the basics of edge extraction

a summary of this presentation, available on youtube here

trading is hard. you should be anxious about the enormity of the task. you need a plan, just as you would with any other business endeavor.

losing money trading is easy. but it's quite hard to lose a lot of money if you avoid really dumb stuff:

you are always trading against other people. so, you need to find occasions when others are forced or willing to trade at prices that are bad for them and good for you. no amount of smarts or will on your part can make this happen. you can only eat what you are fed.

so you need to find occasions when some party (or group) is price-insensitive. maybe they are trading because they have to regardless of price (systematic rebalance, liquidation) or because they are trying to achieve something else (tax loss harvesting, window dressing etc.)

then look to help them out, or take the other side. (we get paid for doing stuff that is useful... things that reduce market impact of others trading, or trades towards market efficiency)

but doing useful stuff is a competitive business. it's quite easy to find donkeys prepared to trade at bad prices (hi robinhood users), but a lot of smarter faster people are competing to trade with them.

so you need to sniff out less competitive places in which to be useful. "doing useful things that suck". if you don't know what sucks about a trading opportunity, it probably isn't a good one. you're smart but you ain't that smart.

maybe the opportunity doesn't come along that often. maybe it involves trading in less liquid markets and it's hard to get out of positions. maybe you're constrained in the amount of money you can make (can only eat what you are fed). knowing why you can compete is important.

this involves humility and getting over yourself. you aren't trying to compete in a battle of wits and science in a zero-sum game. you're trying to do helpful things people value and others don't want to do.

like a plumber.

now having identified opportunities where:

... you design a set of business process to exploit the edge.

his wants to be as simple and direct as possible. don't get cute. market behaviour and outcomes are dominated by noise. the best we are ever doing is throwing shit in vaguely the right direction and expecting some of it to stick.

and remember that the market effect informs the model and rules you use to harness it, not the other way around.

finally, squeeze as much as you can out of each opportunity, but don't depend on any one thing to pay all the bills. market effects are fickle. trade more shit.