ProRealAlgos? Real or fake

Forums › ProRealTime English forum › General trading discussions › ProRealAlgos? Real or fake

Tagged: 

Viewing 15 posts - 151 through 165 (of 227 total)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
  • 03/26/2021 at 1:59 PM #165410
    XORANDNOT
    jmf125 wrote:

    History tends to repeat itself. I am not experienced enough to debate this topic !

    Ask @Vonasi. He might tell you his opinion.

    03/26/2021 at 3:08 PM #165417
    GraHal

    Price Action history is like clothes fashion … repeats, but never quite the same?

     

    03/26/2021 at 3:57 PM #165419
    jmf125

    @Nicolas some indication of optimisation is within the algo itself.

    // **** 4 steps for success ****

    // ** STEP 1 – Strategy idea and alignment
    //Defining an edge or an idea, and enunciating the strategy including settings e.g. timeframe, index, stoploss

    // ** STEP 2 – Source code optimization
    // IS/OOS-optimization of the source code parameters using 60-70% IS and 40-30% OOS.

    // ** STEP 3 – Walk forward testing
    // 60/40 WF-test over 5 iterations. All periods must be over 50% WF-efficiency. Total efficiency over 80%.

    // ** STEP 4 – Demo, trial & Live
    // Algos tested in demo for X trades or 10k bars before going to live, also for X trades or 10k bars

    03/26/2021 at 4:03 PM #165420
    Nicolas

    Thank you for this info, so it matches the idea that I had indeed. The development looks quite decent and sane to me. That's the way most of them here work for their strategy development. So indeed, if we use the latest version with the latest parameters on a backtest on 1M candlesticks, then it seems normal that the performance in this OOS data is less good than on those we validated with the last OOS occurrence. This being the case, it will of course be necessary to return a new version to the period defined by its iterations which validated these OOS, this is what the different iterations are for. So as it stands, I don't know how many months that could represent.

    03/26/2021 at 4:16 PM #165422
    jmf125

    Summary of all the algos backtest over 1 M bars – I believe these are correct but don’t guarantee it.

    03/26/2021 at 5:15 PM #165426
    Vonasi
    XORANDNOT wrote:

    Ask @Vonasi. He might tell you his opinion.

    In my opinion (for the little that it is worth) every strategy is curve fitted to historical data because that is all we can ever do. Will history repeat itself – yes sometimes and no not at other times!

    History is relevant. The candlesticks created before the invention of the computer are very different to those after its invention. The candlesticks after Trumps election are very different to those before and so on. All indicators are created from price movements or comparisons of price so comparing today’s indicator values or movements directly to those in 1972 is pretty pointless.

    A strategy can be a winner in 1972 and a winner today but we must try to fully understand why it is considering that we know that historically price movements have changed drastically over that amount of time. In fact it was easier to make money before the invention of the computer but we only know that now because the computer has been invented! If we were drawing candlesticks with a pencil and calculating an average with an abacus then making money in 1972 would have been much harder than we think it is today.

    So don’t rely on history but as it is all we have to use it is the only thing we have to rely on!

    03/26/2021 at 6:45 PM #165439
    XORANDNOT
    jmf125 wrote:

    Summary of all the algos backtest over 1 M bars – I believe these are correct but don’t guarantee it.

    Can you post the attachment once again, please ? I guess it got lost somewhere…

    03/26/2021 at 7:12 PM #165447
    jmf125

    With attachment as it seemed it did not work first time.

    03/26/2021 at 7:53 PM #165450
    jmf125

    I agree all we can do is to find a strategy that fits the curve as closely as possible, and hoping it will continue.  Going back to 2010 is not really going back when computer did not exist 🙂 and I assume that in 2010 already most of the trading (including professional trading) was already controlled by algos. I also assume this is a constant game of police and thieves, one finds an edge in response to an algo trading and it is successful for a short while, until it gets found out …

    So in that respect, recent events have more weigths than past events. for a short while. We can also hope that a certain logic can be applied, and that this logic remains true.

    Only one thing is certain is that there is no logic in the market !!! 🙂

    03/26/2021 at 10:09 PM #165458
    Vonasi
    jmf125 wrote:

    I also assume this is a constant game of police and thieves

    It is a constant game of buyers and sellers. For every buyer there has to be a seller or lots of smaller sellers if it is a big buyer. This has never changed but the numbers have just got bigger both in price if it is an index and bigger in quantity of buyers and sellers for everything that can be bought or sold. A look at any commitment of traders chart will show us how the small guys are doing what they think is right and the big guys are just using their massive equity to hoover up and make a profit off the small guys no matter what direction price is going in.

    03/27/2021 at 3:49 PM #165498
    XORANDNOT
    Mattias wrote:

    It would be so much trustworthy if that could be a part of the selling package.

    There is no trust and reliability in trading. I think this ought never to be forgotten.

    03/29/2021 at 11:12 AM #165614
    GraHal
    jmf125 wrote:

    Yes excel. Now with totals

    Any reason why you didn’t include the ‘PRTA’ Algos in your excellent Table of results?

    03/29/2021 at 9:31 PM #165681
    jmf125

    It is included, the first row. But sorry I did not keep the exact naming convention , first 10 are the old PRA algos and the next 8 are the new ones PRA+ (I did not do Nikkei and it seems I missed another)

    You should be able to easily match.

    1 user thanked author for this post.
    avatar GraHal
    03/31/2021 at 9:23 AM #165810
    zilliq

    Hi Guys,

    No more answers from @ProRealAlgos ?

     

    05/21/2021 at 10:39 PM #170156
    ProRealAlgos

    We are here. We just don’t want to interfere with the discussions but let people make up their own mind 🙂

  • Author
    Posts
Viewing 15 posts - 151 through 165 (of 227 total)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
and or logic sequence
binary “buy/sell” indicator results display: convenience or further processing
bug or wrong setting?
buy or sell at a certain distance from moving average
buy or sell at a specific value
can’t see procode or screeners in prt 11.
code conversion easy language to pro real time
defparam flatafter for long or short
does pro real time suppor webhooks?
does sest stop trailing command work in real?

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