Outlier Events

Discussion in 'Trading' started by man, Nov 14, 2002.

  1. man


    From my perspective there are basically to ways to look at information to look at prices.
    First is to look at major tendendcies, like any kind of moving average and most of all indicators do. Here the main idea is to get rid of the noise covering the major tendency. In essence it means to reduce information to its essence. I think that is a basic principle of most tools.
    But there is a second look one could apply to the markets: ignoring the essence and looking at the outliers. As an example take any kind of volatility bands. now just plot those points, lets say closes of minute bars, outside of the bands and analyze their behaviour. When they happen quickly one after the other, what does that mean. Is there a constant struggle of some people trying to push the market, with others sill refusing, and finally the pushers win? I did some analysis on that but have not found satisfying results so far. I had the feeling that I could even predict gaps by applying this to the last minutes of trading, but it was not statistically significant over a longer period of time.

    Now my question: is there already research out there on that kind of subject?
  2. Man

    I think the key to your type of analysis is to trade with some kind of 'market context'.

    By that I mean don't look at every example as the same. For example look at AOL. In september prices went here and there, a fair bit of randomness, but then 10/24 onwards there was a lot of context in the market. The context being the breaking of the important Aug high. Will this breakout work or fail? We don't know but I'd say that there might be some better or more reliable tradable patterns at these important points.

    I've always been a big believer that both sides of the random/non-random argument are right. Basically sometimes the market is random, sometimes it's not and perhaps the secret is to realize this, and only bring out your big guns when you believe the markets are tending towards non-randomness.

    See what you think anyway.
  3. man


    I think you are right. Most of the time the market is random, or better to say unpredictable. But there are times when the odds are shifting. Thats the time to trade.
    Predictability is a relative concept. What seems unpredictable to me might offer a clear pricture for you. What was unpredictable yesterday might make full sense today.
    I think troy actually wants to shift odds in his favour by adding tools to his trading.

  4. man


    sorry for my confusing answer. I thought you reffered to another post in another thread. Here is my real answer.

    The trade you did is to some extent what I mean with this outlier thing. But there are others. Consider a standard bollinger bands and only look at the outliers out of the band. Do they tell a story? I do not mean that once you see them you immediately trade them but to follow their behaviour. To me a gap, a breakThrough at support or resistance, or a quick move away from the peergroup ahve something in common: they change the odds.

    The same with volume. Everyhting I saw so far on volume was some line drawn at some chart that should say something, but hardly said anything once you tried to program it. I wonder if volumeOutliers might be a better or equally good indicator for support/resistance as historic highs and lows are. At these points many people place action in the market. Either they went in or out - but they did something and they will remember.

    Another thing is the shape of the volume curve. Usally it has a wellKnown u-shape. what does it tell when this changes.

    As you see: i am looking for extremes and want to know if others do that as well.

  5. man


    I looked at your AOL example. Actually it was intereesting to see that jump in volume - traded 50m stocks compared to the avg 20m stock. Moreover the stock made an outlier gap of about 3%. All quite unusual events = outliers I would say.
    By the way it could be an outlier even nothing happened. imagine the market and the peerGroup gap siginificantly - AOL does not. Another outlier.

  6. Tacsian



    You might want to look into some of the CVR work done by Larry Connors (i think). He examines a number of indicators, such as the VIX, and traded based exactly on outliers..for him, when the VIX pulls away more than 10% from its moving average I believe.

    He runs a site called Tradingmarkets.com, you can search ET for references to what people think of it. I think you can get a month of access for 20 bucks or thereabouts and search all their archives for strategies and formulae etc.

    In essence, many trade with a strategy that looks for overreactions, excessive moves, abnormal volume, pierced Bollinger bands, or other outlier moves away from standard indicators and then they fade or position against such moves. It's a legit edge if you quantify what you're looking for and trade it right. Good luck!
  7. man


    thanks for your input! I will check this and come back. Do you trade yourself in that field?