101) and the market spends 8% above average, the FanGraphs calculator suggests 15% below average. I think this speaks not only to the volatility of anchoring adjustments on the last player at a position but also to the challenge in where to place multi-position players in the replacement calculation. FanGraphs appears to follow my ‘order of operations’ of C/SS/2B/3B/OF/1B/DH. This means that the following players with 1B/3B or 3B/OF are credited to 3B and make it look deeper than 1B and OF: Todd Frazier, Chris Davis, Carlos Santana, Josh Harrison, and Ryan Zimmerman. (Note: the adjustment may lump 1B/3B together which would simultaneously help 1B and hurt 3B)
- While FanGraphs does boost Catchers the most of any position, the investment is (oddly) still below my $ distribution with no positional adjustment. All three sources are in eerie alignment on Catchers.
- The biggest deviation between Razzball/FanGraphs and the RCL ADP is in Shortstop where total investment is 5% lower than the average hitter whereas the raters say it should be about 35% lower than average. Since the average hitter is worth about $13, this means that the RCL, on average, puts a $4.50 premium on SS.
Things get more interesting when running the same analysis using tlhe NFBC 15-team rosters (which use 2 catchers). Instead of 2014 RCL ADP, I am now using 2015 NFBC ADP.
|Total $||% of $||Index vs NFBC ADP||Index per average player|
|2015 Razz Pre||NFBC ADP 2015||2015 FG Pre||2015 Razz Pre||NFBC ADP 2015||2015 FG Pre||2015 Razz Pre||NFBC ADP 2015||2015 FG Pre||2015 Razz Pre||NFBC ADP 2015||2015 FG Pre|
- With the addition of the 2nd catcher, the FanGraphs model following Replacement Theory adds a huge bonus to each Catcher of about $12 – all in an effort to get the last rosterable replacement catcher value to $1. The premium placed on catchers vs my estimates is about $6.60 on the first 15 catchers (about $100 total difference) and $5.56 for all 30 catchers.
- One quirk you may have spotted is that my system values catchers 16-30 at a sum of negative $19.3. Why? Because I am giving catchers a very mild boost (about $1) to ensure the 15th catcher is at $1. I basically then let the $ operate as a meritocracy and the results clearly show that the bottom half of catchers are uniformly below average compared to the other rostered hitters in this format.
- Yes, I have to bid at least $1 for my 2nd catcher and it makes little sense to say they have a negative values. I don’t care. Here is the key decision I make – do I want to invest 6.3% of my total hitter budget on catchers or 14.6%? Personally, I HATE investing in catchers (burns me almost every time) so investing less means I now have 8.3% of my hitting budget (about $15) to spend on my other hitting slots than someone using Replacement Theory. If a Catcher comes at a nice discount, maybe I will grab one in the middle of the draft. Note that NFBC ADP splits the difference at 10.9%.
- With those extra $, the Razzball Player Rater is investing a bit higher in all positions with the FG calculator except for SS (a near draw).
- Based on the ADP indices, both the Razzball and FanGraphs raters feel that the NFBC players spend too much on 2B/SS and should divert it to 1B/OF (mine also suggests 3B).
1) In the key areas where my non-adjusted Player Rater diverges from FanGraphs position-adjusted Player Rater, I feel better about my results. Investing less in Catcher is admittedly subjective. I would argue that the 3B undervaluing is a quirk that one could resolve with a more sophisticated solution.
2) The ‘market’ (as defined by ADP) invests more in 2B/SS than the Raters suggest at the expense of 1B/OF.
So I will now frame the concept of positional adjustment as follows:
1) Forget all the math/theories you have learned on positional adjustments and think about how you want to invest your $ across positions. While I like the simplicity of going without positional adjustments, I am making the decision because I feel comfortable in a) Severely limiting my Catcher investment in two catcher leagues, b) Feel that no position deserves a premium above their position-neutral worth, and c) I want as much of my hitter money as possible going to the best hitters regardless of position. If this means I have a loaded 1B/3B/OF/OF/OF/OF and weaker C/2B/SS, so be it.
If you want to invest more in C/2B/SS, that is fine. Just do not feel anchored by a theory that (in my opinion) causes more issues (i.e. the 3B glitch) than benefits. So instead of doing a full adjustment on Catchers, make a partial adjustment. As long as your total $ adds up to $260*# of Teams, you are fine. If your last 2 SS are worth -$1 and -$2, it is not the end of the world. There is no mathematic imperative that requires you to boost every other SS $3 and re-distributing 2% of your $ budget away from other positions.
2) If you just do rankings, this exercise can help uncover biases.
I ran a similar analysis against Grey’s rankings, ESPN’s Top 300, and the FanGraphs rankings for Mike Podhorzer and Zach ‘Z-score’ Sanders. (Note: converting rankings into $ is easy. Just cut/paste my Player Rater for that league format and credit my #1 player’s $ for your sources #1 pick, etc.).
For Grey, his antipathy towards catchers comes through loud and clear but I think his 2B/SS preference was higher than he might have guessed (He sees this as a coincidence because he just happened to like certain 2B/SS this year. I believe him as last year he was slanted towards 3B).
For ESPN, they are pretty much in line with my distributions with a slightly greater preference for 2B/SS than me at the expense of 3B/OF.
FanGraphs’ Podhorzer and Sanders are eerily similar in their category distributions and invest more in C/2B/SS vs 1B/3B/OF (their low 3B investment mirrors that of the FanGraphs calculator. In talking with Mike, he mentioned the methodology he uses treat 1B/3B the same. Given the wide disparity in their value on my unadjusted PR – 154 v 101 – I can see why 3B is hurt by this methodological choice).
|Average $ Per Position Indexed Against Average Hitter $ (~$13)|
|Razz PR||RCL ADP||FG Calc||Grey 2015||ESPN 2015||FG Mike||FG Zach|
If you found a ranking source you liked in general but felt they were off at a position, this type of exercise could be used to calibrate it as you see fit.
3) By better understanding your dollar distribution by position, you can better focus on the player pool that you will likely draft.
I think most people do this anyway (a comparison vs ADP often does the trick) but it can be overwhelming to have a point of view on every hitter/pitcher. Based on my distribution, I can pretty much ignore the top tier of catchers, 2B, and SS and focus on finding the late-round bargains in those positions.
4) Take a moment to think through your logic behind paying a premium (vs their position-neutral value) for hitters at weaker positions. I understand that because you need C/2B/SS/MI, it does not matter if the market overvalues it. The goal of the drafter is to pay the smallest premium possible. So if someone determines that the premium on 2Bs is always two rounds or, say, $4, perhaps it makes sense to go for Altuve/Rendon/Cano instead of a 1B/3B/OF. Here is the key fault with that logic:
- Once teams fill these positions, they (for the most part) stop drafting these positions. So even if EVERY team values shortstops $4 more than you and they maintain this premium throughout the draft and they have the same top 17 SS as you, that means you are still getting the 18th best SS in a 12-team league at your price.
- Once you get past the first couple 2B/SS, your player values and your competitors’ player values are likely going to be different. You may see Howie Kendrick as the 10th best 2B while the draft room thinks 16th. So, it is definitely possible that by waiting on weaker positions, you will eventually pay face value based on your projections instead of a premium. In my experience, this happens around the midpoint of a snake draft and/or $10 range players in auctions.
- There is also a secondary benefit to waiting on C/2B/SS. While drafting a lower ranked player generally brings more risk in playing time or performance, no one in your league is going to pick a player at these positions up for UTIL. And those who invested heavily at these positions are ignoring the waiver wire. Thus, these positions have less competition on the waiver wire than corners and OF. (This reason alone makes investing in Catcher in shallow leagues a dumb play IMO. I still fooled myself into doing it the last two years in RCL but fool me twice…)
In Part II, I will be tackling the concept of balancing $ values across categories.