I am developing a day-trading system in liquid and volatile stocks. What are my best broker options(in terms of commissions as well as possibility of trade automation) if I want to be able to trade relatively short time-windows, such as 3-5 minutes, placing a large number of trades a day? If anyone has tried something similar, how difficult is it to trade such short time frames with equities? Thanks in advance, for any responses.
I can program very well (cs phd), so that is not an issue. I am developing my system my analyzinga pretty huge amount of tick-level data from scratch, and can implement the system as fast as needed. I am wondering what sort of broker I need to make this feasible. To the extent that it is possible I would like to keep fixed costs down (and obviously commissions relative to trade size.) Timely real-time data delivery will be important..
Since you have an advanced degree in Computer Science... You should be able to comprehend that sub 1 millisecond platforms are already 3 year old technology... Such as this: http://about.reuters.com/productinfo/informationmanagement/material/Reuters_DataFeedDirect.pdf What you are planning to do... Is compete directly with 100s of players with 7 figure technology budgets. That's like flying to Vegas and sitting in toughest game at the Bellagio... Instead of finding and crushing soft games at lower levels of Party Poker.
Bear in mind that the players with 7 figure budgets need much more scalable strategies to make it worth their while. I am still in the exploratory stage, and am researching data feeds/broker options in order to add realistic slippage/commissions estimates to my analysis. In addition, it seems to me that trading S&P futures is also similar to the bellagio game you described, given the liquidity and the number of players there. Given your pessimism, what time-frames for intra-day equities trading would YOU consider doable?
I'm not being "pessimistic"... Just realistic based on > 1,000,000 trades over 15 years. I have a Computer Science degree as well... And build ATS... But would never even consider trading the most liquid stocks... Where latency = $$$$$$$$$ is paramount. You mentioned a KEY point... scalability. Look at the bottom 30% of NYSE stocks by volume... Find an interesting niche and specialize and automate there. Because lower volume or exotic or complex stocks do not scale well to the Big Players... Your competition will be weak... In other words... you will be able to find and exploit market inefficiencies.
Thanks a lot for your input, I will definitely consider it. It is still worthwhile to me to evaluate what latency and transaction costs my current approach can tolerate.
I think that a better way to ask my question might be "What sort of transaction costs do you factor in when backtesting intraday equities trading strategies?". Is it reasonable to assume average bid-ask spread + some flat commission? I also have some systems that execute fewer trades at half-hour time-frames or so, so that is also relevant there. I want to backtest in a realistic manner, and I am aware that the broker/links will have an effect on what 'realistic' is. I'm sorry to respond to my thread so much. :eek: