Not logged in
Viewing 15 posts - 1 through 15 (of 21 total)
  • #241359|
    Customer
    473 Posts

    notch wrote:
    … was built in 2004-5 and has maintained its performance:
    The length of time I’ve been doing this is worth emphasizing. I invested in my first algo’ builder in the year 2000 (I actually think it was 1998 but the memory fades because I#m not as young as you may think) called TradingSolutions and Neurosolutions. I first started performing walk forward (rolling) tests in 2002 before it went mainstream. I am formally educated in statistics, computer science, A.I. , finance, economics and to top it off Accounting – formally: UG, PG & PhD. Whether you decide to believe it is unimportant however I only offer opinions I KNOW work based on 20 years of research and practice. So apologies if occasionally I brush off remarks such as the P-value comments because much of it is genuinely elementary stuff for me. I do not write this to be better, I just don’t want to be constantly flaming the innocent folk of the forums. It is tiring for all. You don’t have to agree but if I take the time to offer an opinion, it is because it works. There is zero guesswork at play here. Zero. However, I still cannot build scalpers!!!! p.s. I cannot believe how well I profited from that GBPUSD 100 pip upward spike especially starting from a short position!!! ;) I should start my own thread called “Notch’s rambling”. lol

     

    A thread started to prevent hi-Jacory…a crime I am most guilty of committing.

     

    I’ll use this thread as a place to share my thoughts mostly on SQ but sometimes trading generally.

     

    This thread will also serve as a question and answer location.  So if you have an SQ or model building question that can be condensed into a short sentence feel free to ask.

     

    I may take interesting posts from other thread and post them here just to have one location of the discussions I find the most interesting and rewarding.

     

     

    #241473
    Customer
    473 Posts

    I had a good day in the GBPUSD manual trading trenches; bought the low and sold what I hope is the high (trade sell running).

     

    SQ wise, I didn’t learn anything new.  However, I had a number of things reinforced:

    1. Support and resistance reign supreme;

    2. buying tick volume and selling tick volume are the closest things to a crystal ball in Forex trading.

     

    What does this mean for SQ model development:

    Serious and profitable manual traders are always aware of the location of key support & resistance areas.  The most common of these is

    a. D1 O,H,L,C & (H+L)/2;

    b. W1 O,H,L,C & (H+L)/2;

    c. MN1 O,H,L,C & (H+L)/2;

    d. Floor trader Pivots or simply Pivots in SQ;

    e. simple moving averages 50, 100,200,500,1000

    f. do not ignore tick volume.  Tick volume inclusion generates some surmountable challenges which you can debate between yourselves; just leave me out of that debate.  Volume is an essential component for accurate forex model development.

    g. raw price trumps all indicators 5 days of the trading week, followed by simple moving averages.  By now you know monte carlo with replacement (random) can be used to generate p values or in other words, P-value 0.01 basically means the models passing an MC at the 99%  significance level and P-value 0.05 means models passing at the 95% significance level.  Raw prices are noisy so use MC to remove the most contaminated.

    Please note the MC assumes a Gaussian distribution and price data is non-stationary (non-gaussian) therefore you will still have to wrestle with many false positives (shit models) until you have mastered more advanced techniques to remove the feces.

     

     

     

     

    #241477
    Customer
    473 Posts

    On the subject of poor/bad models:

    The beauty of a model is in the eye of the beholder.  However, the selection of a model based on the aesthetic qualities of the equity curve and the resulting performance stats doesn’t make sense to me.

    Most users who have more than 12 month’s SQ experience and have deployed live models must realize the chances of obtaining a model that mirrors the perfect equity curve in live trading probably has a P-Value of 0.001.  Why does this even matter?  It matters because when you combine those models as a portfolio in QA, the portfolio performance in live trading is a complete fantasy (as many experienced users now know).

    Give me a backtest performance that reflects the prevailing economic events at certain points in time over a ruler-straight equity curve.  At least that way you actually may produce something of value when combining strategies into a portfolio.

    #241488
    Customer
    473 Posts

    Taken the first 3% from the GBPUSD from 2662 to 2684.  3 pip SL.

    Please with it but wanted to hold to 2711 but then I realized there is high impact news in just over an hour and there is nothing like high impact news to get in the way of the ‘near perfect’ trading plan.  Price is now falling back to the point of control and may even fall to the initial balance.  Trading from the initial balance seems to be a straightforward way to make a few %.

    Question is: how the hell do I model ‘time at price’ in SQ?  A question I’ve had since starting to use the software.

    Or even better: volume at price!   These are the features that would deliver extreme modeling value and relatively few false positives.

    #241489
    Customer
    473 Posts

    Price fell back to the POC at 2664 then immediately moved higher…great support the POC.  I managed to take a small bite long at 2668.  SL 10 pips which is very large for me so not feeling great about this entry because I’ve overpaid by 4 pips.  However, the price has move immediately into profit so it looks as though I may score a few before the news show.  I’ll close 30 mins before the news.

     

    I wonder if we could incorporate high impact news modeling in SQ.

     

    Note: volume rarely lies.  Buying volume is 3 times higher than the selling volume.

    I wonder if SQ can create a way to separate buying from selling volume.

    • This reply was modified 6 months, 3 weeks ago by  notch.
    #241491
    Customer
    473 Posts

    Closed for 5 pips.  Not much but a profit is a profit; I suddenly remembered the cardinal rule of support and resistance: the price is going to test it at least twice.  The price will be rejected on the first attempt, it is the 2nd test that counts.

     

    I wonder if there is a way for SQ to enter a stop or limit order after the 2nd attempt.

    I have 10 more minutes to reenter if the POC is retested; if no retest I’ll call it a day for manual trading because I’ve bagged enough % and I’ll move on to some grid EA work; I’ve discovered a way to prevent grids from adding additional positions until the greatest likelihood of the price moving in the correct direction thus overcoming one of the greatest weaknesses of the grid: being caught on the wrong side in a strong trend.

    #241497
    Customer
    473 Posts

    Please with it but wanted to hold to 2711…

    GBPUSD price is rapidly approaching the 2711 mark without me on it!!!  However, trading before high impact news for me is a no-no even if it means I miss out on double-digit account growth from a near perfect trade plan!  If you are near certain where the price is heading one can naturally load on the lots.

     

    Price should fall between 2655 to 2665

     

    #241498
    Customer
    473 Posts

    Grid on the TDS backtester!!!

     

    Price should fall between 2655 to 2665

     

    There’s no point writing I’m an expert when I can just show it!

     

     

    P.S.

    Closed for 5 pips. Not much but a profit is a profit; I suddenly remembered the cardinal rule of support and resistance: the price is going to test it at least twice. 

     

    We got the double touch of 2664, I just didn’t think it would hit 2710 first!!!  Now for a double touch of 2710!!!  Fingers crossed.

     

    Who the hell isn’t going to sit at the computer when they can read price like that?

    #241499
    Customer
    473 Posts

    P.P.S.   I only ever give an opinion I KNOW works.

    #241502
    Customer
    473 Posts

    Surprise 2710 double test and more profits are taken.

     

    I’m expecting price to move much higher but need 2711 to become support and be tested twice from above.

     

    Generational wealth trading at its finest!

    #241504
    Customer
    80 Posts

    I wonder if we could incorporate high impact news modeling in SQ.

    Technically this should be very easy to implement…I wonder why we do not have this feature yet. Did you already make a new feature request ticket for this one?

     

    #241505
    Customer
    473 Posts

    I wonder if we could incorporate high impact news modeling in SQ.

    Technically this should be very easy to implement…I wonder why we do not have this feature yet. Did you already make a new feature request ticket for this one?

     

    Shame on me.  I haven’t requested it yet.  I will add a request.

    #241506
    Customer
    473 Posts

    Took a short just below 2711.

    I can’t allow what I believe will happen to get in the way of what is actually occurring.  I’m targetting 2655 but may attempt to hold for longer in case price moves towards yesterday’s LOW, I also have a buy stop above the HIGH in case price suddenly breaks out to the upside.

     

    Volume related decision.  Closed short for a few and switched to long:

    #241507
    Customer
    391 Posts

    Nice. I Have 2 live GBP long from bottom and one Short from the top . did not even look at the chart  :)

    #241509
    Customer
    473 Posts

    Nice. I Have 2 live GBP long from bottom and one Short from the top . did not even look at the chart :)

     

    Yes mate and I sold the EURUSD in 2007 at 1.59841, closed it at  1.050548 in 2016 then entered long at that price and I’m still holding.  All trades by flipping a coin and reading my Tarot cards.  :)

     

    But seriously Mabi, you are referring to algo trades.  manual trading I use stop-losses of 2 to 10 pips whereas with an algo you’ll be using much larger stops meaning for 1% risk I am able to trade a much larger position size.

     

    1. Stopped out the tick volume long trade for a few.

Viewing 15 posts - 1 through 15 (of 21 total)

You must be logged in to reply to this topic.