Are we all deluded?

Forums › ProRealTime English forum › General trading discussions › Are we all deluded?

Viewing 9 posts - 46 through 54 (of 54 total)
1 2 3 4
  • 07/01/2018 at 9:52 AM #74988
    Vonasi

    I like a good challenge! Maybe I will move volatility based conditions further up my to-do list.

    Do you have any particular favourite ways of measuring volatility?

    My previous attempts so far have mainly centred around range, ATRP, deviation bands, Vix, averages of high and low as well as general band squeezing and expanding with things like Bollinger bands etc.

    07/01/2018 at 11:41 AM #74995
    Nick, The Tall Guy
    Vonasi wrote:
    Nick, The Tall Guy wrote:

    These longer TF strategies would need to work over months and years

    No they don’t. Just because you are using a daily chart does not mean you have to be in the market for months. You can be in for seconds if your stop and limit orders and the market decides that that is what is going to happen.

    Sorry @Vonasi, I meant that for systems based on longer TF there would be fewer trades initiated in a shorter amount of backtesting, not that trades would be in the market for long periods. In order to generate a meaningful number of trades to increase confidence in backtest results these higher TF strategies need to be tested over much longer times. I like to see a minimum of 100 trades in any backtest to feel any level of comfort, and I would prefer it to be a lot higher.

    07/01/2018 at 12:13 PM #74997
    Vonasi
    Nick, The Tall Guy wrote:

    I like to see a minimum of 100 trades in any backtest to feel any level of comfort

    There are about 300 candles a year give or take on a daily chart if Sunday candles are included. Since the beginning of 2000 there have been 19.5 years so that is 5850 candles approximately to find your 100 trades in. That means that you only need to find a reason to enter the market in 1.7% of the candles. That doesn’t seem too difficult to me.

    I would rather have a lot of trades over a larger period of time instead of a lot of trades in a short period of time because that way I have seen everything that the market has done in say 18 years rather than just what it has done in the last  few months. Now if we could test 18 years worth of data on a 1 minute chart that would be a whole different kettle of fish.

    07/06/2018 at 12:55 PM #75473
    Leo

    Hi all,

    I do not like to test my code on demo because if I test them  during  3, 6 or 9 months, or even years I do not means that the next day I activate them live they work.

    What I do is always testing them with mini contracts then I risk like 20 0r 30 € per trade. If it is working then the next time is entry with a bit more of size. It also have a “quit” if it lost from his Maximum. I know it is time to re-optimize the variable or modify the code, etc.

    I got very late the perspective that exiting a trade is even more important that the entry.  You can see an example with all that included here: https://www.prorealcode.com/prorealtime-trading-strategies/bollinger-bands-contraction-5min-forex/

    I do really believe tha adgo trading can provide us a sort of income and get this finantial liberty we all seek.

    Cheers.

     

    07/06/2018 at 5:10 PM #75488
    Vonasi
    Leo wrote:

    I do not like to test my code on demo because if I test them  during  3, 6 or 9 months, or even years I do not means that the next day I activate them live they work.

    The only way to know truly if a strategy is curve fitted to historical data is to forward test it. You just choose to do it with real money instead of fake money. The result at the end is the same except those who test with fake money have a better chance of holding on to their real money. This is because a strategy that has proven profitable during an IS test and then equally profitable during an OOS test has a far higher probability of continuing that success in the future than one that is only known to be profitable on an IS test .

    1 user thanked author for this post.
    avatar Leo
    07/30/2018 at 2:38 PM #77082
    grimweasel47

    I have a strategy trading on IG Demo that I backtested as much as possible and optimised within certain parameters for each FX Pair that I trade (such as trade time zones etc to turn strategy on/off). One thing to bear in mind with backtests is slippage. In the backtest the system gives you the indicated prices. In the live environment your trades will be susceptible to slippage and no-fills in some cases. This will have further impact on your system. Whilst IG Index have indicated that the demo environment is the same as the live environment, I’m not so sure and I think we should accept lower profit expectations from demo into live.

    My system worked well in backtests so I switched to demo and forward testing then live. Initially the system lost money as the backtest code included the command ‘if onmarket’ which was fine in backtest but not in live trading. This blew a small account (£2k) as I didn’t want to override manually something that I had coded to be systematic. Trust the system and all that!

    Once I removed that command, the system performed well (forward testing) and ran up 100% over 3 months. Then, over the past month it has lost 50% since the PRT update, as I was using a ‘while’ ‘do’ command, which PRT post upgrade did not like. I replaced (thanks to PRT tech team) with ‘if’ ‘then’ command and all is well again.

    Once the system starts trading back into profit on fwd testing I will allocate capital again. All in all, I expect this whole process to have lasted around 12-18 months from inception, coding, testing, live, testing and back to live. There are no short-cuts to easy profits in my mind.

    07/30/2018 at 2:59 PM #77086
    Vonasi
    grimweasel47 wrote:

    Initially the system lost money as the backtest code included the command ‘if onmarket’ which was fine in backtest but not in live trading.

    Good to hear a story with a good outcome even if it was a good then bad then good then not quite so good story!

    The only bit I do not understand is the ‘if onmarket’ issue. This is a major condition to be the difference between working in a back test and forward demo test and then not working in live. Are you able to explain it any more?

    07/30/2018 at 3:00 PM #77087
    GraHal
    grimweasel47 wrote:

    the backtest code included the command ‘if onmarket’ which was fine in backtest but not in live trading.

    So should we cease using If Onmarket  … seems no point if it gives problems when running Live??

    What does If Onmarket do or not do in Live running anybody??

     

    07/30/2018 at 3:06 PM #77088
    Vonasi
    GraHal wrote:

    What does If Onmarket do or not do in Live running anybody??

    It should just be a condition for whether a trade is open or not. At the close of a candle I assume that it just checks the countofposition and if it is different to zero then onmarket = true.

    Very useful for having one set of conditions for a first entry on the market and a slightly different set for any further positions to be opened.

    For example:

    I’ve never had an issue with it in demo or live which is why I asked as well.

     

    1 user thanked author for this post.
    avatar GraHal
  • Author
    Posts
Viewing 9 posts - 46 through 54 (of 54 total)
1 2 3 4

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
help with optimazation of code and we need a snippet on all things optimazation
all custom indicators return zero!
are partial fills possible?
collapse trading toolbar all charts
having a list of (xls, txt) of all the equity options that offer prorealtime?
hide all objects drawn
my eyes are burning staring at this code all night help please!
sorry, we cannot log you in – ig

Create your free account now and post your request to benefit from the help of the community
Register or Login