Mike Elias, The Baltimore Orioles, and Astroball
Basic principles and theories Mike Elias (GM and Executive VP of the Baltimore Orioles) will have worked under and learned from as laid out in Astroball by Ben Reiter. The idea is/was to try to see the game from the point of view of the people making team wide decisions about who is drafted, traded for, and ultimately does or does not get to play on a club that was rebuilding. Elias worked for the Astros immediately before coming to Baltimore. For the Cardinals before them. All entries are direct quotes from the book.
Related people: Sig Mejdal.
Goals
- The only goal was to reach the top as quickly as possible.
Motivators
- Longed for a system that better allowed him to settle disagreements between his scouts and his analysts...in a way that didn't rely on custom, or habit, or intuition.
- [To] avoid ever again having to stand in front of a room at a crucial moment in which scouts were still pulling one way, and the analysts another entirely.
- How do you get things on the upswing as soon as possible so you can get to the point where you're consistently competitive.
Achieving Goal
- Every decision made would be based on the probability that it would prove helpful in the long term.
- Identify players who are unsatisfied with their lot, who possess an uncommon drive and ability to improve.
- [The] days of spending big for losing players should be over.
- Unlike other rebuilding teams [they] would not waste money on a free agent or hold on to old and expensive fan favorites in an effort to keep up appearances, particularly when those players might instead be immediately converted into future assets.
- They wouldn't spend money until they had it.
Decision Making Process
- An overreliance on [Big Data] as a decision-making engine could lead to middling and one-dimensional baseball teams.
- [He] promised he would reintegrate information from another source into his decision-making process...scouts.
- STOUT -- Half stats, half scouts.
- Make humans responsible for triaging -- and sometimes rejecting -- the results.
- Inputs included not only a player's statistics but also [information from scouts] about health, family history, pitching mechanics, shape of swing, personality [and then] ran regressions against all the information against a database stretching back to 1997.
- The idea was to scout the scouts in order to determine which of their judgments had real predictive value and which were the product of cognitive biases -- and then to properly weight them along with performance data. [Compared to actual data from similar players of previous years.]
- [evaluation system] boiled down every piece of information [on every prospect and player] into a single language, and even a single number.
- The end result was expressed as a numerical projection that roughly translated into how many runs a player could be expected to produce measured against the salary he was likely to command.
- "We're operating with information and techniques that are more or less out there in the baseball community. More than half the teams use most of the information that we use. [The] competitive advantage for us is having the discipline and the conviction in our information to stick with it even when it feels really wrong." -- Mike Elias.
Scouting Candidate Keys
- GROWTH MINDSET: ...an intangible -- the illusive but discoverable qualities of persistence and adaptability.
- It was up to the player to have the willingness to act upon that information [provided by coaches/club] in order to grow.
- As players...grew -- or didn't -- so did their projections.
- Coaches spent every working day with the players in question and filed reports back [home] after every game.
- ...models [were] updated constantly.
- The [Analytics Office's] algorithms suggested when [a player] was ready to be promoted to a higher level of play, or to be cut.
- ...system designed to guide [the player] toward the changes he would need to make in order to thrive.
- People who focus on process over outcomes, and who have a growth mindset, don't allow one setback to derail them.
- Every month mattered... a high school senior who received a virtually identical evaluation as another who was...10 months older...had a surprisingly elevated chance of future success.
- Carlos Correa: [Information] came from scouting him and meeting him. Glowing reports from good scouts had predictive value, too. [The] system concluded that Correa was the draft's top talent...especially because he was so young.
Drafting
- Each club was assigned a pool of money to compensate its picks, based on the sum of recommended dollars for each pick in the draft's first ten rounds. If you could pay an early draftee less than the collectively bargained bonus his particular draft slot called for, you could spend more on players picked later.
- The recommended slot bonus for the one-one in 2012 was $7.2 million.
- Draft evaluation process: Area Scouts > Eyewitnesses > Head of Player draft > Analyses of swing/pitch by organizations coaches > Statistical Projections.
- Each scout had focused on delivering information that would benefit the entire organization.
- ...one-one meeting...designed, in part, to get all the front office excited about whatever player it was selecting...
- [player] had received the draft's best grades from at least three-quarters of [the] executives. [The] system stripped those grades of biases and incorporated other predictive information like age and mechanics [to prove the top draftee]
- Years of scouting reports, funneled into and regressed [by the] algorithms suggested [he] was the best player. So did the less quantifiable facts. [Work ethic, etc.]
- ...they'd made the big [decision], and virtually everyone either agreed with it or understood it.
- ...Metrics had always prized: a high on-base percentage … an excellent OPS … a disdain for striking out... strong student suggested a growth mindset; and the right kind of confident: not performatively cocky.
- Metrics had always flagged as significantly negatively predictive: strikeouts.
Trading
- No one was safe if he could be leveraged into a more promising return.
- Trades made out of emotion, even desperation, would have likely been bad ones. Most deadline trades are ill-advised.
- Deadline trades required the sacrificing of years of cheap production for, sometimes, a couple of months of starts from an expensive and aging pitcher.
- Re: Building sustainable winners -- Talented prospects who could continually replenish the roster at sub-market wages, were the key to the effort.
The Clubhouse
- ...the clubhouse was missing a player who...had experienced virtually everything a player could in professional baseball.
- It was no longer enough to witheringly conclude that good teams always had good chemistry, and bad teams had bad chemistry, but that bad teams were never bad despite good chemistry.
- The study calculated a 20 percent variation in team wins from what might be expected if victories were simply a result of the sum of a club's player's performances, and further asserted that 44 percent of that difference could be accounted for by chemistry. The David Ross Effect.
- Demographic differences, called FAULT LINES, could affect performance. Leading to in-groups and out-groups.
- FAULT LINES: included position groups (pitchers/hitters, relivers/starters, etc.), status (well-paid/under-paid, etc.), nationality, culture, language.
- Fault Lines shift the focus from winning to "task-irrelevant cues" (competition and distrust between isolated subgroups) and restricted communication.
- Fault Lines could account for three extra wins or three extra losses in a season.
- Teams that performed best were not demographically similar but had players who could crosscut between a mix of subgroups.
- A clubhouse with good chemistry could persevere through periods of failure better than one without.
- Deactivated Fault Lines could prevent poor results from snowballing.
Advice
- "The feeling that we are smart is our enemy."
- Just because it feels right doesn't mean it IS right.
- "If there is a change taking place, it's not likely going to feel right. If it felt right, it would have been done a long time ago."
- On combining scouting and data analysis: Don't frame it as "either or", frame it as an "and" question.
- Every draft class is a portfolio...You've got to mix up some big bets with some fliers.
- Shy away from taking high school pitchers, with their proven risk of injury and failure, in the first round.
- One of two types of owners: A rich guy who viewed [the team] as a plaything, a guy who wanted something to impress his friends, and a rich guy who wanted to transform them into a sustainably profitable business.
- ...some organizations have...tried to maintain a .500-level team as they prepare to be good in the future. ...it takes ten years.
- "If you say the world is round and I say it's flat, we don't agree to disagree. There is always a right answer."
- [This would not be the right strategy] for someone who had a great farm system, younger up-and-coming players at the big-league level[.]
- "When you're in 2017, you don't really care that much about whether we lost...in 2012. You care about how close we are to winning a championship in 2017."
- One of the problems with regression analysis was that it had trouble identifying outliers.
- High School pitchers were terrifying for any club with a pick in the single digits. "...the list of those picked high is littered with injuries and disappointments."
- ...the importance of collegiate park factors and strength of schedule were no longer a secret.
- In a short series, few...advantages were significant. "The decisions you make during the postseason are not going to impact the outcome too much."
- Data could help guide best practices, but it was unwise to confuse those with perfect practices.
- "All models are wrong. But some are useful."
- The best clubs are not constructed under the assumption that they will finish the season with the same 25 players with which they began it. The best teams have built-in backup plans for when things inevitably go wrong.
- “Only a fool would ignore the potential value of the shit he didn't know.”
- "Just because you can't quantify it doesn't mean it doesn't exist."
Supporting Evidence
- Not even 10 percent of amateurs who are drafted will ever [make it to] a major league field...even for a single inning.
- If a club's draft produces nine players who appear in the majors for even a single game, it ranks in the 95th percentile.
- 95th percentile for the number of everyday players...is four.
- 95th percentile of above-average players (WAR +6) is three.
- The Cardinals fielded a 25-man roster that was 20 percent composed of players they had picked in a single draft. (2013 World Series, game 3)
- On Moneyball: Lessons had been absorbed by most front offices -- On-base percentage was better than batting average. Wins were a poor indicator of a pitcher's value because, unlike strikeouts...they were so dependent on the play of the pitcher's teammates.
- Kahneman and Tversky introduce cognitive biases. [Leading to irrational choices]. And human reliance on heuristics. [Mental Shortcuts].
Conclusions (my own):
At some point a book will likely be written about the Orioles widely acknowledged rapid turn around from a sub .500 losing team in 2017 to a team that won 101 games in 2023 and clinched a playoff spot. When that happens, it is entirely likely that the teams of MLB that are not already doing so will adopt as much of the recruiting strategy as they can see from Baltimore, thereby once more leveling the playing field amongst all teams just as they did with Moneyball and the Athletics earlier in the century. To the point where it was no longer an effective strategy for seeking advantage over other teams because all teams were doing it. Thus negating the value. Similarly, the “Astroball method” will cease to produce as strong results as it currently does or did.
Even now, about half the league has caught on or already been brought on board as staff who know the method disperse throughout the league. At the point which the majority of teams can data mine and interpret data successfully, they will be able to implement the methods on their own without needing the insider. This will happen before the potential book comes out. At that point, competition will settle and once again a majority of teams will be more or less (allowing for individual performance) equally competitive.
Until then, Elias and his staff may well continue to be at the forefront of competitive team management and produce a team which is very capable of a long term string of successes. And that success is not limited just to acquisitions. It effects every moment of play on the field and usage of the players.
By and large, the vocal majority of the fan base does not understand the mindset or methods on display in Astroball or in the Orioles organization. Nor are they interested in trying to work it out. They continue to bemoan every trade and acquisition, they fail to understand when and how prospects advance through the minor league system, and lack any knowledge of the overall new reality of modern baseball, remaining mired in an almost religious obsession over old statistics that, if they ever did, no longer have any value in determining a player’s worth when compared to the newer stats available. The majority have yet to absorb even the lessons of Moneyball, let alone the Astroball method, and certainly not the Elias/Mejdal iteration upon it. Something that this document does not purport to see into. Elias plays very close to his chest and rightly so. This only shows the base he started from, not any modifications to the method he has made. Which will turn out to be considerable.
To understand how the Baltimore Orioles work and how decisions about the team are made and why they make the moves they make, even down to play on the field, you have to begin at the lessons of Astroball and work from there.