Profitable strategy that work on any market

Forums › ProRealTime English forum › General trading discussions › Profitable strategy that work on any market

Viewing 15 posts - 1 through 15 (of 74 total)
1 2 3 4 5
  • 07/24/2017 at 1:58 PM #41501
    juanj

    I suppose we have all searched for, or at least though about the idea of a strategy that could be applied to any market (regardless of the spread) and be profitable.

    So I want to put a challenge out here to create such a strategy:

    I have myself been trying to develop such a strategy for quite some time and think I might of finally found one. And I will be sharing it here, but first I want some participation.

    My criteria for such a strategy is simple:

    1. The strategy must remain completely un-optimized (i.e. other than the spread all code must be static) – average spread have to be taken into account.
    2. The strategy does not have to outperform buy and hold or have exceptional returns but it has to be in the blue regardless whether the market ends in the blue or not.
    3. The strategy must work on any market (Forex pairs, Indexes, etc.)

    Some pointers from my experience:

    1. Obviously less is going to be more, simple high probability triggers and modest exit criteria.
    2. Indicators are useful but keep it basic (i.e. RSI[2],  200MA)
    3. Smart trade management

    What I was able to achieve so far is a market neutral strategy that could remain profitable in all of the major FX pairs as well as the US500 and other major indexes.

    07/24/2017 at 9:03 PM #41529
    Eric

    are we talking about daily tf?

     

    07/24/2017 at 9:25 PM #41530
    juanj

    Timeframe does not really matter, was thinking along the lines of 5min to Daily. But then again the idea is that it must work on any time frame.

    Have done some tests on the 5min using my strategy (originally tested on 1Hr) and it does not perform well unoptimized in that arena (allot of noise on the lower time frames).

    So I would say 1Hr to Daily is more realistic.

    07/25/2017 at 8:29 AM #41539
    Pepsmile

    Excellent (probably idealistic) topic of interest to all.
    Some markets have specific behaviours, oscillation (Forex), directional (shares/indices), quickly reach the objectives, etc. What could be the good filters that works ?
    Intraday tests (excluding spreads first) are more efficient due to the availability of the length of the history and the range of the range of patterns, etc.
    You should identified an overall acceptable ratio of success for this strategy, ie by market/timeframe, etc.
    At all events, your topic could give good ideas to define appropriate rules to build robust strategies.

    07/26/2017 at 8:13 AM #41626
    Wing

    The best way do to this I think is with a “self-learning” system that adapts to market conditions. Machine learning and such are obviously not simple, but if you truly want something universal I believe that is the way.

    07/26/2017 at 6:44 PM #41719
    juanj

    @Wing I agree with what you are saying with regards to the machine learning algorithm (a.k.a. genetic algorithms), however unfortunately that is out of scope for this forum as ProRealTime does not support the type of functions required for such an strategy (i.e neural networks). However I have managed to write a simple strategy quite some time ago that utilizes a very simple ‘pseudo random’ approach to determine the direction of the trade and then using slightly more advanced position sizing and exit criteria based on ‘smart’ variables that are automatically adjusted (incremented) based on previous trade performance. The strategy ran on a 1min timeframe as variables could only be adjusted after each trade and depending on the trade performance. The strategy showed some potential but I abandoned the idea after feeling ProRealTime is not the right platform for this type of strategy. Maybe sometime in the near future we can open a thread on the topic and I will share some of my work. Would be interesting to know if @Nicolas has tried venturing into something similar using ProRealTime?

    1 user thanked author for this post.
    avatar GraHal
    07/30/2017 at 9:29 PM #42034
    juanj

    I have to say it is kind-off sad that nobody is contributing to this?

    I realize there is allot going on throughout the forum but I believe this is something that could add value to all aspects of our strategy building endeavors.

    I have continued to work on this concept myself and I now have a strategy that is profitable on all of the below completely un-optimized. With only trading hours and spread being adjusted.

    US500, FTSE, DOW, DAX, ZAF, OMX, EUR/USD, AUD/USD, GBP/USD, EUR/GBP (have not yet tested all the others, but I am sure they will yield the same outcome)

    As promised, I will share this strategy on the library soon enough. But I want to encourage you to look into this yourselves as well, reason being I would like to see different approaches to the one I am using.

    5 users thanked author for this post.
    avatar Marlon, avatar Juan Salas, avatar Yannick, avatar Wing, avatar Pepsmile
    07/31/2017 at 1:01 AM #42043
    Wing

    Good to hear. I think one reason people do not contribute is that they don’t believe in the idea, or lack the knowledge to understand how a system can adapt.

    Like you I have made a self-learning/adaptive system in PRT long ago, so if you post yours it will be interesting to compare. I never posted that system to the site since it is a complex program with a lot of indicators, and performance has been only break-even for it during this year.

    Even if you don’t post it now, it would be interesting to share the time-frame, average trades/day, time-in-market, etc.

    As for the system being un-optimized, how many moving parts does it have? How many variables have you set yourself? My system suffers a bit from being too complex, and is therefore less reliable than something simpler.

    1 user thanked author for this post.
    avatar Juan Salas
    07/31/2017 at 6:14 AM #42048
    Inertia

    Hi JuanJ,

    I am sure that your very interesting topic has been red by mostly all of us ! From my side, a straddle strategy could theorically work in any instruments…

    07/31/2017 at 6:33 AM #42049
    Nicolas

    To answer @juanj question about machine learning with ProBuilder language, the lack of arrays has pushed me to not investigate further into my own ideas. The fact is that I believe pattern recognition / machine learning is effective, stable and more accurate with non-time dependent timeframe such as tick charts, but we also all know that ProOrder is not compatible with this kind of chart. The lack of time also plays enormously against me unfortunately 🙁

    I used to play with Perceptron (a single layer neural network) on another platform, the formula is simple and can be feed with any indicators or prices arrays. I don’t know where it would lead us or even it would be useful for your ‘universal strategy’ project, but I may post the adapted code for ProBuilder if you ask.

    07/31/2017 at 8:41 AM #42053
    juanj

    @wing I have decided to go with a 1hour time-frame as to find a balance between market exposure in terms of signals vs. noise. Another interesting fact is that I have made use of Fibonacci and Ichimoku number sequences (which pleasantly surprised me as to their market neutrality). I also made use of a RSI 2 period and ATR 2 period along with a 100MA. Then of-course I added one last ingredient which acted as my trigger (which I will disclose when I post my code). Initially I did not add a trading time filter (to keep it completely market neutral) but decided to add it in as I were catching allot of false signals in the after-market hours, making it less effective on some indices and forex pairs. Actually adding the trading time filter made it profitable on every market I tested it on. Also quite interesting is that it outperforms a buy and hold approach on the greater majority of markets tested while always making the equity curve less volatile in terms of draw-down.

    07/31/2017 at 9:57 AM #42066
    Nicolas

    Ichimoku is “market neutral” as you stated because it is only built with highest high / lowest low.

    07/31/2017 at 11:26 AM #42080
    juanj

    @nicolas that is correct although I didn’t actually use any Ichimoku based techniques in this strategy. Just the number theory. Goichi Hosada (founder of Ichimoku)  in his development of ichimoku, spent 4.5yrs of his study just on number theory. 9, 17, 26, 33, 42, 65, 76, 129, 172

    07/31/2017 at 12:35 PM #42085
    Nicolas

    That represent a lot of time spent for studying only 9 numbers 🙂

    07/31/2017 at 1:03 PM #42095
    juanj

    @nicolas I think rather that the 9 numbers is the result of the time spent rather than the focus. It is interesting though if you notice how often numbers like 9 and 26 for instance comes up as optimum numbers in optimization. I used 42 as one of the variables of my strategy that just so happens to be the best suited variable between the different markets.

  • Author
    Posts
Viewing 15 posts - 1 through 15 (of 74 total)
1 2 3 4 5

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
array/bubble sort in strategy doesn’t work, but works as an indicator
buy on certin dates, not on technical singal
buy on retest on a certain candle
buy or sell at set time and price change on any market.
do the market replay reports actually work?
exchange time on market
exit within the next 4 bars if higher than entry price on market
how to get any previous open market day in yyyymmdd format?
is it true that you can’t invest in eu stock market shares with autotradin
market cap on prt

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