Algorithmic Trading

Discussion in 'Strategy Building' started by Scottsdale, Aug 18, 2005.

  1. cosine

    cosine

    There are three mathematical "effects" at play here (from simplest to more complex): gradient, divergence and curl. A former French poster in ET would acknowledge this smartly and thoroughly. The algorithms' propensities work inversly with respect to the importance of the market flow variations.

    There are very few threads in ET that look into sizing up how market maths can be used to focus on more or less important things in profit making. To reach the compromise among the money making factors, it is an interesting consideration of fixed costs, anomolies and being on the real right side of the market. Optimization takes these out of the picture in the order listed.
    [/QUOTE]

    I would like to see your maths and theory so I could aknowledge them smartly and thoroughly too.
     
    #11     Aug 18, 2005
  2. What I am trying to get at is that, imo, very few of the players can play these games, although three people that I know of on ET claim to be able to trade stocks 100% by computer, and no one in their right mind is going to tell you how they are making money doing that (if they are, which is a big IF)

    THIS IS NOT TRUE>>>>>>>>>>>>>>>>


    i personally know of 5 differnt individuls at my firm that run black boxes...100% automated trading boxes

    the WORST of them is up 100k on his boxes this year, the others are up many TIMES that

    obviously they are not out there tellin eveyone their stragies...quite opposite...2 of them are in dedicated rooms that NO ONE...and imean NO ONE can enter...period

    we know they are out there....there is even "box wars" goin on...where there are boxes that look for other boxes to "trick" them..high bid, high bid...smack

    anyone who trades nazzys will occoassionally get hit on 1 share while they are hidded in front of size of a thin issue......thats done purposely...its not grandpa billy buyin $1 share in his 410


    In terms of reversion to mean theory...if you have UNLIMITED CAPITAL...( i know this doesnt make sense) ..but saying IF...then you could theoretically double down on INTC, QQQQ, etc....every say 10c...until you get a 10c bounce in your favor......u'll never have a losing trade (trust me ..i thought of doin this...but i konw eventually i'll get blown out in 1 trade)

    its like sittin a casino and just bettin red ALL NIGHT
    u'll walk out with cash...but ur pickin up peenies in front of bulldozers
     
    #12     Aug 18, 2005
  3. if it is so easy to make money with automated trading why do prop firms still hire traders. why wouldnt they just program computers to do the trading?
     
    #13     Aug 18, 2005
  4. The edge I can gain over a machine is achieved when I figure out what that machine is reacting to. A simple example was trading the large cap nasdaq stocks off of the S&P and NDX futures back in the old SOES and SelectNet days. Because I knew that a huge amount of index arb was being done, I could trade in front of the arbitrageurs when cash and futures became out of line. By trading a few NDX stock which were highly correlated to the index, and through my superior order routing knowledge I was consistently faster than the arbs who were trying to buy or sell a much larger basket. Also my feel for the momentum allowed me to anticipate when programs would kick in before they actually did. The problem I encounter with today's nasdaq is that it's not clear what variables are triggering automated buy/sell decisions, and there is no advantage to being a highly skilled order router. However, I do believe that with the knowledge of what variables a machine is trading off of, my observation skills would allow me to anticipate the triggering of buys/sells, and I could then get in front of them.
     
    #14     Aug 18, 2005
  5. I coded that strategy once in Tradestation....
    long term: didn't make any money
    Good news: didn't lose much either.
     
    #15     Aug 18, 2005
  6. duard

    duard

    Indeed. At about 2:30 EST (1:30 Central) on the ES @ 1220 an ask for 10,500 lots appeared. Of course it was a spoof and was meant to "confuse" algorithmic boxes.

    I rarely see that big a spoof. Of course the limit is 1,500 lots per trade and the 10K contracts were 2 ticks above the market but you get the point. Unfortunately this didn't yield any significant trade off the info mainly because today was consolidation.
     
    #16     Aug 19, 2005
  7. I saw that too, however, only simple order book balance algorithms would get confused by such as size ask, it depends on if the algo maintains a theoretical price, so the size of the book is only one factor in one set of possible algorithms. Most black boxes I know takes a variety of algorithms and then select a statistical most probable fit ...
     
    #17     Aug 19, 2005
  8. This is a reasonable statement if you talk about a particular machine that you happen to have some kind of knowledge about.

    The newcomer 'machine-traders' in the market are likely pretty dissimilar. So your 'edge' over one particular player may not mean anything, excluding perhaps some weird frontrunning/size scenario.

    One further hypothesis would be that many identical machines are trading in the market. This is as unlikely as many identical human players making identical moves based on fixed decision rules in machine-free environments. 'Anything that's known by many ain't worth a thing'.

    Here you kind of touch upon the cardinal belief & hope of all TA players, machines or not. We all try our best at this. :cool:

    Pretty true rufus! And if you got it, better not talk too much about it!
    PS: I have my doubts whether putting 10 people on it will bring you much.
     
    #18     Aug 19, 2005
  9. It is by no means easy, finding a profitable strategy from high volume trading is quite hard, and then it takes a fair bit of work to develop the automated system itself. It is not un-usual for such a system to take *years* and 3-10 people.

    Most prop firms I know do not have the technical or the quantitative resources available. Surprisingly, individuals with expert knowledge of quant analysis and system are known to build entire systems by "themselves", but they have no incentive to share their work with anyone else.

    The only few prop firms I know that have the pockets and infrastructure to compete are Wolverine, DRW, Peak6, to name a few. Not surprisngly, this came across this morning, even Goldman are thinking about getting rid of exchange presence, somehow I don't think it will be Peak6, having know their internal profitability over the last 1-2 years.

    http://today.reuters.com/news/newsArticleSearch.aspx?storyID=271059+16-Aug-2005+RTRS
     
    #19     Aug 19, 2005
  10. This above is pretty on point and similar to my experiences with my own automated traded and that which i have seen over the last year.

    Pertaining to early comments, i don't have any stats to back it up but most of the higher frequency strategies i have seen or heard about were mainly mean reversion.

    This makes sense if you think about it. Market Makers and specialist are basically liquidity providers and lots of the boxes are set up to trade similarily.

    As far as the "what happens when the market goes nuts or has a big trend day", well everything has a shut off switch. Generally the algrothims are robust enough to know what unusual is, and when "unusual" happens they either shut off or the algo adapts to trade differently (more conservatively).
     
    #20     Aug 19, 2005