Basic Example of Automated Trading in Equities

Discussion in 'Automated Trading' started by wilburbear, Aug 12, 2006.

  1. Anyone have a basic example of an automated algorithm for equities? I don't know how these work. So, I will venture my own basic idea to get flamed or added to. If you know how these really work, post it here.

    For stocks having a certain average daily volume, say, over 200,000 shares a day, find stocks with a bid ask spread of one half of one percent or better. Also filter with stocks having a beta under 1.25. Now we have stable stocks with workable volume, and wider bid ask spreads.

    Place bid and ask one penny ahead of the current bid and ask. Adjust bids and asks with beta-adjusted percentage movements in the S&P mini.

    It's incomplete I know. I thought of it, and typed it, in 5 minutes. For those who know, what's missing?

    What I don't understand most of all, is an automated system in stocks with news, or strongly trending stocks. I hear algorithms trade those markets as well as other markets. But, doesn't the algorithm keep selling, or at least selling and covering for small losses as soon as a loss is incurred?
  2. Anyone wish to comment on how automated trading works in equities?
  3. A paradigm I call XOT trading. I have fleshed it out the big picture in detail and worked out the scaleability out to hedge fund proportions regardless of market conditions and available EQUITY instruments. Futures are a bit more complicated. I have come to understand it to be the personal ultimate in efficiency of capital and market participation. ET does not like lenghty posts and for not so obvious reasons, I do not divulge. At the heart of it, like all trading is, the PV relationship and more importantly, their mathematical derivatives. Most do not understand the V and just watch the P. The V is the fine tuner and it all boils down to WHEN to do WHAT your are informed to do. It's almost like an "are we there yet?" cycling followed by a DECISIVE, "we are here"...

    In a nutshell, it works optimally with using a dozen or so very liquid high price (~$70+) assets who have decent daily and fractal volatility. The idea is to ride the mover and to do this you have to use percent bars so as to put everything on even keel. It is one of those frameworks in which juggling many balls comes in useful...

    There are a number of fine folks who do the automating thing well over in Most of them see the tip of the iceberg. It was always the 90% that I did not percieve that I had been after for a very long time... I'm sure you can guess as to what such a paradigm deals with...
  4. When you trade stocks at least on short-term basis unusual volume is a key. Just ask yourself about major advantage of stock trading over futures trading. Look at ES, YM or ER2 intraday dynamic. You’ll see that only about 50-60% of days contain low risk trade opportunities 1-2 times daily when market trends strongly. Market profile traders can say “market is looking for new value”. Then look at stocks – you’ll see dozens strongly trending stocks almost every day. What is common for these opportunities? Right – unusual volume. So you know what to monitor. The hard part of this monitoring is technical side of the problem. Most strong trends develop during morning session so you have 0.5-1.5 hours of active time. It is the case when automatic trading can greatly beat discretional. You need right software to 1. monitor unusually high volume stocks, 2. select stocks, 3. trade selected, 4. manage risk.

    You can actually create asset specific profiles of assets. It is very neat. I do this by choosing some not so arbitrary time intervals to look at 5/10/15/30/60/90M. 30M and 15M charts are pretty nice. Then I establish the "Mean" (as in mu) profile. To do this, I grab the last 60 days of historical data (180 would be better). Then I sort and group the bars according to their interval so that I get 60 9:30AM bars, 60 10:00AM bars, 60 10:30AM bars, etc... Then for each interval, I will sort each bunch of 60 according to their Volume in order to get an unbiased distribution. From this I then generate a frequency distribution plotted across time (9:30AM to 4:00PM). What you get is a complete equity profile with volume histograms plotted along time in order to normalize things so to speak. It is very nice to watch during the day and what's even more spectacular is that you don't find any jumpiness as the active cell moves along during the day. It works great if you trust the PV relationship. Most traders don't, so I guess it can be considered an edge. However, I am certain you understand that unusual volume indicates unusual returns... You can gauge this unusualness by using instantaneous derivatives like (Pro-Rata-Volume "PRV")... Do a search in this forum to find the concept and exact calculation. :cool: