Capital, Costs, and Execution Risks in Automated Trading

# How much capital do I need to create a competitive automated trading strategy?


# How much capital do I need to create a competitive automated trading strategy?












I'm a relatively small investor, and I'm interested in building my own fully-automated quantitative trading strategy. I also read about dark pools, and how difficult it is to get good prices on orders.

My question is, in the context of the other players on the market, how much capital is necessary to create a viable trading strategy? I realize that getting a high profit margin will be nearly impossible.

The kind of trading strategy I'm referring to would hold securities anywhere from a day to a year. I'm not looking to do HFT or scalping.

## Answer by Strange (score 8, accepted)

https://quant.stackexchange.com/a/4111

Ha, interesting, so many responses with "negative" expectations. There are plenty of people that have successfully gone down this road and are producing pretty nice returns, so obviously it is possible.

A trader with a smaller capital has better chances of producing good ROC with very reasonable risk parameters, simply because he's would not be constrained by liquidity and ability to deploy capital. In fact, for someone trying to deploy a million or two, there is a number of capacity-constrained opportunities that can easily produce high double digit returns with Sharpe ratio north of 2.

Market data and hardware are pretty cheap these days. You can get tick-level data for 10k and you can get 1 min bar (all that most non-HFT systems would need) for under 5k. A reasonably powerful box will run you another 2k. Pretty much all software need are covered by FOSS.

So, to answer your own question, assume that you can produce returns around 30 percent with Sharpe of 2 (a very reasonable assumption for small capital deployment) and that you need about $500 a month in data fees and various fixed costs (e.g. co-location, brokerage charges, new hardware etc). In short, if you have the requisite IT and quant skills plus the requires market experience, you can do it on a very tight budget. In fact, I know people who do this sort of stuff and they started at 300k.

## Answer by SRKX (score 7)

https://quant.stackexchange.com/a/3874

As a starting point to my answer, I would say that reading a book is not sufficient to start doing automated trading on your own as chrisaycock suggests in his comment.

I would answer your questions in 3 different ways.

First, building your "AI bot" which I would rather call a systematic algorithm not only requires programming skills, it also means having access to market data and, if you want to make it fully automatized, access to brokers APIs to send your orders. All this costs money, so if you're not a big, big investor or a firm, it's not likely that you will have the sufficient fund to really tackle the world of systematic trading. Don't even start thinking about feeding prices manually to a systematic strategy as an individual; you'll go on holidays or get drunk some night and forget to do it which will screw it up.

Second, building an automated trading strategy which looks good in a backtest does not mean that it will do well in the live markets. This phenomenon is called the Implementation Shortfall and it adds to all the usual biases such as data-mining, survivorship bias and so on. Essentially, depending on the amount you wish to invest, you'll just give away your gains in transaction costs. What do you do if there are bugs? What do you do if you have no internet for some time?

Finally, even if you're successful for some time, keep in mind what Nassim Thaleb calls "Monkeys on typewriters" in his book Fooled by Randomness (which I suggest you should read if you haven't already). Basically, this concept explains that out of billions of monkeys who would decide to write randomly on a typewriter, one will finally end up writing the Odyssey of Homer. So, it is possible that you will start by making profit, and it is possible that you know someone who's been doing so on his spare time for 10 years but bare in mind that it might just be luck. Being successful in systematic trading requires a lot of experience which comes with making mistakes and hence loosing money (so, if you're not ready to lose some big amount of your own cash it's gonna be difficult). Also keep in mind that even the most brilliant minds combined together to from systematic systems sometimes get it wrong - very wrong, so being alone won't help.

I'm assuming here that your strategy would be looking to generate alpha rather than beta. If you want to do an asset allocation system trying to optimize your risk-adjusted return as a small investor, then it makes more sense to me.

## Answer by Ricky Sahu (score 1)

https://quant.stackexchange.com/a/9516

Speaking from personal experience, you definitely do not need anything like the first response alluded to. I'm running a pretty sophisticated outfit without the need for any monthly fees for data, you can find everything online these days. Monthly fees for trades of course can run into a few hundred bucks a month. If you work on your programing skillz and get informative data, you dont need a fancy machine just a cheap server (mine's under $30/month).

Basically, you can scrape by with $5000 and a good automated trading algorithm (sharpe in the low 3s and trading stocks/etfs that move more than the market usually does). 15K to 20K will of course mitigate losses from fixed trading costs.

Like the second response suggested the AI bot is the most important piece. It just comes down to hard work, stay up late and work on this most weekday nights and weekends. Don't read books, just learn how to code and stats from the internet.