Some thoughts on the Library of Strategies

Forums › ProRealTime English forum › General trading discussions › Some thoughts on the Library of Strategies

Viewing 15 posts - 16 through 30 (of 32 total)
1 2 3
  • 06/13/2017 at 2:58 PM #38224
    Nicolas

    @Eric

    There are more than an hundred strategies in the library 🙂 Yes, seems quite impossible! While some are not optimized, some are and would need re-optimisation from time to time ..


    @Noobywan

    Even if I shared the same thoughts as you, I must say that everyone is responsible to trade a strategy or not with his own money. Everyone must analyse first a strategy with backtest(s), WF if needed, and even demo trading if he wanted to, before going to live trading. Harmonizing all backtests with the same starting capital, same manner of analyzing drawdown% or position sizing would say that every strategy should act the same and you know just like me that this is not the case. For instance, how could we compare the profitability of a strategy like the well-known “Pathfinder” (which accumulate orders and change position size according to the statistical seasonality of the traded instrument) with a common double MA cross with fixed LOT/SL/TP? While one would benefit of this comparison, the other one would become an uninteresting one.

    Since the beginning of the website and with all the strategies shared, we always tried to fought the overfit phenomena with different approach: explanation on forums (so many..), blog articles, FAQ about WF, MC analysis, .. Now that we have tool in the platform to test if we are not over-optimized (more to come hopefully), we may continue with this approach and that’s why using the ranking scale, manel was talking about , is IMHO a good idea.

    Other ideas are welcome 🙂

    06/18/2017 at 11:35 AM #38539
    Derek

    This is so important! It is a step towards ProRealCode.com becoming even more of a knowledgebase.

    Tag navigation is often found on community sites that have to manage wildly different perceptions for the same content from different users. For example is Band XY a rock band or a pop band? We have something similar here. Maybe some members want to tell their buddies that a strategy is overoptimized, maybe some members think it is a clever adaption of a traditional approach.

    The proposition is to utilize Tags more by incentivizing the community to do so. From this spawns a second layer of navigation. If it is possible to easily see how many times a post has received the same Tag there is an order to the new navigation. No need for any further calculations.

    This means that everybody can and should tag quite regularly. How about giving points for contributing a certain amount of tags? Or, we already have quite a few badges. How about one for tagging?

    If it works we end up with some type of crowd knowledge index for the contents of this website. Could be helpfull to resolve quite a few problems when trying to move unrelated forum topics into a navigation.

     

    06/18/2017 at 1:31 PM #38544
    GraHal

    Well there’s a different slant on it, thank you Derek for your lateral thinking!

    Would this mean loads more work for Nicolas? Do we already have the tag option on this site?

    What is the insert read more tag (next to Toolbar Toggle) supposed to be used for on this site?

    06/18/2017 at 5:20 PM #38558
    Derek

    Tags are a standard WordPress feature. When submitting a strategy the poster can select some and you see them under the strategy name in the grid view of the strategies.

    How much effort depends on the capabilities of the theme. I don’t expect it to be a biggie but assume that Nicolas can answer better what the starting point is.

    06/18/2017 at 7:25 PM #38565
    Nicolas

    I can code almost anything for the website, even if it’s not already in the WordPress API.

    Tags are already used in the Library, but not like the way you think about, authors can add them when they post  just like they could use them when they ‘tweet’.

    A separate tag system could of course be added, but if I understand correctly it would be like a voting system?

    06/18/2017 at 8:23 PM #38567
    manel

    I agree with Nicolas re. his point about using the WF module more in order to differentiate strategies better. Since it’s new, it will take some time for users to adopt this regularly but it’s something that should be encouraged in order to achieve the ranking system we are discussing here. If we were to put forward as a community a list of standard components that contribute to overall ranking of a strategy then once the more experienced posters start to use this methodology then hopefully others will follow when posting their strategies. It’s not easy definitely but the first push is always the hardest 🙂

    With regards to this I would also suggest that in PRT we need a “Points Profit/Loss” metric in backtests and results, not just £/$/Eur monetary value. As Nicolas and noobywan mentioned earlier, there are numerous strategies that adopt increasing size, position management etc in their code. Stripping out the effect of these allows you to analyse the underlying profitability of the technical strategy separately from the scaling/money management portion. This also allows comparison across markets by removing the monetary value of a point in each contract eg. a mini contract vs a standard contract vs a £1 contract, and across asset classes eg comparing an index vs a commodity would be easier. It also removes the issue of comparing results between strategies with different starting capital as some use 5k, 10k, 25k etc because we would also have the draw as a points (and a % of points) number not just a % of value number so we would not have to “force” people to use the same starting capital. This allows different users to choose a strategy and use the contracts and starting capital that best suits their risk profile.

    The end result being, if we are allowed to compare strategies easily it would engage users to look into deeper into each strategy to see how/why/if one gives a better £ profit than another etc etc which is one of the aims of this forum I think, to try and understand and improve strategies. Eventually I think we would have a rating (see my earlier recommendation) incorporating a points profit parameter and position management parameter – I have some additional thoughts on how we could incorporate both into a single rating. But critically we need this points based system I feel to get started because as we know, if we don’t get the technical entry points correct, the position management of the strategy becomes irrelevant.

    Nicolas – apologies, this suggestion probably belongs in a PRT development thread, but I felt it relevant here and hopefully if people see the value of this in comparing strategies then you may see fit to try and get it “fast tracked” by the software team.

    2 users thanked author for this post.
    avatar Nicolas, avatar GraHal
    06/18/2017 at 8:41 PM #38569
    Derek

    Every user could have the option to tag a strategy, an indicator or a topic.

    Depending on how often a post has been tagged with the same tag the more important it becomes, so it will show up higher when looking for all posts with this specific tag.

    Hope this makes sense.

    06/18/2017 at 9:16 PM #38572
    Eric

    Now that time has passed since many strategies been added to library it would be possible to backtest them on new data to see performance

    if every member should test 1 system and share the result on the forum instead of every member testing 100 system it would be a improvement, but i suppose its wishful thinking?

     

     

    1 user thanked author for this post.
    avatar GraHal
    06/18/2017 at 10:38 PM #38573
    Eric

    and by every member i meant active members..

    and after the backtest on new data then do a walk forward optimazition

    of course the best thing would be if the creator/author of the strategy did it but they have done their job and others can help

    06/19/2017 at 10:25 AM #38582
    GraHal

    Yes  it would be fantastic if all Strategies could be backested and fitted into the categories or whatever method is decided upon. We then have a baseline.

    As you say Eric, if we split the task between volunteer members, but maybe just backtest and walk forward the latest version of each Strategy … can be hard enough in itself to find latest version? This means anybody should be able to backtest and walkforwrd test and categorise as no need to get involved with the code at all? This task may therefore actually get completed as it is not too optimistic?

    Optimisation could be left to the original code author and / or  skilled coders who understand the code and can be trusted to ‘get it right’? Unskilled code authors may screw up or change conditions without even realising!?

    I like your  tags idea also Derek  … please can we have everything!!??  🙂 😉

    GraHal

     

    06/19/2017 at 10:38 AM #38584
    Nicolas
    manel wrote:

    I would also suggest that in PRT we need a “Points Profit/Loss” metric in backtests

    The next probacktest generation will have it for sure (already discussed as a part of a next update with PRT). Would be definitely more constructive to build serious strategies around risk instead of profit 🙂

    manel wrote:

    I have some additional thoughts on how we could incorporate both into a single rating.

    Please share them with us, even if we do not already have the points profit in backtests.

    Derek wrote:

    Every user could have the option to tag a strategy, an indicator or a topic.

    Could you explain what you think about a “tag”? a simple example would enlighten me 🙂

     

    06/19/2017 at 9:41 PM #38620
    Derek

    Currently only the poster of a strategy can set a tag. What if  everybody can add tags when they read the post. In this example I would tag it “multi timeframe” and show it to the next newcomer that asks about this feature 😉

    12/10/2017 at 7:36 PM #55288
    GraHal

    Implementation of ideas / features discussed in this Thread would be a big benefit to all users of this Community.

    Last post was nearly 6 months ago … is there support to resurrect the desire for existing and future Strategies to be rated?

    It is a big task, so a simple method that can be further developed / expanded may the easiest way to move this forward?

    GraHal   

    12/11/2017 at 8:11 AM #55309
    Nicolas

    6 months ago right, during that time I did a lot of things 🙂 But you are right, this is something that could enhance the website library of codes. I have many others developments to do first, I will certainly come back to this in the next months.

    12/11/2017 at 10:10 AM #55322
    GraHal

    Nicolas I know, you are never still and all for our benefit! We all thank you from the bottom of our hearts! I hope you have already made your millions!? 🙂

    Please don’t overwork to the extent your health and family life suffers!

    Best
    GraHal

  • Author
    Posts
Viewing 15 posts - 16 through 30 (of 32 total)
1 2 3

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
buy on certin dates, not on technical singal
display indicators only on some timeframes
division by zero parameter error on 3 different strategies
trailing stop on ib – stop repositioning on dax
10ema/30ema crossover on weekly charts?
2 accounts limitations on ib
adding entry filters to your strategies
array support resistance buy on midlevel
automatic trading on ib stocks and etf
build ea bassed on indicatir

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