Ask HN: is algorithmic or HF trading something I could do on my own?

[nearestneighbor]

Assuming I had the necessary skills (and I do, in math and programming, but not finance or trading at the moment), could I do algorithmic or high-frequency trading on my own, or is it (participation in the market, low-latency information, etc.) priced too high for an individual?

[yummyfajitas]

Yes. I know of a person trading profitably with about $200k capital. The company I work for started off this way as well, and it's considerably larger than $200k. I also know of a person who went this route and had trading profits but his trading profits didn't cover expenses (he lost something in the neighborhood of $50-100k).

You'll typically need to buy a server and colocate it at your broker's datacenter.

Brokers to talk to:

http://www.limebrokerage.com/offerings/by_role_highfrequency... (This is the same Lime as LimeWire.)

http://www.gndt.com/

http://www.interactivebrokers.com/ibg/main.php

Unless you are very rich, you probably can't afford to do latency arbitrage, which is only a subset of high frequency. The capital expenses (custom routers and the like) are probably too high for individuals.

[nearestneighbor]

Thanks very much for the links!

[yummyfajitas]

Also, if you are interested in learning HFT without risking your own capital, my company is hiring. Instructions on how to apply are here (in encrypted form):

http://meshcapital.com/application.tar.gz

[yourabi]

If you are asking the answer is probably no.

It is a very capital intensive business.

You need a large amount of capital to get started (single millions? maybe up to 10 million?)

You have a high barrier to entry: colocation at exchanges (expensive), expensive data feeds, sophisticated event-processing software...etc

Then after you have all that you need the brokerage relationships.

Then you need to actually develop a successful algorithm - and your competitors have many years more experience than you.

...etc...etc

You get the point.

However, you could look at automated technical analysis / to supplement your own retail trading, but HFT is its own game.

[nearestneighbor]

There's the possibility of middle men who are supposed to make some of these obstacles easier to overcome: http://news.ycombinator.com/item?id=1560072

Are you speaking from personal experience by the way?

[nl]

Yes.. maybe.

I developed an algorithmic trading program for fun once. I never let it loose on a real account, but it did better than break-even on a few months of data I collected.

BUT: the trading costs killed its viability (even using interactivebrokers), unless I upped the capital I was using to an amount that I wasn't comfortable risking.

There's a useful forum at http://www.elitetrader.com/vb/forumdisplay.php?s=4a039395d01...

Personally, I think you have less of a chance in making HF work for you than other methods though.

[jasonmar]

You are probably not going to be able to do enough volume initially to get a discount on your execution. In futures the top tier requires 10's of thousands of contracts per month, in equities you'll need to be trading millions of shares per month. It would be extremely hard to overcome this even with great algorithms.

[carbocation]

Are you colocated at the NYSE?