Time in Market

Discussion in 'Strategy Development' started by nonlinear5, May 8, 2011.

  1. Let's say you have two strategies, A and B. Their profiles (as measured by profit factor, net profit, max DD, Sharpe's ratio, and number of trades) is identical. The only difference between A and B is the average time in trade. Average time in the trade for strategy A is 4 hours, and average time in trade for strategy B is 1 hour. I think most people would agree that strategy B is superior to strategy A, since it takes profits and losses quicker, thus potentially diminishing the probability of "getting stuck" in adverse events.

    My question is, how would you incorporate the average time in trade into your "strategy quality" equation? My instinct tells me to divide whatever your "strategy quality" by the square root of the average time in market, and compare the results. For example, for the original strategies A and B, if we choose profit factor as the "strategy quality" number, the adjusted quality would be:

    for strategy A: PF / sqrt(4)
    for strategy B: PF / sqrt(1)

    As can be seen, the adjusted quality for strategy B would evaluate to be twice that of strategy A.

    What do you think?
  2. You can only truly compare the "benefit" of the longer duration strategy if the other strategy is not present.... so the absence of the shorter strategy confers a benefit to the longer term strategy if you think of your allocations in terms of:

    if this event is happening.. --> do this.. type of statements..

    in other words, your trading becomes contingent on trade events and the allocation and trades themselves take on a kind of structure that suits your models assumptions about the price action. In a case like that you can't reduce your trade signals into values comparing one trade against another like that.

    .. Also the timing within that timeframe itself ceases to be part of your trade valuing technique for the reasons mentioned above.

    If you cannot isolate the shorter strategy consistently, than you can infer that the trade value is no different than the longer term trade, because you weren't able to identify timing parameters to distinguish it and take advantage of the short term trade. But you wouldnt be doing this anyways.. giving that to value trades in an event type scenario would require a different technique than "valuing" them in a strategy quality kind of way.

    If you actually do have the ability to isolate the trade in the shorter timeframe it then becomes subject to other considerations as well- like liquidity, which when trading in large enough size, a price action model would have to incorporate that as well.