Would members here recommend we - with no trading experience - embark on algorithmically trading our own strategies with millions of our own money? What are some of the pitfalls of trading an algorithm? It's not just ALL automatic, for instance, is it? The algorithm does everything?
One of the pitfalls is the cost of a low latency set up. You have the cost of IT hosting and non-display market data. The smaller the account, the more the cost will overwhelm your returns. If you can along without low latency, you can keep your cost down with a solid API. If your missing trades in testing with small lots, you will know it won't work. With the API, you might not need millions to run your system. If you have a full time job outside this project and can't monitor the trading of the algorithm, you might have to pay someone to do it.
It's low latency. It's a daily, zero investment (or long/short), SPY strategy trading 100 winners and 100 losers
Then that will be your big issue. See attached. In addition to IT hosting, non-display market data, you need to pay for a cross connect to your broker's FIX connection. That runs around $500/month at NY4. This set up can easily run as much as $10,000/month. If your account starts with $1mm, you have to earn 12% per year to just cover your IT costs. It really depends on what you need. You might be able to cut it down a little.
Start testing with just 100 shares... but also have a basic knowledge of your platform so you can trade out of your positions manually if something goes wrong.
Performance monitoring should be your foremost task with this endeavor. Next, think about building into the algo bet-sizing based on performance.
There are a lot of cheaper ways to do this. For instance I can rent a shared co-located server that has my own copy of multicharts on it for about $500/mo.