Optimizing "Time" as a parameter

Discussion in 'Strategy Development' started by logic_man, Jun 1, 2012.

  1. Question for the traders who have, in some way or another, worked time into their trading models. More specifically, this is not about cycles, but about actual "clock time". For example, in the ACD method, the way you "confirm" a trade opportunity is if the asset remains below or above a specific price for a specific amount of time. That's the kind of time I am talking about. After all, isn't that what traders are trying to do, "time the market"?

    Anyway, the question is, if you have worked time into your models, have you found that the optimization of that time parameter has been more robust than optimizations for other parameters?

    Sticking with the ACD method example, the guy who developed it recommends waiting at least the same amount of time as the opening range you use as your confirmation period. So, if your opening range is 15 minutes, your confirmation period is 15 minutes. At least that's what I remember, it may be a little off on the exact amount. But, the point is, it's an exact amount of time. One would assume that this value shouldn't ever change, right? It would never be better, i.e. more profitable, to wait half the time of the opening range or twice the time of the opening range, for example. Presumably because if you wait half the time, you get more "false positives" and if you wait twice the time, price will move away from you too often and you get bad entries. Also, you would think that you could test the time period and determine that it's actually 75% of the opening range that is the best amount of confirmation time and that would never (or extremely rarely) change.

    That's my theoretical thinking about time as a optimizable parameter, but if there are actual experiences with this sort of optimization, I'd be interested to hear anything you'd be willing to share.
  2. ===============
    1]Yes,Above/below price, for 2oo days;
    above[uptrend] or below [downtrend] 50 day moving average.

    Many experienced traders are not so much timing the market ;
    more like time in the market..........................Even the more profitable daytraders,LOL -look @ how much time Don Bright Trading has been in market.LOL.. But see what i mean??

    Sure. Long story short i think it was Bill Brodsky who said ''innovation deserves more than 15 minutes of fame.'' We have to say amen to that...............................................................:cool:
    5 minute chart is noise/static
  3. Using your 50 above 200 day example, what I am asking about is more along the lines of optimizing the number of days the 50 has to stay above the 200 before it confirms an up or down trend.

    To me, 50 and 200 are completely arbitrary, but if you were to do the analysis and show that if the 50 stays above the 200 for five consecutive closes, buying at the open on the sixth day and holding until the 50 goes back under the 200 was profitable after costs, then that would be something similar to what I am asking about. Not that particular rule, but a rule of that type because the optimization of adding in the qualifier "for five consecutive closes" adds objectivity to the analysis by qualifying it with an optimized time parameter.
  4. jcl


    Most optimized parameters in strategies are time dependent, f.i. the bar width, or a filter constant. At least in my strategies I did not find any more or less robustness in time dependent parameters than in others.
  5. When you say bar width is "time dependent", do you mean that you would optimize to find the optimal bar width or that you would optimize to find the optimal number of bars of a pre-defined width?

    My question is oriented toward the first kind of optimization, since I am saying that rather than take any time-related parameter value as a given, I would want the freedom to use time however I saw fit.

    Going back to the ACD method, the guy who developed it uses a 15 minute opening range, but he had the freedom to use a 25 minute opening range or a 10 minute. You could call his opening range 1 15-minute bar or 3 5-minute bars, but the more important factor is that it is 15 minutes of actual time. During that time, price can move any amount it wants, but once that time period is over, it sets an objective benchmark for the trade set-up. I assume the guy uses 15 precisely because the traders who set the opening range are nearly always done with their trades within 15 minutes and that from the 16th minute onward, all other trades are basically reactions to those first 15 minutes and this can be shown empirically.

    Then, my question is whether if he optimized it to be 14 or 16 minutes, would that optimization be more likely to work out of sample than an optimization that wasn't based on clock time or an optimization of some parameter which is time-dependent, but not time itself? If you could actually determine what it is, why would the time period used to set the opening range ever change?

    And if the answer is that it wouldn't ever change, are there other ways of using time which could achieve the same level of constancy through all market conditions?
  6. hey man, I have no business on this thread. I don't even know what you guys are talking about. But time is the most important factor for me in analyzing a market.

    How much can it move in a day?

    How long did it take to get from price A to price B?

    How much has it moved in an hour?

    How much can I make if I hold a bond to maturity?

    Is it underperforming or outperforming time?

    How much time do I have before it turns?

    Nobody can beat time, so if I am, maybe I should get flat.
  7. I agree that time is the most important factor in trading. Yet, there are threads about channels, trendlines, Fib numbers, breakouts/breakdowns, etc. that have tons of pages, but no threads about time of a similar size. At least not one that I've seen.