Optiver - 3σmales

Inspiration

Money basically.

What it does

Make the market (reducing the bid-ask spread) while making us money

How we built it

python and using their interface and library

Strategy

math(best bid of stocks in avg) > best offer of etf ← buy etfs, sell stocks
- demand is higher than supply ⇒ buy first, then sell
- supply is higher than demand ⇒ sell first, buy later
math(best offer of stocks in avg) < best bid of etf ← buy stocks, sell etfs
- demand is higher than supply ⇒ buy first, then sell
- supply is higher than demand ⇒ sell first, buy later

Problems:

Problem 1

imagine scenario math(best bid of stocks) > best offer of etf so we want to buy etfs (for the lower price) and sell stocks (for a higher price).

say there is a high supply for etfs (say 50) and a low demand for stocks (say 2). this means amount(stocks) << amount(etfs). Here it is a relatively safe trade to short sell stocks first, and then buy etfs, since we are pretty sure there will be etfs available
Now say there is a higher demand for stocks (say 50) than a supply for etfs (say 2). this means amount(etfs) << amount(stocks). Here it is a relatively safe trade to first buy etfs and then sell stocks, since we are pretty sure someone will still buy our stocks.
Now we come to the “risky” parts. what if the demand of stocks == supply for etfs. Here either trade we execute first, we are at high risk and very unsure if we can still sell/buy accordingly. thus we need a delta, a range, how far away we want to stay from this case! (Fist risk hyperparameter that can be learned)

Problem 2

what if we buy but can't sell for our calculated/estimated price (sell for profit) and thus take a potential loss and vice versa. Solution: include a loss tolerance (first implementation taking infinite losses), after which we discontinue hedging (then we would need to compensate and recover hedge ratio via strategy change) (Second risk parameter)