What do you mean, exactly, by starting a trading shop? If you mean becoming a broker dealer, that will require certificates and capital - and lots of it! :P
I was thinking more along the lines of proprietary trading. There is no regulatory oversight for risking your own money, and the amounts required to get started are reasonably modest. See, e.g., http://www.reddit.com/r/IAmA/comments/9s9d7/iama_100_automat...
Ahh - agreed. Getting around PDT for $2,500 and having incredibly cheap leverage makes for an interesting market (though one bad trade and your done - you have to put it all on the line with such a small account)! Though if you do not have that much capital, jumping into futures/options trading against the best in the world - you will get your head chopped off. The far shot of a reality such as the one on reddit is not a good one to try to reproduce. I definitely recommend he read "Fooled by Randomness" by Nassim Nicholas Taleb. While it's obviously a biased book (even just taking into account the title) - it does shed some light on the subject.
Regarding "Fooled by Randomness". I haven't read the book, but assume it's brought up in the inherent randomness of the markets, and people's ability to try to rationalize said randomness.
Trying to understand the market is surely a shortcut to crazy town, and certainly trying to rationalize it is a poor mechanism to profit. But it's clearly evident that the market isn't uniformly random, and that there exists exploitable patterns in the market that are reliably (to a degree) profitable over certain time periods.
I think the thesis of "Fooled by Randomness" is best taken advantage of by understanding that you need to recognize the pattern, and exploit it directly, rather than trying to extrapolate meaning and use that directly.
The market is clearly a truly chaotic system, but you just need to take advantage of patterns that are statistically relevant and profitable.
Edit: After additional contemplation, I'm not sure I actually agree with what I just said.
I currently have been making money the last 3 years by trading futures. It's been consistent, however, my co-founder of our fund has been trading for 20 years, lost his family, lost all his money, his house, everything. He was working at dominos when I took a little risk to give him some cash to trade, and boy was it a good decision. You could even say then, that the fooled by randomness theory exists - I was _really_ lucky to have found him and he was _really_ lucky to have been able to succeed. Regardless, it is possible. For every successful trader I personally know, I can give you 10 that majorly failed. It's human nature to want to try it - to be in (arguably) the most competitive business in the world. I offer my advice on where to start for free, but jumping into HFT in the really big markets is where people get crushed. Our current trading system can read a quote, send an order, and get confirmation about 2,000 times before you can blink! With that said, there are a lot of places (penny stocks, spread strategies, smaller equities) where you can thrive, and not have to trade against big machines. I will post to hackernews one day with an overview of his story and what we do
It seems that doing HFT without being in a datacenter close to the exchanges is a loosing strategy, since at these frequencies, a delay of milliseconds can make a huge difference. Do you agree with this assessment?
100% absolutely - here is a really good TED talk that puts things in perspective (sorry if it's a repost, but it's definitely worth your time to watch) : http://www.youtube.com/watch?v=TDaFwnOiKVE
For just momo or momentum traders that deal in extremely liquid trading instruments, microseconds can even make a difference. Our current feed from a US equity exchange is about 220 microseconds or .22 miliseconds - and we are on the "slow" side of things. The guys that actually read, process, send transactions usually go as far as using FPGAs even (some of the bigger firms)
That said - there are still lots of places in the market where opportunities exist!