Sox Machine Book Club: ‘The Inside Game,’ Chapters 10-13

The Inside Game by Keith Law

Let’s wrap up the inaugural edition of the Sox Machine Book Club with the final four chapters of Keith Law’s “The Inside Game.”

(Bookshop.org link; Sox Machine receives a portion of the proceeds.)

The first nine chapters were divvied up over three posts:

And if you missed it, I discussed parts of the next three sections with Law on the Sox Machine Podcast.

Chapter 10 — Pete Rose’s Lionel Hutz Defense: The Principal-Agent Problem and How Misaligned Incentives Shape Bad Baseball Decisions

Picking up on the previous chapter’s theme of moral hazard and how incentives shape actors, here comes the principal-agent problem involving literal agents. In this case, he uses the motivation of Ozzie Albies’ laughably low pre-arb contract extension, in which his small-fish agent wanted to lock in the commission from the best contract he could get before a bigger agency stole his client.

It’s a bit more specific from moral hazard, because in this case both parties have the same goal, which is a long-term contract extension. It’s just that the agent has time-based complications that result in him not getting the best possible extension for his client. Law also uses real estate agent — who should want to negotiate a lower price for their buyer, but would earn a greater commission if the sale price were higher.

Law also provides an example of a principle-agent problem that doesn’t involve an agent of any kind, which is Pete Rose’s defense of his gambling patterns. Rose said he didn’t bet on his own team to lose when managing the Reds, but the lack of action could still shape the way he’d behave managing games. He might pull out all the stops to salvage a game he had a 40 percent chance of winning because he had money on it, while easing up on a game with a win probability of 60 percent because he wanted to rest his players for more personally vital situations. He and the Reds technically wanted to win both games, but Rose would’ve been incentivized differently depending on whether he had a personal, private reward for doing so.

Chapter 11 — Throwing Good Money After Bad: The Sunk Cost Fallacy and Why Teams Don’t “Eat” Money

The sunk cost fallacy is well known to any baseball fan who has watched a team run out an expensive player well after his productive days are behind him. There’s no economic justification for playing a more expensive player when a rookie offers more upside, but whether the team doesn’t want to admit it made a mistake or appear wasteful, it keeps devoting resources to a player who won’t make good use of them.

Chapter 12 — The Happy Fun Ball: Optimism Bias and the Problem of Seeing What We Want to See

Law writes that optimism biases “are a class of smaller, related cognitive errors that all resolve to us thinking what we want to think.” He points to overconfidence bias and the planning fallacy, both of which don’t devote enough consideration of the way forecasts can go wrong.

At this particular juncture, Law uses the out-of-whack offensive numbers that resulted from the juiced baseball, and how that runs the risk of leading teams to think the 10-homer guy who hit 23 homers is a 23-homer guy in perpetuity.

Chapter 13 — Good Decisions™: Baseball Executives Talk About Their Thought Processes Behind Smart Trades and Signings

After spending the first dozen chapters of the book pointing at all the ways teams struggle to make good decisions, Law ends on a more charitable note by showing how teams can navigate such heuristic minefields and come out ahead.

Or at least he does with Jose Bautista, whose swing change resulted in a season that didn’t resemble the rest of his track record as he approached free agency. The Blue Jays signed him to a five-year deal, giving it all the markings of a recency-driven deal. Then-Toronto GM Alex Anthopoulos paints the decision as a much more tortured process, involving weighing the contact quality data and open market value. Ultimately, keeping Bautista looked like the best road, even if the price wasn’t one he was excited about. It worked out.

The heuristics in play aren’t spelled out as clearly when talking about the Tigers signing Magglio Ordonez in 2005, the Red Sox trading away Nomar Garciaparra and drafting Dustin Pedroia in 2004, or the Cardinals declining to sign Albert Pujols after the 2011 season. There’s groupthink, base-rate neglect, status quo bias and so forth in play, but the decreasing references force readers to figure out the specifics for themselves.

Take a second to support Sox Machine on Patreon
6 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments

Chapter 10: This one was self-explanatory but I liked the in-depth breakdown of exactly WHY it wasn’t OK for Rose to bet on his team even if it was only to win. I see this is in my own job as well as multiple other professions where your income is based on what you produce the business financially. It also makes trusting certain professionals more difficult. The realtor was another great example.

Chapter 11: Seems like every team has dealt with this issue at some point or another. Too painful for me to go back into White Sox history to find examples so I’ll move along.

Chapter 12: Can anyone describe the major difference between optimism bias and recency bias? They felt very similar at least with the particular examples Law used with the juiced ball. I suppose recency bias can also cover negative outcomes where optimism is looking only for confirmation of what you want to find.

Chapter 13: Breath of fresh air to finally get some good decision making although I agree that how they tied in with all the various biases wasn’t exactly laid out cleanly. Or maybe my brain just was ready to be done with this particular book.

With that said, I thought this was a good pick for a group read. Topic density wasn’t too bad and it read fairly quickly. My advanced metric knowledge is fairly limited overall but I had no problems with the figures used in this book. I particularly like when an author notes what “average” is supposed to be when using any sort of statistic. Gives me a baseline into how good or how bad any particular example is.

My biggest complaint here if we are talking heuristics, is that my memory really needs a jump start sometimes to really lock a particular topic in. And I could have done this myself if I was more motivated but some sort of summary table for reference would have been very helpful. Every bias outlined with a basic definition is all that would have been needed in the index section. The concepts would be easier for me to apply to daily existence if I had a way to reference them quickly.

Either way, these sorts of books do get me to examine choices and outcomes with a bit more brain power than I would otherwise.

Another book I would recommend for anyone interested in how our brains are wired for communication would be “crucial conversations.” Really interesting stuff although a lot more academic than Law’s work here. Geared towards how we actually communicate with one another and how to mentally slow things down when dealing with a difficult conversation. Still a relatively quick read which is ideal for denser subjects when you aren’t familiar with the material.

Shingos Cheeseburgers

Chapter 10 — Pete Rose’s Miguel Sanchez Defense: The Principal-Agent Problem and How Misaligned Incentives Shape Bad Baseball Decisions
Keith dropped the breadcrumbs to it but didn’t quite come out and say it: Lots of players get to their first walk year right around the peak of the aging curve. It would make sense that out of a huge sample size guys have their best years in walk years since for most players their first (and potentially only) walk year is at their statistical peak. My guess is that the frequent pointing out of players having their best years in walk years in a forced continuation of the ‘greedy players’ narrative. Not that Keith is doing that here.
A good oversimplification of the Pete Rose anecdote is that inaction does not imply indecision.

Chapter 11 — Throwing Good Money After Bad: The Sunk Cost Fallacy and Why Teams Don’t “Eat” Money
This one hits close to home for the Sox in baseball context. I think we’re all painfully aware of it even if we didn’t have a proper name for it for the last 15 years.
Not to argue with a well-respected academic paper but I had a hard time wrapping my head around the interpersonal sunk cost effect, particularly the basketball game tickets example. In this theoretical there is a discernible difference in the outcome of the decision: attending (and presumably enjoying the game or not. It’s a risk reward calculation, not a biased decision. Furthermore, there are other considerations to be made such as reputational risk: if I received these tickets from my client will they not do business with me if I waste their $200? Will my friend not give me tickets in the future if I elect to not go tonight? I feel like there were too many uncontrolled variables to clearly assign causation to the bias effect.

Chapter 12 — The Happy Fun Ball: Optimism Bias and the Problem of Seeing What We Want to See
I’m pretty sure that Statcast breaking HR was off of Reynaldo.
The juiced ball influence wasn’t quite uniform, we know if favored guys with certain swing paths over others, but I would think that relativities in projections under uncertain conditions remain the same. If Jose Abreu is projected to be the sixth best 1B in the AL in 2020 under a juiced ball scenario what’s the likelihood he’s ranked significantly differently in a normal ball scenario? I would think somewhat low. That relativity is useful in terms of targeting players but obviously flops when trying to price contracts. It’s not dissimilar to trying to project a player who has made a material change to their approach, the system has fundamentally changed and you’ll have to find the closest analogs.

Chapter 13 — Good Decisions™: Baseball Executives Talk About Their Thought Processes Behind Smart Trades and Signings
I was glad to see the book end on a more positive note although I’ll point out Dombrowski was also the GM for the Expos too. What kind of bias made Keith forget them here?!

It’s mentioned here or there but never really neatly packaged into an overarching thesis: the analytical influence on baseball should be limiting, if not eliminating bias, from most decisions. He did mention in the last chapter the non-baseball playing derived ‘value’ of Albert Pujols for the Cardinals but in the end, they decided that unquantifiable value wasn’t enough. I suppose the implicit message of the book is that even though we have the ability to overcome biases they are still present, inert until they are exposed without the victim even knowing. I can see why an influx of non-baseball folks into organizations has coincided with the diminishment of biased decisions (although those non-baseball folks have their own set of issues).

Overall not the best baseball analytics-adjacent book I’ve read but a nice introduction for me to behavioral sciences within a familiar context. Nothing revolutionary, but an easy, enjoyable read. As I mentioned in a previous discussion I wasn’t crazy about Keith’s tone but that eased up a bit in the later chapters. I’d recommend this to someone who isn’t looking for a stats-heavy but progressive book about modern baseball.

I’d give it Six out of Ten Middle-Middle Reynaldo Lopez Fastballs

This was fun – thanks to the other readers/commenters. What’s next?

Shingos Cheeseburgers

He gets to that idea to some extent near the end – the vast amounts of data teams are leveraging is sure to produce false correlations. …Fooled by Randomness if you will. So some level of human intervention is necessary to give statistical conclusions the ‘sniff test’ but any level human of intervention will introduce bias.