Detecting volatility regime intraday

Discussion in 'Strategy Development' started by clearinghouse, Sep 14, 2012.

  1. I have a strategy that does extremely well in slow uptrending markets, sideways markets, and lazy-slow down trending markets. It does extremely poorly in very fast popping-up trend markets and very fast popping-downtrend markets. The more "pop" in the ES, the more my performance slips into oblivion.

    I've been observing it run for about 4 months and I can pretty much look at a chart historically and point to parts where it does well and does very poorly. So, to juice the results, I've been toying with increasing size during "good times" and reducing size during the bad times. Doing this has smoothed out my results, but I've been doing this by eyeballing it. There's no rhyme or reason, just me looking at trading results, the charts, and turning up the knob (or down.)

    I'm trying to figure out some way to automate this. One idea I had was to look for changes in the thickness of the books, since fast and "poppy" markets tend to go hand in hand with thinner, flimsier looking books. However, this doesn't work that well always because sometimes there are algorithms from others that come into effect and cause slippage/bad-performance to increase.

    Is the correct automated approach to look back at my own results, track performance, and incrementally step up and down the size based on historical results? I was trying to find a post in this forum by a guy who had a system for doing this, but I was unable to find it.
     
  2. No, the guy was talking about feeding in his equity curve back into some diagnostic function which then reduced risk.

    It seemed really complex, but I've essentially been doing the same thing. I cast out a wide net, see where the 100 lots are hitting winners, then crank up size until I start seeing failures. There's always a pool of losers, but they're not that heavy relative to the winners.

    I was thinking at the very least I probably need to visualize the equity curve or trade-by-trade results so I get a more concrete grip on things.

    Although, generally speaking, for me high volatility is bad. The problem is that volatility strikes out of nowhere sometimes. An example : you're in a long side-ways range, making your range over and over, and then *blam* someone comes in and blows out the range, for example.