Summary

We tend to overestimate the strength of signals in data, are too confident in our own skills, and are thus overconfident in our projections. The most important piece of prediction is weighing information correctly using Bayes’ theorem. Humans are naturally results-, not process-oriented. Approach pundit predictions with healthy skepticism. The stock market is a negative-sum game - so unless you are insider trading you are better off investing in an index fund in the long term.

“It is hard to tell how many investors beat the stock market over the long run because the data is very noisy, but we know that most cannot relative to their level of risk since trading produces no net excess return and entails transaction costs - so unless you’re insider trading you’re better off investing in an index fund.”

Key Takeaways

• Rely on many small pieces of data, not heavily in one big data point (fox vs hedgehog)
• Use Bayes’ theorem to adjust expectations
• Find fields where the “water level” is low to excel in
• Better off investing in an index fund (efficient market hypothesis)
• When the facts change, change your mind

Notes

• Fox vs hedgehog
  • Fox: relies of many small pieces of data, less sure, willing to admit wrong
  • Hedgehog: believes heavily in one big data point, overly confident
• People misinterpret probabilities (Thinking, Fast and Slow)
• Political, economics pundits perform little better than chance
• More data can lead to more bias
  • eg. Predicting red vs blue seat gain vs an individual race (can create narrative more easily)
• “When the facts change, I change my mind”
• Most important part of predicting is weighing information correctly
• Self-fulfilling and self-cancelling predictions
• Bayes’ theorem:
• Model / heuristics: a simplified way of looking at a system
  • Watch out for bugs, assumptions
  • Developers’ biases show through
• Find fields where the “water level” is low
  • Tip of iceberg is profit among mountain of work
  • Eg. Stock market companies
• People are results, not process oriented
  • Don’t factor in luck, probability
  • Eg. statistically, a pro poker player can make $270k to -$30k / year
• Averaging multiple models almost always works better than relying on one
• Efficient market hypothesis: it can become impossible to make a profit in the stock market if priced perfectly reflect value (therefore truly unpredictable)
  • Market-wide PE ratios is promising
  • S&P: 1900-’70: ~13.5, ’71-’99: ~15, ’00-’16: ~25
• The unknown unknown: what you don’t know you don’t know
  • Know what you don’t know!