The Invisible Hand and the Price of the 36th Man

At #TDGX, a silent auction system for bidding future keepers on a draft slot was introduced. A couple of the experts in that league broke down their thoughts on bidding for Mike Trout.

Mike Buttil at TDG wrote about his team’s decision to bid 15 keepers for Trout

Mike Newman at RotoScouting wrote about his decision NOT to win Trout

In The Dynasty Guru Reader League (#TDGR), we copied the invisible hand system and the winning team for the first pick bid 20 keepers. This, at face value, seems to me crazy. BUT its impact on the economy and value of 2014 keepers is worth exploring. It’s early still to see how this plays out, but I’m going to hypothesize.

First, some breakdowns:

  • Everyone in #TDGR knew about the expert league, so we all knew what was spent for the slots in that format
  • In every other way this reader league mirrors the expert league, so the economics of draft slot bidding should be more or less identical EXCEPT this is a brand new system so the market hasn’t had a chance to correct itself yet
  • The Reader league is dealing with readers, but combined the 20 managers have over 230 years of fantasy baseball experience. 

Now I’m going to review the bids for the first slot, and the winning bids.

Breaking Down Bids

 

Image

As you can see from the table above, there are only 4 teams who didn’t submit ANY bid for Trout. The average bid rounds to 8 keepers. The top 3 bids were 20, 18, and 16. Perhaps notably, no one was willing to bid the same amount that Mike B at #TDGX bid for Trout. 

So what happened here?

My theory is that 17 teams saw 15 as the going rate for Trout, put in a token bid thinking maybe they’d get lucky (maybe everyone else saw 15 as too much too and then getting a tie for 10 could be really worth it!). 3 teams saw 15 as the going rate and decided they wanted Trout enough to bid up. These guys knew if you wanted Trout, you had to get crazy. They were willing to get pretty crazy.

The Mikes’ linked above get into the economics of Trout more, so I’m going to leave this for our edification.

Now let’s look at the total bids across all rounds. I’m going to show the winning bid compared to the average:

Image

I calculated all bids here, so if a team won pick 3 but also bid on every other pick, that is part of the average.

FWIW, that’s me at pick 19 bidding 1 keeper. 

Why it’s worth bidding

Nearly everyone in #TDGR bid on some pick. Remember, by default everyone carries 35 keepers into the next season and throws back 5 players. The keepers bid are EXTRA on top of the 35, so the Mike Trout team will be throwing 25 keeprs in, pick 2 will throw 15, etc. To decide whether to bid or not, you basically must pick from the following choice

After the 2014 season, my team

  1. Will have 35 players worth carrying into 2015
  2. Will NOT have 35 players worth carrying into 2015

You have to make this decision for yourself. HOWEVER, knowing that the first 3-5 picks will be worth substantially more than every other pick, you have to consider what the 2015 player pool will be.

Normally, the player pool for this league in each year’s 5 round draft will be the available MLB players (including the 100 players we’ll always throw back, but also any otherwise unsigned player) PLUS any new Rule 4 draftees or International Free Agents who are not eligible to be picked up the previous year. Most years, the bottom 5 guys on any given team will probably be busted prospects, retirees, or players so far below replacement level that they’re essentially free via FAAB. One should figure that of the 100 players not kept, maybe 25%-50% of them could be drafted in the last rounds but probably weren’t worth acquiring via trade.

HOWEVER, after this first year the player pool will be the new rule 4 draftees/int’l FA’s, PLUS 163 players not kept, plus everyone else.

My theory here will be that of the 163 players not kept, more than 50% will end up being drafted in the 2015 5 round draft. Coupled with what is supposed to be a strong crop of 2014 rule 4 draftees, the 2015 draft player pool in this league should be fairly deep.

BASED ON THIS, I’m speculating that the value of the 35th man on my roster changes for 2014-2015. When I’m going to have to decide who to keep and who to cut at the end of the season, I have a few choices to make. Primarily, where am I going to draft? Then, who do I think will fall to me in the first round and is he better than my 35th man? 

My guess is that, with the depth we’re facing, for EVERYONE in the league the answer to this will be “yes, the player in round 1 of the 2015 draft will be better than my 35th man.” Now, suddenly, it may not be worth carrying all 35 keepers. Getting into that supplemental round may return some value. I wouldn’t be surprised, for instance, if some of the teams who didn’t lose any keepers still decide to throw and extra 1 or 2 back just to see what they get in the supplemental rounds.

Another Minor Factor

For me, it was worth bidding something to get near a turn. This is going to be a slow draft and knowing I’ll have picks close together had value to me. With that, I wish I had bid more for picks 3-5, but I’m happy I got pick 19. If you decide to join a similar league with a deep, long, slow draft, take this into account. If you’re doing a live draft, this isn’t really that important and I think that significantly impacts the way one should bid.

Parting Thoughts

The value of one of the first few picks here can be debated, but at least for 2014 I do believe there is value. There will be a total of 800 players drafted in 2014, and ~737 coming back to 2015. With 23 man rosters, that means 460 players will be actively rostered with the remainder as bench pieces, depth, or minors. The first round of the 2015 draft in this league should be rich with talent, either via upside in prospects who are now eligible or keeper pieces who could bolster a bench and provide depth. By the time the first 100 guys are taken in that 5 round draft, the supplemental rounds will begin and at least 7 people will participate in this first round. I took a gamble, betting to get near the turn, that I’d get something back in this supplemental round better than whoever my 40th man will be at the end of the season. That’s a gamble I think everyone in this league should have been willing to make.

 

#TDGReaders League

I’ve been listening to the TINO Podcast and following The Dynasty Guru Expert league draft, and decided now is the time to take the plunge and initiate a new, first year dynasty league. The #TDGX format is so intriguing for a few reasons that I wanted to exactly mirror that format. Here I’m going to track some of my thoughts in why I set up the league and how I felt I needed to modify it for readers rather than “experts.”

For reference, here is the constitution I created

What’s Great about #TDGX Format

  • Depth– 20 teams, 40 man rosters means 800 guys are drafted. 35 are kept per team which means every LOOGY worth his salt, every UTIL player, every hyped Low-A prospect guaranteed to bust by 2017, will be owned by someone at some point. 
  • The Draft System– the “invisible hand” method outlined in The Dynasty Guru blog post is, to me, an incredibly smart way to handle the “first year draft” problem. Lottery is patently unfair, letting teams bid on picks and placing a real penalty on it adds a level of complexity to the initial draft and future drafts. 
  • Experts. Setting up my league to be identical to a specific, publicly discussed league, and opening it to fans of the experts involved means we’re all working with identical source information. It will be fun to see how the ‘reader’ draft compares to the experts, and how the 2 leagues start to differ over time. Participants are going to know about every draft pick, free agent, transaction, and strategy that gets written about in #TDGX posts, which means any individual team advantages are going to be all the more subtle. 

What’s Different About a Reader League

Putting together the reader league, I was faced with a few key challenges, outlined here:

  • Keeping it free: #TDGX will be a CBS league. This lets them manage everything outside of the draft, including minor rosters. For the first year, I’m more interested in filling the league and getting it started quickly, so putting a cost on it seemed to be a barrier to me. This meant CBS was out of my budget, so we’re using ESPN for the Active Roster/bench management and will need to track minors via spreadsheet. More commissioner set up time
  • Keeping it competitive– I imagine the #TDGX league should have no attrition problems or owners who bail in July. This isn’t guaranteed in a reader’s league, especially a free one. Identifying some ground rules for involvement requirements was necessary for me to draft a constitution.
  • Filling in the holes– there are some things that #TDGX’s blog posts and tweets don’t specifically detail, so I made some personal gut-level decisions. For instance, #TDGR will allow trade of draft picks (even in draft) for the current and following year. There will be penalties for improper roster management (if a owner claims a minor FA via auction then forgets to add him to the minors spreadsheet, that owner can lose both the player and the auction dollars). There are gut-calls I’m making as a commissioner (I probably won’t ever veto any trade). 
  • Balancing power– This is a lot of work for me to do, so I’m going to need to figure out how to divide responsibilities. I figure I’ll ask another owner to step up and manage tracking the draft, since it will probably last 10-20 days. I may need to put together an ‘advisory board’ to discuss potential issues or constitution changes. We’ll see how that goes.
  • The traditional roto format– some people love this format and have been playing in it for decades. I, personally, have philosophical issues with it as a baseball fan. But it’s the rule of #TDGX, and it may represent the biggest challenge to me.

Winning the Unwinnable: The War Room (Snake Draft Version

Writing about mock drafts is one of the more common, perhaps easy topics for most fantasy sports writers/blogs/sites. Most mock draft articles focus on the results, rather than the process, perhaps with a sprinkling of insight or strategy beyond “what do you think of my team?” or “I can’t believe Anthony Rizzo fell to the 7th round!” or whatever. I have no time or patience for this sort of article, which is why the ‘Snake Draft 401‘ and ‘Auction Strategy and Strategies‘ articles on FanGraphs+ were a pleasure to read. The $5 a year’s subscription to FanGraphs+ is one of the better fantasy sports deals, and I’d recommend the investment highly.

Regardless of your draft type (auction or snake), the snake draft 401 article specifically has a lot of good advice on creating and working with spreadsheets. There are plenty of sources to download war rooms, player raters, draft trackers, and other jawns but there’s some understated joy in the time, energy, and complete geekiness in creating your own spreadsheet to evaluate players and manage your draft.

While the draft itself is the battlefield, your war room spreadsheet will save your neck in battle. Using Google Docs, I’m putting together my own war room to test in mock drafts and auctions and ensure I’m meeting my team goals and not getting too caught up in the excitement of the next pick that I miss my overall vision for my team.

My war room is probably in Beta mode, and already has a separate sheet for each position (with rp and sp separated), a sheet of all hitters/pitchers, a sheet for all players, a sheet with draft round tiers, and a home sheet for managing in the draft. Each sheet has its own function, which I’ll outline here in the order of their utility during the draft.

A Picture of The War Room Sheet

A working version of my war room spreadsheet

A working version of my war room spreadsheet

The most important sheet is that which I’ll use to track my draft picks for each turn, and manage my actual draft. Starting from the bottom and working up, I’ll explain what I have here:

  • At the bottom is a list of rounds and picks. Following the Snake Draft 401 article, I created a simple sheet listing every pick in a snake draft in a row. Then, when I found out what pick I had, I could copy and paste that row and know which number I’d be picking for the entire draft. This makes it easier to know when you’ll need to reach for a player, or when you might be overpaying
  • On the right  is a list of the players picked with each round. Whenever I picked a player I would copy and paste their stat line from the ‘all players’ sheet.
  • Top left is the cumulative projected stats of this team. I have this set to add up all the stats from the right-hand list, so that as soon as I add in a new player their projected stats are added to my team total. This is indispensable in figuring out what categories to target with the next pick. I’m using a slightly modified version of the steamer stats found on FanGraphs, simply because Steamer uses FG playing time projections and my research has shown these to be reasonably accurate. I probably trust the ZiPs projections a little more, especially with young players or projected rookies, but especially in mixed leagues where there isn’t so much depth, I’ve found Steamer to be fairly reasonable.
  • Below this on the left are a few fields that I’ve put together based on speculation and projections.
  • Team Goal: This is my guess at what it will take to win each category, based on the winners from my leagues last year and some inflation or deflation based on other research. Above this, just below the draft projection  total, I’m calculating the percent of my projected team stats against the overall goal. I’ve targeted a balanced team focusing on 75-90% in each category, but you can see if my team performs as projected I should expect to win the Runs category and could win AVG and RBI too, meaning the weaknesses of my pitching staff will be overcome.

I have tested this against a few mock drafts, and the format makes it easy to track draft progress without clicking between too many items. In a fast moving snake draft it’s still possible to keep up with this without missing a pick or overlooking a need. The true benefit to this is in testing my team in real time against a goal, so that I can determine with pick 20 whether my team needs SB or pitching depth more, or fill other holes.

It’s easy to modify and add in an auction calculator. I’ve begun testing this and will write about that in a future update.

Winning the Unwinnable: CAIRO Mock Draft Results

R HR RBI SB avg
1015 274 1119 116 .273
W Sv SO ERA WHIP
80 123 1107 3.78 1.284541

Take a look at those numbers. How many wins do you think that’s worth in an average 12 team 5×5 roto league? I’d guess that team would place between 3rd and 5th if it can finish in the top 2 in R, HR, RBI and SO. There are obvious areas to improve but it’s not a bad place to start.

Now imagine your offense looks like this:

Brandon Belt 1B
Ben Zobrist 2B
David Wright 3B
Pablo Sandoval 3B
Pedro Alvarez 3B
Asdrubal Cabrera SS
Jed Lowrie SS
Buster Posey C
Mike Napoli C
Matt Holliday LF
Carlos Beltran RF
Josh Willingham LF
Andre Ethier RF
Adam Eaton CF

Think that team can put up the numbers you see above? I don’t. You probably don’t. CAIRO does.

As part of my ongoing Mock Draft series I am reviewing various mock draft strategies and results to see what this can teach us about planning for a draft, learning from mistakes, and judging various methods of ranking and selecting players. I’ve outlined best practices for mock drafts and established my method and guidelines for testing. Here is the first in a series of results of testing

Mock Draft I

  • Date: January 29, 2013
  • Site used: Mock Draft Central (see draft results here- teamname Bushleague Avocados)
  • Scoring/Roster: 12 team 5×5 roto; 23 team rosters (2x c, 1b, 2b, 3b, ss, 5x of, ci, mi, util; 9x p)
  • Draft position: 6th
  • Strategy/Ranking System Used: CAIRO
  • Strategy notes:  I used CAIRO v 0.3 projections, sorted by oWAR for batters and WAR by pitchers. When in doubt I picked best available for a position I had not yet filled. Note also that officially my last pick was supposed to be Joe Nathan, and ended up being Nathan Jones. This is what happens when you are hungry and search by RP named “nathan”.
  • Projected end of season rank: 12th (ouch)
  • Gut reaction, hitters: CAIRO has 8 3b and 4 catchers in their top 30 by 0WAR, compared to 1 2b (Cano) and 1 SS (Reyes). It seemed to me at least that CAIRO is optimistic on older players or undervalues younger; either way it frustrated me to have Josh Willingham as my next top player (ranked 39 by oWAR) in the 9th round when it seemed like a lot of potentially more valuable guys were ranked as worse by CAIRO.
  • Gut reaction, pitchers: CAIRO’s top 20 pitchers by WAR is a decent list ranked indecently. That it predicts Fister to be approximately as good as Kershaw is good news for Fister, who is being drafted fairly late. Matt Harrison also had an impressive projection that placed him as a top 15 pitcher.

Parting Thoughts

  • This draft was purposefully rigid in sticking with the results of the ranked CAIRO list against a very specific metric of oWAR/WAR. These are not ideal fantasy ranking systems and to arrive a better ranked list I would have put more effort into isolating the fantasy-specific projections. This list was also compiled using v0.3 of CAIRO, which is still under the works. I chose it because at the time of the draft it was the most up-to-date complete projection publicly available
  • A simple lesson from this, one that most of us probably already know: regardless of the list you are using in the draft, don’t be too rigid. Rigidity will find you overvaluing or undervaluing players you may be targeting, resulting in Mike Napoli in the third round (even if I do intend to use him as my 1b, that’s early for a guy whose hip is about as healthy as my grandmothers’).
  • Another, equally simple lesson– while not without fault, projection systems such as CAIRO are a smart place to look for sleepers and breakout candidates. I will trust a projection system that utilizes a few years of data more than I will trust my gut when evaluating a player I don’t know much about.  A great or horrid 2012 does not necessarily mean a repeat in 2013.
  • In my real live draft prep I will use a few projection systems to compare and supplement my personal targets. The benefit of even a faulty projection system is that it projects all players using the same metric. Looking at 2 or 3 systems and finding players that are ranked highly across all of them could give you an indication of that player’s projected skill.
  • In the end this mock draft was a success not because I love the team I ended up with, but because I tested a strategy and identified areas to improve upon it. I won’t be drafting from an unedited projection ranking much in the future, and no one would ever recommend that be the end-all of your draft prep, but it does give a healthy baseline to test against for future draft practice.

I’ll discuss some of the available projection systems’ pros and cons in a later post. For my next mock draft post I’ll discuss a Yahoo! mock draft, and the problems that face drafting in low stakes situations with a bunch of strangers.

 

The Quality Start, pt II: Projecting

In part I of my n-part series on the Quality Start, I outlined why this stat should A) be understood to have no real baseball value and b) be understood as a fairly ideal fantasy stat. To sum, the QS is an ideal replacement for ‘wins’ in assessing fantasy pitcher contribution in any format, because it relies more on the skill of the pitcher than on the circumstances beyond his control, because it is easy to understand, and because it is projectable.

As a minority fantasy stat, the QS still does not appear in most easily attainable projection systems when you are doing your own player evaluation. But given a few bits of easily attainable information, it is much easier to predict with some accuracy how many quality starts a pitcher can be expected to have.

A simple cursory web search revealed a few different equations for predicting QS, and I tested two of these against 2012 data. Mr. Cheat Sheet and Cup of Fantasy Joe both outline pretty good forumlas, and you should read those links for insight into how the formulas work and what they suggest you learn from them. Comparing the formulas from each site against all SPs who had 10+ QS, I found the following:

Cup of Fantasy Joe and Mr. Cheat Sheet Quality Start formulas compared against actual 2012 data for pitchers with 10+ QS

Cup of Fantasy Joe and Mr. Cheat Sheet Quality Start formulas compared against actual 2012 data for pitchers with 10+ QS

From this data it’s fairly clear both CoFJ and Mr Cheat Sheet are close enough in projecting QS numbers against actual data that I would leave it to preference in choosing which formula to work with. I found Mr Cheat Sheet’s formula a little easier to wrap my math-inhibited mind around. It could be that against a larger pool of data, the discrepancies become clearer.

 

If you really like tables, here is a list of a few top, middle, and bottom pitchers with their 2012 actual QS, compared against what COFJ and MCS’s forumulas project using the same 2012 data:

A sample of top, middle, and bottom-tier pitcher Quality Starts compared with various formulas to project the same

A sample of top, middle, and bottom-tier pitcher Quality Starts compared with various formulas to project the same

 

Looking at this chart the discrepancies between the projection formulas appear a bit more extreme, and it becomes clear that in reality a pitcher will over- or under-perform his projection based on a variety of factors. Both COFJ and MCS weigh games started, total innings pitched, and ERA in determining QS, and while it is clear these numbers can get very close to accurate results, it is also clear that the margin of error is probably about +/- 3.5 QS or so. Again, I’m math inhibited so this is an eyeballed guess.

One thing that I feel this table DOES reveal is the degree to which a Win is less representative of a pitcher’s true talent than the QS. Again, fantasy stats are much poorer at evaluating true talent than advanced stats, but if we all agree that Ross Detwiler is worth a bit more than a 2012 Tommy Hanson or Ivan Nova, QS better reflects that.

 

You can view the full spreadsheet of 2012 quality start comparisons here.

Some speculative conclusions

A few quick, dirty conclusions based on this info:

  1. Even if you do uses Wins, the QS formula can give you a measure of talent for a pitcher to help ground a Wins projection more into reality than a simple guess.
  2. You could avoid projecting QS or any game-performance projection by focusing instead on IP and ERA. A 1, 2, or 3 starter who avoids injury and gets deep into games is a good bet to put up strong QS numbers.
  3. As with any ranking, the results of the formula you use can only be as good as the projection system you rely on. Projection systems are generally more reliable than individual gut-feel, but the general rule of thumb is that more meaningful data leads to better projections. Trust the numbers you get for an established pitcher with several years of service time more than the numbers for young guys.
  4. Especially in head-to-head leagues, QS is a useful SP evaluating stat. In roto leagues, a more ideal stat may be some WAR-based evaluation that is more context neutral. But in leagues where you have to win week to week, QS is less likely than Wins to punish good starters who pitch for crappy teams or reward mediocre starters on offensive powerhouse teams.

Regression to the Mean, with Chart

In the most recent episode of Fangraphs Audio, host Carson Cistulli speculated on the merits and possibilities of illustrating, graphically and with all emphasis on brevity, the very concept of “regression to the mean”. Below, in at least 50% graphical format, is a chart meant to convey this very idea.

The problem, as discussed, is that fans of Team Y or Player X will assume a year of poor performance is an outlier while an especially good year is representative of true talent level. What this chart shows is that, as with most things, reality is more complicated than we may speculate and a player can both be better than their worst year and worse than their best year, and that indeed such an assumption is healthy.

Michael "OK at Baseball" Saunders

A graphical representation of the concept of regression to the mean as illustrated using Seattle Mariners CF Michael Saunders 2013 ZiPS projections

To illustrate this concept I chose Seattle Mariners CF Michael Saunders, simply because

A) the comments of FanGraphs ZiPS projections of the Mariners featured an especially dimwitted response about Saunders’ 2012 performance compared against his 2013 projection

and

B) I can think of no things in god’s creation worth getting upset about less than the projected 2013 line of Seattle CF Michael Saunders, who as Szymborski noted could reasonably be expected to perform within a margin of error of +/- 10% of this line, based on, one would assume, circumstances.

also finally

C) compared to his performance over 4 years of data, Saunders’ 2013 reveals a trend of general performance improvement, close but not exactly the level of the previous year but certainly ahead of, say, the year before.

For these reasons, Saunders is  a good representation of ‘regression to the mean’ because this case shows that the mean doesn’t actually mean the average across X time, but instead the average of X time with the time weighted in such a way that does give more value to the performance of the previous year. And also because if you look at things stacked against  4 years of data suddenly you’re not comparing A to B, but instead A to B, C, and D, and having more to reference is hardly a point to complain about.

Winning the Unwinnable Pt II: My Mock Draft Rules for 2013

In pt I of my mock draft series for 2013 I outlined best practices for successful mock draft planning and thinking. I’ll be participating in multiple mock drafts this off season to put these ideas to the test as I plan for various leagues. In this short post I outline the rules and goals I’ll be using in all mock drafts

  1. pick one list/strategy and stick as close to it as possible
  2. don’t draft a pitcher before round 4
  3. don’t draft a relief pitcher before round 10
  4. in a 2 catcher mock draft, don’t draft the second catcher before round 20
  5. if possible, target 2 top 15 pitchers (unless this is impossible based on rule 2)
  6. in roto mock drafts, set the goal of finishing at least 3rd in all 5×5 categories except wins and save

 

If I add or change any of these rules, I will update this post accordingly.

Follow

Get every new post delivered to your Inbox.