The Laws of Trading

Everything is a trade – even if we aren't aware of it, we are almost always trading time, money, or attention whenever we make a decision.

Adverse selection - you will almost never be happy with a trade! If you get a good deal, you will have wished you had bought more. If you get a bad deal, you will think that you have overpaid. The counter-party almost always has more information than you do.

Types of risk – this was a new section for me, it basically outlines 7 different ways to look at risk. An important but under-mentioned type of risk is "liquidity risk" (which I have been seeing now). The idea here is that when you want to go an sell an equity or derivative... if the market is very much correlated with it, everyone will be trying to sell it at the same time. Your liquidity might dry up!

Generally speaking: you want to make the trade that minimizes liquidity risk.

There's also an interesting framework here around what your beliefs are and how you might want to trade. If you really believe in the stock out-performing, you might want to buy the stock itself. If you believe in the sector more than the stock and think the stock will be a bellwether, buy the sector ETF. If you think the economy will outperform, just by SPY. Each of these pools have greater and greater liquidity.

Buying options requires less capital, but almost always makes you less money.

Find the edge – generally the edge is something that 1) you recognize and nobody else does 2) you can actually act upon 3) is simple. The more words it takes to explain your edge, the worse it probably is.

Stories are useful – instead of trying to endlessly model the numbers, it's worth coming up with a simple yet powerful story of why you are making money. It's always possible to find some sort of random noise that indicates a strategy will make money (Stochastics vs Statistics)... but that rarely proves that this will be a good trading strategy.

G models and P models - models come in two forms: generative and phenomenological. Generative models are the “gears” approach to explanation, they predict the underlying mechanisms of why something happens. P models are only based upon empirical results or particular outcomes.

We’d generally prefer G models, but obviously you can’t get those in most cases. One interesting area discussed that I’d like to work on is forming G models of other people. I find myself more often afraid to joke with someone I don’t know very well, because I’m not sure they’ll take it the right way. With more practice g modeling, I can understand more of the motivations behind what they do. That said, we need to be careful-problems can come from incorrectly assessing why someone does what they do. The stories we tell ourselves are usually just that: stories.

Revealing edge – whenever someone proposes a trade, you have to ask yourself: "what is their edge?" And if their edge is really so good, why would they trade with me?

Aligning incentives – the section on aligning incentives is really interesting, in part because it discusses the fact that any sort of structure requires incentives to be aligned to solve the principal-agent problem. In finance, there's a problem where the fund managers are generally incentivized to take big, risky, bets, and then either close down the fund or raise another one. When fund managers are also investing the majority of their capital alongside the fund, that's when the incentives align.