The Laws of Trading

Motivation

1. You might say "to make money obviously" but there are tons of other reasons, some good but mostly bad:
• insurance
• responding to base emotions like greed and fear
• bored!
• to get in the zone
• big score
• being right (this one is a pitfall for me)
2. Why does motivation matter?
• need to be aware of your motivations, even if they arent "bad"
• you need to make peace with your motivations or change them
• make yourself small
3. In the end, it comes down to knowing yourself
• tendencies, flaws, blinds spots, strengths, motivations
• do not rely on willpower
• precommit when you can so that you stick to it

Adverse Selection

1. Information and information asymetry
• if you are trading with someone, you are subject to adverse selection
• if you are right, you traded too little. if you are wrong, you traded too much
• use Bayesian reasoning to update your models
• a market is really just a distributed aggregator and integrator of information & Bayesian reasoning
2. Why Trade at all?
• different participants are better at different things (fundamentals, liquidity, activists)
• different time horizions
• you can integrate all this info to find out what things are worth probabillistically

Risk

1. What is Risk? Most important to realize looking at one measure of risk is not to only look at it one way
2. Market risk, liqudity risk, operational risk, technological risk, political risk,
3. Risk Aversion and catastrophic risk
• Losing 100%, not only worse than making 100%, but infinitely worse, you are out of the game

Liquidity

1. Once you have found a + EV trade, there are tons of ways to express it. So you like a stock:
• buy shares
• buy clls
• sell puts
• buy sector etf, etc etc
2. Depending on EV and downside, can use multiple expression to tailor risk to where you want it

Edge

1. Marginal Traders and why only they Matter
• Bids and offers are given by marginal trader
• you have to be better than this skill level AND
• The marginal trader in a mature modern financial market is very skilled indeed!
• You need to constantly be getting better
2. Edge as a Story
• Sounds like you'd rather have math but NOT TRUE
• Just looking at the data wont get you anywehre: WHY does it make money?
• Logic and math can help describe/bring clarity to a fuzzy edge, but they are not edge in and of themselves
• Backfitting 100's of strategies will find one but who cares?
3. How are edges dsicovered?
• usually starts out with noticing something: "hmm that's odd"
• a hunch of why gives you license to test
• if the math checks out, have to dig deeper on the qualitative side - why/how long/when?
4. GO TRY IT

Models

1. Types of Models
1. Generative model - describes a true underlying state of world
2. Phenomenological models - describes an observed irregularity (good enough)
3. Ex: Black scholes option pricing model is generative, the implied vol smile is phenomenological (doesnt actually make sense since should all be same IV)
2. Dangers
• Dont want to over rely on generative models as being the TRUTH.
• yet phenom models do not tell you why something should be happening, so you lose the story and the why/when
3. Theoretically, models should be generative but this is not always possible in the messy, real world.
• Another reason stories are important (& should be re-evaluated) to models
4. Good models
1. Robust - not overfit, simple, few parameters where returns are not super sensitive to inputs
2. Inpsectibility - should be able to reverse engineer and see how it is capturing the stated edge
5. Data
• How much do you need?
• You want enough to reach significant conclusions, but no more. dont want to overfit, different regimes, etc
• Partition data into in set and out of set (for training and testing)

Costs and Capacity

1. Categorizing Costs
1. Costs can be linear (brokerage) non linear (margin, market data)
2. Costs can be visible or invisible (commission vs bid/offer)
3. invisible non linear cost like herding (everyone short vol at the same time) can be deadly
4. Every profitable strategy has a depreciation cost - it is getting arbed away over time

Possibility

1. Induction - this is really how we come up with (hopefully) profitable strategies
• how do you know the sun will rise? induction
2. Feedback both positive and negative really matter
• reflexivity
• mean reversion vs momentum trading
3. Correlation and probabilites
• Even the tiniest of correlations (like two engines on the same plane thought to be independent) means the risk of failure can go up by 100x

Alignment

1. Have to align incentives of traders with firm - blow up risk heads i win tails you lose
2. but there are tons of other pitfalls:
• capital, high water marks, lemon problem with traders (for firms and investors), culture

Technology

1. Data is everywhere
2. Why do we still need human traders
• creativity
• models are not the real world
• come up with conclusions/hypothesis to test. if the numbers work you have a better basis for why you are making money
3. Software engineering and debugging are significant challenges

Adaptation