March 24, 2014

Fantasy Freestyle

Early-Season Strategic Decisions

by Wilson Karaman

Printer-friendly

Contact Author

The old cliché that “you can’t win a championship in April, but you can lose it” is a good one to keep in mind when approaching early season team evaluation, particularly as it relates to trading. When you react to April and May performances, you react to inherently small sample sizes, and that’s a generally dangerous thing to do. Players get lucky, other players get unlucky, some other players try to play through undisclosed injuries… it’s a minefield of variables out there if you’re looking at a few weeks’ worth of performance data.

Still, the fantasy baseball season is only so long, and waiting too long to address your team’s Achilles heel(s) can mean digging a hole too deep to claw your way back out of later on. Waiting too long to make moves can also lead to increased limitations imposed externally by the established market; a seller’s market may develop around a player or players you need, driving up the costs of acquisition. Still, while early-season roster shake-ups are a necessary and sometimes-advantageous part of the game, approaching them with caution is of critical importance.

Know what to look for
Once the season starts, I tend to gravitate back to a landmark 2008 study published by Pizza Cutter at the now-defunct Statistically Speaking*, mostly because I’ve yet to find a subsequent piece that comes close to its depth. For math nerds, it’s a tremendous study on statistical variation and stabilization. For everyone else it offers an accessible series of punch lines for spotting trends in the making. Cutter found the following baselines for hitters and pitchers regarding sample size validity:

Hitters

• K%: 150 PA
• Contact%: 150 PA
• LD%: 150 PA
• BB%: 200 PA
• GB%: 200 PA
• GB/FB: 200 PA
• FB%: 250 PA
• HR%: 300 PA
• HR/FB%: 300 PA
• BABIP: Doesn’t reach a 0.50 r-squared at 650 PA or below.
• AVG: Doesn’t reach a 0.50 r-squared at 650 PA or below.

Pitchers

• K/PA: 150 PA
• GB%: 150 PA
• LD%: 150 PA
• FB%: 200 PA
• GB/FB: 200 PA
• K/BB: 500 PA
• IF FB%: 500 PA
• BB/PA: 550 PA
• BABIP: Doesn’t reach a 0.50 r-squared at 650 PA or below.
• HR/FB: Doesn’t reach a 0.50 r-squared at 650 PA or below.

The above figures represent the turning point at which a sample size of plate appearances becomes more or less coherent enough and free of noise to be statistically meaningful. The most interesting takeaway from these findings is that all sample sizes are not created equal. Some small samples are smaller than others, insofar as certain performance indicators are more helpful in evaluating players over shorter timeframes while others are much less helpful without nearly a full season’s worth of data (at least). These benchmarks are extremely helpful to keep in mind, as it’s easy to get trapped into reductionist thinking along the lines of “Well, Player X’s BABIP is low, so it naturally stands to reason that it will likely regress positively as the sample grows.” The above numbers tell a cautionary tale, though, insisting instead that the component elements of BABIP provide a much more stable collection of data points for analysis. If you’re so inclined to attempt isolating real performance trends in the first half of the season you’ll be much better served turning to things like contact, strikeout, and walk rates for hitters and batted ball profile for pitchers.

Beware of biases of perception
Jeff Quinton’s excellent piece last week touched on some of the biases our pre-conceived notions related to past success can have on our draft strategy. Just as dangerous are the biases we harbor about the teams we thought we were drafting this year. If my draft unfolded in such a way that I grabbed a bunch of players I projected would be strong Stolen Base and Runs contributors, the natural inclination is to use that as a starting point for in-season adaptation, regardless of how things actually play out in practice. I may be eighth in steals in a 12-team league come June 1, the thinking goes, but I drafted Brett Gardner and Michael Bourn, so clearly that’s just a small-sample blip that will self-correct as the season continues. Thing is, that’s not necessarily true (as managers who owned that combination last year will vocally attest). Utilizing the above-referenced metrics can be helpful not only in spotting trends, but in so doing helping us to be more dispassionate and objective in our analytical methods. The team you field is not necessarily the team you thought you’d be fielding, and it’s of critical importance to base your adjustments on the former.

Evaluating your team is only one piece of the puzzle
I know your mama probably told you differently, but it’s important for you to recognize that you are not the center of the universe. For all the time and effort you devote to evaluating your players and figuring out your team’s strengths and weaknesses, your team does not operate in a vacuum. In Roto leagues especially, your ultimate fate hinges not on the performance of your team, but on the performance of your team relative to the other teams. This is probably the biggest reason that I tend to avoid consummating trades early in the season, barring situations of catastrophic injury and necessary production replacement. You’re not privy to the strategic intentions of your opponents at the season’s outset, nor do you likely have the time or energy to take the kind of fine-toothed comb you’ve taken to your team and run it through everyone else’s lineups. The latter issue speaks to the limitations of playing a recreational game, while the former is a much more systemic factor of structured competition.

There are a number of ways to skin the proverbial cat when it comes to building a winning team. There are balanced winners who are strong across the board (think second-, third-, and even a couple of fourth-place finishes in every category). There are weighted teams that finish first in a bunch of categories and fifth or sixth in a couple others. There are teams that punt a category entirely, yet cobble together enough first- and second-place finishes elsewhere to overcome it. The point is, you’re not party to the kind of strategic path each of your opponents is angling to take. And the earlier you look to wheel and deal the more blindly you’ll be flying in this regard, as intentions will not have become clear just yet. By moving early you’re further giving up valuable intel that will help mold and shape the development of your opponents’ strategy, as their subsequent strategic choices will now include the knowledge of your deal.

All of this is to say that if you feel the tradin’ jones in your bones early in the season, tread cautiously. While there is indeed some merit to the idea that moving early allows you to “set the market” and acquire helpful players before others have the chance to drive up the cost, the impediments you face to clear-headed decision-making are more numerous and have that much greater potential to produce negative, unintended consequences.

*The original studies no longer appear to be available at their original homes online. Summaries courtesy of the Hardball Times.

Wilson Karaman is an author of Baseball Prospectus. Follow @vocaljavelins
Click here to see Wilson's other articles. You can contact Wilson by clicking here