Post
A friend (thoughtful, academic type, reads the newspaper) recently asked me why I preferred to talk about topics where I was on the wrong side of expert consensus (like BTC, Internet Finance). His point (probably true) is that while experts can often be wrong, you are likely to be right >50% of the time by starting with expert views and then adjusting from there. I told him that markets people make the best returns when they are contrarian and being on the wrong side of expert consensus is usually correlated with low consensus / high payoff opinions. You want to max Edge i.e max the sum of log2(your probability/market probability). It's pretty hard to max edge by finding 80% probability opinions that you think are 85%. You want to hunt in underpriced markets / low consensus opinion space. The interesting thing is that most people are not attempting to max Edge at all but min Brier i.e. (your probability - outcome)^2. So if you say there is an 80% that it rains tomorrow and it rains, your Brier Score is (1-0.8)^2 = 0.04 and had you said "No Rain" it would have been (0-0.8)^2 = 0.64. Lower is better. If your relationship to predictions is basically [saying your opinion to friends / going on TV / writing articles] then you just want to be right pretty often especially when you say you are confident. That directly translates to min Brier. That makes a ton of sense! Who wants to be the guy who is edge maxxing by betting on 5%s he thinks are 20%s but is basically wrong about 80% of everything he ever says? The guy who wants to make money, that's who. Min Brier is a disaster for returns. Portfolio growth is directly proportional to edge and orthogonal to brier. here is a 1000 sim monte carlo comparing a min brier guy who is right 85% of the time (consensus 80%) and an edge maxxor who is right 20% of the time (consensus 5%). min brier: mean final portfolio: $1.07m mean return: +6.6% win rate: 85% max drawdown: –3.6% prob of loss: 10% edge maxxor: mean final portfolio: $19.9m mean return: +1,889% win rate: 20% max drawdown: –14.4% prob of loss: 0%
@sacha I’ve met two billionaires in my life and the common pattern i found is that they were both extreme risk takers. obv small sample size but feels directionally right. iq alone won’t get you there. these people had made enough to retire and they’d still risk it all again and again
