break Out and jack H questins

Discussion in 'Strategy Development' started by madmaxer, Jun 12, 2005.

  1. Hello all Dear ETers.

    I have a couple of questions but I need to clarify a fact before I go to my questions.

    I certainly do not want to make a dead tread alive after 2 years, I just would like to ask your opinion about what I have researched and what I think.

    1- A few months back on ET, I encountered the method by Jack Hershey for Trading ES contracts. I looked at it and tested it on 6 months worth of data -- seems good, some losses, but overall the concept of confirming trend with MACD and Stoc is an interesting concept to me. However I am interested to see if anyone is implementing the same concept for Russell 2000 ( ER2 ). I, personally, think that the same thing with lower MACD should work ( something like MACD histogram around 0.27-0.30 )

    2- My second question is about the Break outs. My general perception of a break out comes from Tony OZ stock trading methodology. I like to utilize them for trading Eminis. What I do is basically look for a relatively big jump or fall with a volume higher than average volume on the 1min or 133 tick charts -- after a consolidation time like ½-2hrs and Continution on the next tick. I was wondering if I am missing anything else in defining a break out. Also, in order to enter a position what other things are essential to be looked before entering the position on BOs.

    I appreciate your advises
  2. answer:

    1. Changing the values on the histogram would be good for any index or for any season of any index once you are under way. Usually back testing is not a workable thing for many reasons. The manner of your personal mental processes overrides any work that can be done on back testing whether the approach is canned or discressionary.

    2. To get to a good effectiveness and efficiency on BO's it is best to concentrate on detecting what comes before BO's. Seeing a BO is too late to make money on a BO.

    People who detect for BO's are really trading whether a BO is going to trend or be a failure to trend. Sorting out thse two conditions is not usually a capability a person has who is watching for BO's. They see BO's as BO's which work and BO's which fail all as one situation. So most people blow being very successful trading BO's because they do not discern between the two possibilities.

    So you may want to change what you are doing and looking for. Once you get past the Tony OZ level of thinking, consider going to the place of considering what comes before a BO and using that to suceed in trading BO's that succeed and BO's that turn into failures.
  3. OK fair enough , but I am looking for some thing more specific than “"considering what comes before a BO and using that to suceed in trading BO's that succeed and BO's that turn into failures””

    I still need to know what do we define as a break out , or if you say look what comes before the BO’s , what area we need to consider, Slow consolidation with low volume ? or is there more out there that I do not see

    Also thanks for comment about MACD but if I want to just simulate your Rocket strategy on ER2. do you thin MACD histogram around .27_.30 is a reasonable number to be substituted with .4 that we use on ES?

  4. mad,

    This is perhaps an optimization question. With the right software, you should be able to verifiy/assertain this for yourself. I might recommend Tradestation and/or Grail software. I'm pretty sure Jack does not have/use any backtesting or optimization software, so I wouldn't expect a definite answer for a parameter setting question like this.

    As far as what comes before a breakout which can tell you whether the impending BO will fail or not, I would not expect a definite answer there as well. Once the breakout occurs, you might be able to watch volume. You could develop rules for reversing from a suspected failure. I just recommend coming up with ideas based on patterns that you see in charts, and find ways to test your ideas. If you don't have software to backtest, like Jack, you'll have to go with other types of paper trading till you come up with a winning percentage, net profit per trade, sample set ect. that you feel confident enough in to test with real money. Search girlpower for an example.

    Good luck.

  5. Hi madmax,

    You put your finger right on where it hurts with Jack: "I still need to know what do we define as a break out..."

    "or is there more out there that I do not see": That's why Jack's disciples always sit in the legendary 95% camp. We badly need these in order to pay for our fat profits.

    I am still puzzled on how you managed to 'backtest profitable' your MACD thing. Take a good look at it once more - don't get stuck with the 95% grouppies, NLP or not. You would be the first one to have come up with this. If in doubt, publish the particulars. Some may be willing to help.

  6. Mad. What comes before a breakout? Nothing. By definition the range is in a tight (unnnnnh!) consolidation, 2 to 4 ticks. It has been that way for longer than it takes to reach a slow orgasm. The volume is so low it sounds like popcorn in the microwave when most of the corn has popped. And price either moved up or down strongly before the consolidation began. Neither price nor volume will tell you shit. What counts is the internal and external market factors you aren't watching. Mike.
  7. OT

    Regarding back testing. There is some bad news. Since the purveyors of such stuff do not know how markets work, it is difficult to provide product to those who, equally, do not know. Actually they do provide stuff and it sells. Thus a significant quagmire is continually supports in more and more ways and depth a mistaken effort. It is one tough situation to expect iterative refinement of an ill chosen concept to ultimately be successful. This is why I believe the railroad industry did not invent wings. I enjoy working contemporarily with the RR industry today, however, on variable voltage EMF efforts to get trains to compete with wings. I was able to ride from Chicago to Albany with one of the lead design engineers. Watch the introduction this Fall on the NE corridor. RR will at last be able to beat the point to point time of travel.
  8. Jack. Re backtesting, the available products are only a trap if you use their canned solutions, crap like conventional TA. I use E-Signal. It allows you to capture tick-by-tick the tick time, the inside market, the last transaction, and its volume. That's the whole market, except for DOM, which IMO is useless. Profit comes from using that data innovatively in your own personal codes. Even though it is written in Java, on a decent machine you can perform very complex operations on that data in real time.

    If I could make only one distinction between you and the B-Team, it would be that we require algorithmically computable rigor in our trading systems. Now such rigorous systems might or might not be practical to test, but they are at least unambiguous. Of the few elements of your system which had sufficient algorithmic specificity to be codable, not a one has tested out with positive expectation. Yet somehow the whole is supposed to be grater than the sum of its parts. Mike.

    BTW, I may be in Tucson consulting soon, and would like to buy you, Kokomo, and Bogart a taco.
  9. Mad,

    Try looking for things that improve the effectiveness and the efficiency of making money. You might consider this *efficiency factor* formula:

    EF = 1 - [(1 - Win%) / Win%] * (Average Lost/Average Win)

    Such formula emphasizes more on Win % and Average Win/Lost rather than a plain Total Win/Total Lost in the PF formula. It "penalizes" systems that have high win % but low $ win/lost ratio and vice versa.

    You can use this when you are backtesting Jack Hershey suggestions for beginners to try. Change parameters/inputs around (such as MACD histogram) to see what's most efficient, or what works at all. Don't forget to use a large sample set, yet reserving some data (called out-of-sample, or OOS) for walk-foreward testing to see if it holds up to the non-optimized data. (Of course, include commision and see what various slippage amounts will do to the data.) You will then be able to look at the data and ask, "what wasn't that"? You can see for yourself if there is any bad news.

    Begin to consider the data as your partner and look at the facts continually to find out what it is contributing to you to make money for you.

  10. I dont know of anyone who makes money using Jack stuff on eminis. Position trading stocks - yes - Spydertrader. Eminis - no.
    #10     Jun 19, 2005