Notes from Capital Allocators: Annie Duke

Linkhttp://capitalallocatorspodcast.com/2018/02/05/annied/

About Annie: Professional poker player and author of Thinking in Bets


All decisions are a bet

When you choose x you forgo y. The decision is a bet that x is a better outcome than not x.

Beliefs are formed then confirmed

Dan Gilbert’s known for happiness research but his 90s research which is lesser-known was focused on ‘belief formation’. We are hardwired to not vet beliefs since beliefs are typically perceptual. Hallucinations and mirages are rare. However, abstract beliefs that emerge from our social interactions, language, symbolism are incorporated via this same mental machinery which was really designed to assimilate perceptual beliefs.

Gilbert showed that by default we accept the belief is true BEFORE we vet it.

Research shows:

    • We often fail to later vet the belief
    • If we do vet it, we are biased
      • Kahneman’s idea of “motivated reasoning”: our beliefs drive how we vet the belief
      • Confirmation bias
      • Blindspot bias
      • Smart people are often more extreme in their biases because they rationalize with a greater repertoire
        (This was tested by first evaluating subjects’ statistical prowess then comparing how they handicap a neutral versus emotionally-charged outcome)

How do we improve?

  • Frame decisions as bets. 

    1. Assigns probabilities to outcomes
    2. Define what we explicitly are evaluating
    3. Invites others into the truth-seeking process which is also good for social reasons
      • When acting very certain we can suppress or intimidate other’s views
      • Avoid the pitfall of confusing certainty with accuracy
      • Makes the communicator more believable
      • Avoids biasing others before they start the vetting process

  • Define winning as being more accurate inoculating ourselves against self-serving bias.

    • Use Mertonian norms which comprising the ethos of science (acronym: CUDO)
      • Communism: Standardize how data is presented so members of the community cannot present the group biased picture
      • Universalism: Ideas have objective truth regardless of the messenger; “Don’t shoot the message”
      • Disinterested: Do not infect the group with your beliefs
      • Objective Skepticism: seek counterfactuals and dissent
        • In groups, use ‘red’ and ‘blue’ teams
          Red team’s function is to rebut or dissent. This instantiates a role in which being a team player is actually to challenge.

  • Be aware of our tendency to “temporally discount”.

    • A dollar today appears worth way more than in a year
    • Tonight’s wine is tomorrow’s hangover (Seinfeld’s “Day Jerry” vs “Night Jerry”)

If we associate better-calibrated beliefs with better outcomes and a happier life we should strive to be honest with ourselves while reasoning even if it sacrifices our ego/fun in the moment.

Risk Management

  • Why we fail to apply the principles of Kelly betting:

    1. Garbage In/Out: Poorly calibrating the edge and/or variance
    2. Our ability to apply rational System 1 rules are compromised when we are in an emotionally charged state (“limbic system firing”), which is when the rules matter most. “Stacking Irrationality”.

  • The merit of risk limits irrespective of risk/reward or expected value:

While irrational, they are less damaging than allowing yourself to continue betting when your “emotionally unfit”. A bias which gives us the chance to play again tomorrow when we are not emotionally unhinged is adaptive for the long-run. (Me: Reminiscent of ergodicity discussions about maximizing compounded expectation)

Tells and body language

  • Good to follow: Joe Navarro. A FBI operative specializing on body language

  • How poker players read opponents:

    1. If never faced them before, start with base rates
    2. As you learn how they bet, Bayesian update base rates
    3. Merge tells with updated probabilities

  • Signs of being relaxed vs discomfort:
    • embodied by distance from table
    • self-soothing behavior

Leave a Reply