peculiarity this year

Discussion in 'Automated Trading' started by ssrrkk, Oct 24, 2011.

  1. ssrrkk


    I have several automated strategies that I am back and forward testing. There is one in particular that relies on morning volatility / reversals. This one was a consistent money loser up until the beginning of August, when the VIX shot up. Since then the consistency of this algorithm over the last 50 days or so is quite amazing. When I plot the profit curve for the year, it is a V shape with consistent losses up until August, then consistent gains since then (of course there are losing days even post August, but the expected value of the algorithm appears to have turned positive). Anybody else seeing something like this?
  2. Yes. If you have studied the market for any amount of time, you will see many things very similar. It makes sense as well. The market is simply a summation of human behavior, which can be rational/irrational, but is always and forever changing & evolving.

    You should test your strategy for more than just 1 year, for more security.

    There are several (apparent to me) relaitonships between the vix and startegies (or money that can be made). Any strat employing 'letting winners ride' will enable those to make more $ when the VIX 'shoots up' (VolatilityIndeX) because when you are right it will be more obvious, if you can sit tight. Volatility tends to spike at market bottoms, therefore any long strategy will fare better when VIX 'shoots up.' Again these, are just a couple of my own observations.

    Look at the equity curves of some of the journals on this website. It's a common thing to see 'v(s)' in the equity curves, of infinite variation of time, trade # and $ size, because of the factors listed above, mainly, point #1.

    The market will never play 100% to any systematic formula you can apply to it.
  3. ssrrkk


    Yes I am well aware of the non-stationarity of PL expectation. But I just felt in this case, the difference between pre-August and post-August this year was quite striking and I was wondering if anyone else had an algorithm that exhibited this two-phase behavior this year in particular...