Hacker News new | past | comments | ask | show | jobs | submit login
You Bet [pdf] (oaktreecapital.com)
97 points by RickJWagner 33 days ago | hide | past | web | favorite | 8 comments

Well written piece on expected value. Something he doesn't get into is the shape of the skill factor

There's quotes out there about how poker is 9x% luck, & that last stretch of skill is a game of chess. What they're talking about is skill cap. By this we mean that poker is such a hard game that not even the best players are playing it perfectly. See a recent AI Pluribus for some articles on how AI's pushing this game

Essentially it's about having a 3d graph with the three labels:

X: Player1's skill Y: Player2's skill Z: Player1's chance to beat Player2

Issue: skill may vary with different strategies, so that you end up in rock-paper-scissors scenarios

Then there's a function over how much effort is required for a player to increase their skill. This is how we get games described as "easy to learn, hard to master" they're games with high skill caps but which a player should be able to quickly have more than single digit odds against an experienced player

An interesting thing with this 3d graph is that you can make transforms on it while treating the game itself as a blackbox. For example, in StarCraft players will often play Bo3 in tournaments. Plenty of commentators will go on about how this let's people bring a variety of strategies etc etc, but all it really does is avoid letting a single map's balance be decisive & exaggerates the advantage of skill. If you have a 70% winrate against someone, in a Bo3 you have an 80% winrate against them

This ends up creating interesting decision making scenarios when one can choose strategies which shift the role of luck in a game. It may make sense to play a strategy where even with perfect execution you only have a 40% winrate if the skillcap for that strategy is low enough & your opponent is better. This sort of thing makes it that there can be a bit of a divide where players are stuck at a 40% winrate against players who are better than them until they've become skilled enough to rationally play the higher skillcap strategy

If you like his memos I recommend the "bubble.com" memo from 2000:


Timely written, at the very peak of the mania (though Marks correctly admits that such precise timing is accidental, knowing well the difference between luck and skill :) ). You can find all Howard Marks memos on Oaktree's website.

The PDF has been corrupted since going up a few hours ago. Here's an archived copy:


If you enjoy this you might enjoy Thinking in Bets by Annie Duke.

It's got a lot of the classic "here's how humans are irrational" stuff that's in other similar books of it's genre, but I found it full of very practical tips as well.

One of his main points is that value is based on fundamentals and good companies can be bad investments if there is not additional value to be uncovered. This is sound advice but I think the definition of sound fundamentals is changing. The lens through which we value companies probably always needs to evolve as the global market evolves and I think this makes identifying “good bets” more challenging.

There is one other factor. Reputation. Imagine being at a poker table where the dealer preferentially deals aces to one player. That's exactly what happens in venture capital, and it's the only asset class where the asset chooses the investor and it's why top funds tend to stay as top funds.

> In other words, we tend to respect people who think like we do. Did you ever hear someone say, "I think Bob's a genius, and he thinks my views are all wrong"? That's something few people would ever say. No, we tend to think highly of people whose opinions mirror ours.

This doesn't seem like a very good explanation, because we can and in many cases do change our views in the face of arguments from people we respect as geniuses. It doesn't have to be a coincidence or evidence of bias if the people you respect the most in some area of work also share your views about that area, because if you respected someone else more then you would change your views to those views.

This is quite interesting, here's the man himself on Bloomberg talking about this memo few days ago:


Applications are open for YC Summer 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact