Feel like giving up?

Forums › ProRealTime English forum › General trading discussions › Feel like giving up?

  • This topic has 29 replies, 7 voices, and was last updated 5 years ago by avatarLeo.
Viewing 15 posts - 16 through 30 (of 30 total)
1 2
  • 10/18/2018 at 1:10 PM #83092
    Leo

    Thanks for sharing your insides here.

    Maybe you will like then the project I am working on. A lot of mathematics there.

    It is a neural network. Very simple at the moment.

    I am getting a lot of issues for detecting or defining what is that explosive movement.

    The neural network then learn about these movements and make his own prediction. The network itself after several tests seem to be right.

    Thanks in advance.

    10/18/2018 at 2:50 PM #83108
    Vonasi
    Brisvegas wrote:

    Ok clearly my time appears to be wasted here , i am not going to argue . There are no patterns and if this is the case writing code is futile for its all random …  massive eye roll   … I was going to show some patterns but im not going to fight against negativity for no reward

    I guess one persons discussion is another persons argument.

    10/18/2018 at 2:58 PM #83110
    Nobody
    Leo wrote:

    I am getting a lot of issues for detecting or defining what is that explosive movement

    Explosive momentum is a fast move , mathematically its distance over time , so we can easily ascertain this . The maths behind this is easy , define your starting point , define the time of interest and apply the maths . This will identify the explosive moves on a chart with a tangible measurement , Then from there we can analyze the patterns created by this explosive move and we can examine common patterns/context ( if they exist ) that precede that move . Is there an edge , does it repeat . how can we exploit that . A process , ultimately many things will fail but at some point you will find an edge  . My bin is full of failed ideas but every now and then you find some gold  , AND the more gold you find the better you will become at looking for it  . You need a system for developing a system  . Tell me the parameters you want for an explosive move and i will write it for you and we can apply it to instuments that interest you and analyze if and what patterns exist , we add context to this and you never know what develops  . Just putting it out there

    10/18/2018 at 3:19 PM #83117
    Nobody

    I came here with good intentions and you as a “moderator ” should be encouraging it but your passive aggressive negativity which seems focused upon myself is making it difficult to continue , i have plenty to offer and would continue to do so if only you moderated yourself  … good luck and good bye

    10/18/2018 at 3:29 PM #83118
    Nobody
    Vonasi wrote:

    guess one persons discussion is another persons argument.

    I am not the one talking in absolutes  , given your opinions of pattern recognition i am not sure what you are even doing here . Apparently its impossible and we should all revert to value investing  , i mean what is probability , its a fricking set of parameters that present a statistical edge  , what is a set of parameters that presents a statistical edge  . ONE fricking guess , its a goddamn PATTERN  .  End of story

     

    Now i might not be head of the class when it comes to writing code but knowing what/how to put into and develop robust algorithims i am in the top decile , i got something to say and its worth listening to  , i suggest you let those who want to listen do that and you go look for value shares and do a DCF or whatever it is you think is the key

    10/18/2018 at 3:40 PM #83119
    Vonasi

    Oh dear…. as a ‘moderator’ I must ask that all those on here who are throwing their toys out of their pram please calm down and try to discuss the subject rather than discuss their feelings on other people’s opinion.

    10/18/2018 at 3:46 PM #83121
    Nobody

    The passive aggressive retort isnt becoming   . Im frustrated for sure because you are not helping this conversation .   Im taking my toys and pram and ^%&ing of and it really is no skin of my nose but for a few members here it will be a travesty , you just keeping posting your secret sauce curve fitted junk that clearly doesnt work and maintain your #figjam demeanour .  I mean value investing in a coding site FFS   .  You need help

    10/18/2018 at 4:04 PM #83124
    Nicolas

    Hello everyone! What a mood in this topic 🙂
    I wish it were going well, because our forum needs this kind of discussion and believe me, this is the kind of discussion that will never have a firm and definitive end point, because each of us has his own sensitivity and his own version of what the market is and how he has to trade it. Recurrent patterns or not, momentum trading or other strategy, whatever! and I do not stop saying it, you have to unlearn everything you think you know, nothing is true, everything is right, I see it every day wobbly strategies that make people happy. So do not be so sectarian and withdrawn on your own positions, everyone has their own opinion, everyone has an opinion on everyone and it does not matter.
    So please, stay calm and keep talking, sharing spirit is our creed here!

    10/18/2018 at 4:35 PM #83125
    Nobody

    Nicholas I am sorry it has gone this way , my objective was to guide and through guidance and exploring the data i was planning on getting to the point of logical evidence  without going down the full data mining process that is prevalent . Datamining is easier than logical decision trees based on quantitative analysis and is a common mistake . Alas it was not to be  . Maybe at another time when the other party is on holidays i will return and continue this journey  , i actually tried to draw him into the conversation as he does have skillsets that could compliment the process , i will never make that mistake again

    10/18/2018 at 4:40 PM #83126
    Vonasi
    Brisvegas wrote:

    Maybe at another time when the other party is on holidays i will return and continue this journey

    As the other party retired at the age of 35 to go live on a sailing boat you may have rather a long wait for that holiday. 🙂

    10/18/2018 at 9:57 PM #83148
    Vonasi

    So – back on subject.

    Leo – your neural network coding has been very interesting and I have followed the progress with great interest but I feel that PRT code may not be the best tool to work with to do it justice which is why I have not done any work with it myself. Somehow I feel that my energies are better used quantifying individual things and then building the best of them into strategies. Sounds like curve fitting but if the quantifying is thorough and the strategy kept simple then it is not. Long periods of forward testing are currently proving this to be true so far. I have some fantastic out of sample equity curves but without presenting the code it seems others may just think they are curve fitted – so I will refrain from posting equity curves in future to avoid raising the blood pressure of others.

    In this topic there has a been a question as to whether ‘value’ has a place in algo trading – of course it does if someone truly understands what is meant by value. In an index value can always be found by identifying low prices and why would a strategy that goes long an index not be interested in buying at low ‘good value’ prices? Buying at higher prices equals poor value and bigger draw down so identifying value is a major part of any trading concept whether it be a long term hold or a one second strategy. Value is value. Buffett understood value and his theories of buying good companies when others are selling holds true in every time frame and for every instrument. He just happens to be running a massive fund and we just happen to be buying a spread bet – for both value is the same thing. Every major fund buys when the market is falling and sells when the market is rising. Due to our small account balances we need to scale this down and make sure that we minimize the chance that we are not buying when it is not the bottom.

    10/20/2018 at 4:37 AM #83208
    Leo

    Thanks again for your opinions (no offense but reading conversation that are turning very hot is also interesting).

    About the “value” of something here is my short story about it: some time ago (summer 2017) when I save some money and time for learn fundamental analysis, I choose some Key numbers for choose companies to buy, like (how much money they have quick ratio, long term debt, etc), how much are the dividend (KGV ratio in german), how good are the company to make money (return on asset, the most important thing for Warrent Buffet) and finally how much it cost (price-book ratio). I choose Verbio and Decheng Technologies… WHAT A FLOP!!!

    It that moment I look for the stocks everyone choose: Amazon for example, it cost around 900€ per stock and his price-book ratio was more than 20! ohhh man, who pays for something that it cost 20 times more of their assets ! not to say the technical aspect (everything shows overbought)… then what?… stock price got around 2000€.

    I do not believe any more on value of something, rather I believe  markets is a very chaotic system (chaotic in the meaning of mathematics).

    In case Brisvegas still reading this topic:

    Brisvegas wrote:

    Explosive momentum is a fast move , mathematically its distance over time , so we can easily ascertain this . The maths behind this is easy , define your starting point , define the time of interest and apply the maths .

    Then we choose two parameters starting point and time. I tried I am not still not satisfied because the definition of  “time of interest” is very subjective, I still missing explosive movements that in another time frame they are indeed explosive or a very slow but steady trend.

    I post here 3 different version of it, hope you give me your opinion, everyone has different approach ( i didn’t posted yet in the neural network topic because I do not feel like advancing then will be confusing). The ideal for me is to set a range of pips and look for a steady or explosive movement without influence of how much time it takes to do so and not so much pull back that can execute trailing stop.

    Please moderator, do not open a new topic. I know it is a very technical issue here but as long as someone is willing to work and develop a better “classifier” for the neural network, I don’t want to open a new forum topic.

     

    10/20/2018 at 5:40 AM #83212
    Vonasi
    Leo wrote:

    I do not believe any more on value of something, rather I believe  markets is a very chaotic system (chaotic in the meaning of mathematics).

    Even in chaos moments of value exist. We are not considering the value of a company based on its metrics but the value of price compared to previous price.

    1 user thanked author for this post.
    avatar Leo
    10/20/2018 at 9:38 AM #83221
    GraHal
    Leo wrote:

    Please moderator, do not open a new topic. I know it is a very technical issue here but as long as someone is willing to work and develop a better “classifier” for the neural network, I don’t want to open a new forum topic.

    A better idea might be a new Topic,  but ALSO provide (from your post above) a link to follow for Brisvegas  and / or  anybody else willing and able to help you Leo?

    Reason I suggest above is that this Topic title is Feel like giving up  and so capable coders may not waste time reading the to and fro between contributors on this Topic?

    Leo, your excellent work and now a call for help improving The Classifiers may get lost in the mire of this Topic?

     

     

    10/20/2018 at 7:07 PM #83245
    Leo

    Thanks Vonasi I got your point… I will think on it.

    Hi GraHal, you are right I talked about something very technical, but Brisvegas seems to having done something I am seeking to… anyway I wont open a new topic.

     

     

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

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
algos don’t want to start up in auto trading
backtest giving strange results
bollinger squeeze set up
code trailing stop like “set stop ptrailing x”
how do i speed up backtest optimization?
pre opening up
prorealtime, wake up (your website)
screener for gap up
screener issues with moving averages like others
steep angle up

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