Errors in the Hershey Futures Method

Discussion in 'Strategy Development' started by Joe Doaks, Jul 22, 2007.


    Those of you who chart simultaneously in multiple time frames, as I do, recognize that trend lines and the channels drawn therefrom often look significantly different in the different time frames. This is because the higher the time frame, the greater the uncertainty in where within the bar the price extremes occurred with which the trend line and the channel are constructed.

    For example, in the Hershey SCT five-minute charting system, the price extreme in the bar could have occurred anywhere between the first and the fifth minute, a potential uncertainty of four minutes. The two points used to draw a trend line therefore have a potential uncertainty of eight minutes. Similarly the uncertainty in the timing of a potential FTT is four minutes, making the total uncertainty in the FTT construction timing twelve minutes.

    Now this is not much of an issue with channels with closely spaced highs and lows, of the type you typically see in midday bull or bear flags. But it can be a significant source of uncertainty in long morning or afternoon runs. The result is clearly and frequently witnessed in Spydertrader's thread where there is much lamentation about erroneously declared FTTs.

    You can verify this issue yourself by drawing trend lines and channels in different time frames, for example, five minute vs. thirty minute, and noting how a channel bounce may look like an FTT or vice versa. Or a channel breakout may look like a bounce, or vice versa. What is the solution? Regardless of the time frame in whcih you make your trading decisions, always trend and channel chart intraday in one minute.

    Now the Hershey students will retort that their "gaussians" (that regrettably imprecise corruption of a precisely defined word) save them from the errors of which I speak. If the responses to this thread are civil and serious, I will explain why that assumption is fraught with error as well.
  2. It's good to see an analytical critique of the method.
  3. Thank you, 225. As you well know, my major difficulty with the method stems from that very analyticality. As Spydertrader said early on in his Journals thread "There will be no hard and fast rules here" (apologies if I misquote or inacurately paraphrase). This prevents the method from being coded into the unambiguous binary decision (redlight/green light) which the ordinary trader needs from a non-discretionary trading system. That absence makes SCT essentially discretionary, and therefore not really a system at all.
  4. I don't think that Jack or Spyder ever called it a system. I don't consider it a system. It's an organizational paradigm(at least that's how I look at it).
  5. <i>"That absence makes SCT essentially discretionary, and therefore not really a system at all."</i>

    Is the title of this thread, "Errors In the Hershey Futures <b>System</b>?
  6. Gentlemen, your points are well taken. But what else would you call what SCT purports to be than a system? There are two extremes in approaches to trading. One is the intuitive approach, wherein the practitioner charts those instuments and price/volume functions which he considers to be relevant, and makes "expert" intuitive gestalt entry and exit decisions not constrained by absolute values of those functions or of relationships between instruments and/or functions.

    I stand (admittedly imperfectly) for the other extreme which requires either a single rule-based binary entry decision or an aggregation of them (red light/green light). I think the former approach is overly egotistical. They might say I am overly analytical. In brief, to me a system uses hard-and-fast rules to make unambiguous trading decisions. From that viewpoint, SCT, or Spydertrader's version of it, is no system. Neither here nor there, perhaps, likely a matter of taste, but you asked.

    But I await comments specific to the charge I made that FTT identification based on five minute channel construction has inherent and unavoidable uncertainty.
  7. Jander


    Always happy to carry on a civilized debate, just not sure where to begin here. You went into great detail about how differing timeframes can make your channels look skewed, but what does that have to do with this methodology? We use the 5min ES and do not switch to different timeframes.

    Also, it doesnt matter when during a bar an ftt/channel bounce occurs. Decisions are made intrabar based on projected volume and end of bar readings only serve to reinforce/adjust what you had predicted. The lamentation you speak of has nothing to do with the assumptions you have made, in my opinion.
  8. nkhoi

    nkhoi Moderator

    joe is not interest in proflogic, none that I recall anyway.
  9. Your first ten words are not a good idea. This may be an invention or extrapolation of yours. So drop the first ten words and begin your thought process again.

    Anyone can freely change whatever they want. In your case you did the change and you now have something that is kinda flakey here and there.

    A lot of people are smarter than average. They often approach things in ways that may be oriented to getting gold stars or something. In trading the bottom line is what counts.

    Your invention stated in the first ten words introduces for you (and not for me, by any means) uncertainty. Too bad for you and too bad for anyone who mimics your reasoning process.

    FTT's occur in triples on dominant concurrent traverses of each of the three monitored levels. On, non dominant traverses of each level they appear as singletons.

    If you reflect, the reason the binary elements of the data sets are used is for one reason: to eliminate uncertainty. Why would a person who was independently improving SCT with an invention such as yours or such as anyone else's, go to the trouble of destroying the MADA routine in the process.

    Look, by googling or something, at the primary thing Quants do. They work in support of the corporate book of their employer. The majority of the time they invent instruments and tools in support of getting clients to end the use of prior products and tools and, then, to convert to the new party line of the sales force of the corporation and its corporate book.

    What you have done, and well below the quant level of thinking, is invent a tool that will drive the clients away. Think like a quant from now on and do iterative refinement on what you know and use. You can't or don't use SCT so don't bury it with mistaken inventions.

    What would a person running 10,000 contracts with SCT do? See, he sure as heck doesn't go to multiple time frames. What does he do instead? Think for a moment and see if you can get on some rational thinking track.

    We all love the humor. If that is what you like to do, do humor. Entertain us. You are not making the cut as of now re: TA.
    #10     Jul 22, 2007