Ergodicity - Luca Dellanna

[Saluki]

This book has been mentioned a lot but not reviewed, so I'll take a shot.  Ergodicity is the concept that your expected returns will be close to the average of the population. So if the average adult male in the US is 5'9", then you are probably going to be close to that height if you are an adult male in the US.  If something is non-ergodic, then you might find people who are 20' tall, or 4" tall, even if the average is 5'9". This is the problem that Howard Marks talks about when he says that a person who is six feet tall can drown in a river that is 2 feet deep, on average. 

 

What's really good about this book is that it's one of the few books that is good at explaining mathematical concepts without using a lot of math.  Nassim Taleb is a terrible writer, but a good explainer when it comes to this.  Peter Berstein, Benoit Mandelbrot and Michael Lewis are all good at explaining math with using formulas and math. 

 

So in thinking about non-ergodic things you come across, you must keep in mind that the average won't apply to you, and that in some things, like Russian Roulette, future gains can't make up for past losses, because you are out of the game. It's a great example because if you look at the payouts for something like the Kelly Criterion, there are sub-optimal bets, then a full Kelly Bet, then overbetting, which gives sub optimal returns just like underbetting.  But underbetting won't wipe you out.  Overbetting increases your odds of wiping out, then the odds don't matter because you can't make it up with future bets.  So if you think about yourself in alternate universes, and you want a good outcome in as many as possiblle, then you probably won't hold Amazon since the IPO and get rich.  Because in some universes, that would mean you held onto Netscape and went to zero.  So trimming winners and rebalancing, is not optimal in some universes, but across all univeses, it increases your odds of a getting a return that matches the average: erodotic.  It's okay to have suboptimal bets if it reduces the catastrophic risk. 

 

He uses the example of his cousin who is trying to win 10 skiing races.  If he goes as fast and takes as many risks as he can, he has a 20% chance of winning each race and a 20% chance of breaking his leg.  But if he breaks his leg on the first race, his chance of winning the next races are zero. So it's better to go with a 10% chance of winning, and 1% chance of wiping out because it increases the total number of races you can win.  

 

It's a short book but very readable and with good examples to explain the concepts without getting caught up in math.  I would recommend it for this crowd.  

[whatstheofficerproblem]

Thanks for the detailed explanation Saluki. You are good at explaining things it seems, I suddenly feel like reading this book, going to pick it up.

[jfan]

I've purchased both this one and the one titled winning long-term games. Both are excellent reads that gave me some ah-ha moments.

 

Survival is more important than performance.

Play the games that you can survive and not the one's that others were successful in - reproducibility matters.

Sub-optimization and redundancy are helpful to survival.

 

This was really useful to think about portfolio construction, long-term investing and dispels some common notions that amateur dabblers like myself held.

For example (as it applies to the non-professional)

1) Not being fully invested is not a terrible thing. It gives optionality, allows one to have a clear mind when situational crisis occurs, and achieving a "satisfactory" outcome can still get you to the end-goal without being the best (eg IRR).

2) small position sizes should not be shamed. Lots of unexpected stuff happens both positive and negative. For very long-term holding periods, both are increasing likely over time.

3) Also cutting your flowers should not be shamed. Idiosyncratic things can even happen to really great businesses especially as your time horizon lengthens. The goal is having your portfolio survive for an investment lifetime (50 years), not just 5 - 10 years. This framing also helps with patience.

 

This series of books coupled with ideas from deep survival are super useful (eg survivor mentality).