What am I missing?

Discussion in 'Strategy Development' started by abattia, Feb 25, 2011.

  1. I’m trying to understand better why a successful intraday strategy on one instrument won’t necessarily work on another.

    For example, say something works well on SPY; why won’t it always also work well on F?

    I understand one’s an ETF based on an index, and the other’s a single company stock. But is that all there is to it?

    After all, both go up and down throughout the day, and both are highly traded stocks.

    Even if F typically moves more in a day than SPY does (in relative percentage terms), I would have expected to be able to handle this simply by scaling strategy parameters like target/stop from SPY to F. Yet, in my experience, it doesn’t seem to work.

    And - to my unseasoned eye at least - if you gave me charts for both taken on a random day, and hid the absolute values on the price axes (and didn't start the price axis at $0, so that I couldn't judge which instrument had the larger fluctuations in % terms), I couldn’t tell you which chart was SPY and which was F.

    And the order books/ladders/DOMs for both instruments also look identical (again, ignoring the absolute price values).

    So why do some set-ups work on one, but not the other?
  2. xiaohu


    I think you are asking the wrong question.

    You should be asking this instead:
    "say something works well on SPY; why SHOULD it also work well on F?"
  3. Volatility, young padawan, volatility
  4. I thank thee, O Jedi Knight
  5. Thanks.

    I think the question I did ask is similar enough (it's just the reverse) to what you suggest for rephrasing to be unnecessary.

    Indeed, my post already lists some reasons why I would suggest a good SPY strategy ought to work also on F, and so already attempts to answer the question you are proposing. Feel free to comment on these points.
  6. nLepwa


    You need to understand what are the characteristics of the underlying that make your strategy work.

    Then you can look for the same characteristics in other assets in order to diversify but more importantly you can monitor the presence of the characteristics in the current asset and take action accordingly (i.e. stop trading if needed).

    Volatility might be an important characteristic for some strategies.

  7. vikana

    vikana Moderator

    I would expect a strategy that works on e.g. SPY to also work on QQQQ and likely GE another other broad companies. I would not be surprised if it worked poorly on bonds, currencies or a company with a fundamental driver such as the price of oil.

    Automobile companies have had unique problems the last 3 years, so F is not a good proxy for "any company". Why don't you simply try all companies int he SP100 and see how it goes in general?
  8. Yes, I think this will be the crux of the matter. Thanks.

    ... so I clearly don’t understand as well as I should why it works with SPY in the first place!

    What I have assumed (up to now) is that any widely day-traded instrument (such as SPY or F) will be subject to roughly the same mix of types of trader/algo (i.e. trend-followers, mean reverters, arbitrageurs, news traders, technical traders, value traders, etc), all stalking their set-ups intraday.

    And when the prices of different instruments auction up or down intraday (at different times and in response to different stimuli) to find new supply/demand balance points, if all those auctions are attended by roughly the same mix of trader/algo types, then I have been expecting the behaviour of those auctions (the patterns of price action) to be similar across the instruments.

    This, I believed, is why you can identify common structures in the price action of different instruments, for example, support and resistance, or gaps, or trends and channels, etc.

    So if a strategy were to depend simply on structures like these, and if such structures are present in the price action of any widely day-traded instrument, why can’t one simply “translate” the strategy (by making appropriate adjustments for the differing “dimensions” of the structures between instruments) to get a strategy that also works for other instruments?

    Don’t different volatilities, for example, amount to different “dimensions” to the above examples of structure? e.g. different widths of levels of support/resistance, different sizes of gap, slope of trend, or width of channel, etc ...?
  9. Thanks. I'll try QQQQ, GE, IWM, SSO, SDS, etc ... and EFA, EEM?

    Why would you discount these instruments? Why would their price action be so different to that of SPY? [Apologies, if all I achieve is to expose my ignorance!]

    In time, I may do so... But I am hoping to direct initial efforts at the lowest hanging fruit!
  10. When you trade a strategy the returns will never be linear except for when you’re using a martingale system. Because the returns on the strategy aren’t linear you go through wining cycles and loosing cycles even if your strategy is profitable on the long term.

    Assuming you have scaled the strategy properly to work on a different instruments price and volatility the problem could be that one instrument is going through a losing phase in the strategy and if the system is profitable later on will win to balance it out.

    SPY and F won’t be tightly correlated so you can expect them not to win at the same time, do take in to account that the strategy itself might not be apt to be used in other instruments and may only be performing well because of the current movement in SPY, you should always back test a strategy extensively to find it weakness before you waste time and money trading it.

    my 0.02$
    #10     Feb 25, 2011