on finding an angle and whacking it repeatedly

if you wanted to bet a sports game effectively you might:

  1. estimate the outcome probabilities yourself and compare them to the odds
  2. look for situations where those setting odds tend to get it wrong, on average

we might call the first one "handicapping" and the second one "playing angles".

if we're handicapping, we need to be "more right than the market" on average, across all the possible things that can happen.

if we're playing angles, we can afford to care about one situation only.

how does this apply to trading?

when i had more energy and less money, i used to trade illiquid locally listed ETFs.

these might only trade $100k in a day.

they were local wrappers around US listed ETFs with huge expense fees.

they exist for convenience and tax reasons.

typically, they had one or two market makers who were obligated to quote a fairly large fraction of the average daily volume, about 1% wide.

they quoted wide and picked up little trades during the day, probably making just under half the spread each time.

they're playing the handicapping game.

they have to produce a fair value all the time and quote around it.

they gotta be less wrong than half their 1% spread all the time to make money.

and they surely were. that ain't hard. and they have nice commercial incentives.

i could have played the handicapping game too if i wanted.

if i came up with a better fair value, i could quote inside them and pick up those little trades at good prices and leave the other market makers to pick up the bad prices.
but this is inherently unstable.

i'm competing directly with them in an obvious way.

and, not only that, but they have better commercials than i do. a big advantage. they can easily freeze me out if they want to.

so my job became to look for angles.

first i had to understand, as best as i could, what the market makers were doing.

(they're providing all the exploitable liquidity in the market - there's no point trying to understand the random small flow they're harnessing)

then, having understood how they are quoting, i could try to find some angles.

i'd look for some situations when their quotes would be drastically "wrong" by an amount i could exploit after costs.

i could afford to ignore the rest of the market.

i concentrated on how the two market makers updated their quotes in predictable, systematic ways, based on movements in liquid tradeable international assets (like ES futures) and recent large trades they had recently made.

now, understanding what they're doing, i could try to work out when they're likely to be wrong.

i had a model that (mostly) matched what they're doing.

now i could mark that out and look for circumstances where they're likely to be very wrong.

and having found some angles, i lay in wait, waiting for them, and i pounced when they occurred - taking as much liquidity as i could.

eventually they notice this, of course. my trades were very obvious in a market like this. but it's hard for them to do anything about it quickly.

they are obligated to quote a certain amount a certain distance a certain amount of the time.

my trades were telling them where and when they are wrong - but that doesn't mean they're going to figure out why quickly.

eventually i outgrew this stuff and it became too much effort for too little reward.

but it serves as a good example that you don't need to model everything all the time.
you don't need to know much to trade profitably.

you can find one tiny little edge case that is mispriced - and just keep whacking it.