4 ms·
As an individual over even the fastest internet connection there is no way you can do High Frequency Trading. You need latencies of sub-millisecond to do this.
by retube 16y ago
As an individual over even the fastest internet connection there is no way you can do High Frequency Trading. You need latencies of sub-millisecond to do this. The guys that do do this have very expensive hardware co-located at the exchanges. All this costs a LOT of money (a friend of mine who developed an HFT platform for a large bank reckoned you'd need about a million bucks to have even a vague chance of comepeting). Additionally it's also VERY risky, and not the free lunch it's portrayed in the media. You're "picking up pennies in front of the steam-roller" - one mistake and you're f*cked. And trust me, mistakes get made: if an order doesn't fill/complete and you're left with an outright position you can lose a lot of money very rapidly.
So forget HFT. However yes there may be opportunites for relative value / alpha / rich-cheap type analyses, which AI/ML could help you with. But bare in mind that markets are often "confounding" - and never do what they are supposed to. So play carefully.
Edit: I should also add that HFT requires a very fundamental knowledge/understanding of contracts and their pricing (mostly futures), how exchanges match and fill orders, how curves are constructed, how the markets operate and so on.
- adn37 16y agoAgreed. As a developer in the field, I confirm access is very expensive and the software base deployed is very heavy (custom protocols built over TCP and engines). Basically, if you go through HTTP (efficiency put aside) then you go through a broker. This adds up latency and voids any chance of HFT.
- nailer 16y ago> You're "picking up pennies in front of the steam-roller" - one mistake and you're f*cked. Isn't that why exchanges have a simulation FIX ports?
- adn37 16y agoDepends. We use simulation to validate connectivity and business logic. No link to the potential future bad decisions taken during trading.