Dilemma

Discussion in 'Strategy Building' started by fordewind, Feb 19, 2018.

  1. Why don't get in touch with this guy,looks like he knows his staff and more.
    I never contacted him,but i had this link on file.He was a freelancer years ago and since you dealing with intraday staff you need to talk to an expert.

    https://ua.linkedin.com/in/vadim-zaitsev-b31b1338

    Disclaimer:
    I never visited Ukraine and have no contacts there,this is only my advice not a recommendation.
     
    #11     Feb 19, 2018
  2. linkedin is blocked in Russia
     
    #12     Feb 19, 2018
  3. #13     Feb 19, 2018
    fordewind likes this.
  4. #14     Feb 19, 2018
  5. fan27

    fan27

    This should be relatively easy to diagnose. Try the following:

    1. Instead of placing actual trades, log (to console or file) time stamps for the data being read (attached to the ticker data), for when a signal occurs and when a trade is supposed to be placed. What you want to see here is if there is a lag between when you are evaluating the ticker data vs when the ticker data was printed. If there is little to no lag, go to step 2. If not, there could be a to long of delay in receiving/processing your ticker data.

    2. Manually look at your data and make sure those time stamps line up with what you are able to see when manually backtesting your strategy.

    3. If step 1 and 2 are good, continue to log out the time stamps but also place the actual trades. Compare your fill date/time vs the time stamp you log right before you place the trade. If there is a gap in time, you have high latency in your order execution. That could be your problem.
     
    #15     Feb 19, 2018
    fordewind and tommcginnis like this.
  6. tommcginnis

    tommcginnis

    Regarding your risk of wrist of RSI:D:
    1) Switch mouse hands.
    It is a trivial thing to reverse mouse buttons:
    in Win7: ControlPanel/Personalization/Mouse

    "Back in the day" your fingers never had to leave the keyboard, and the load (per the QWERTY "Origin Story") was shared over all fingers. With the coming of the GUI, and the very poor habit of thinking our 'handedness' needs to follow on to a mouse GUI, we end up with a number pad on the right side of our keyboards, and then a presumed right-handed mouse, and we are (seemingly) doomed to take our hands OFF the keyboard, move a mouse pointer to a specific spot and click-in to identify it to the computer, then let go of the mouse to twink in some numbers from the number keyboard, then back to the mouse to click-in again to confirm it all. THIS USED TO BE doable from 2-3 keystrokes from that older keyboard interface.

    AT ANY RATE, by simply switching to the left hand (and by switching mouse buttons, it becomes perfectly in tune with what we used to do with the other hand...), you will cut down on a HUGE amount of metacarpal synovial sheath load. "Ouch!" no more!

    2) Glucosamine HCl is your friend. (And, get the HCL, not the Sulphate, which is only 2/3rds as effective, but 3/4s the price. Pay for the HCl, and get your money's worth.) Glucosamine/Chondroitan/MSM are items that our 21st Century diet seems to do poorly. (The white, crunch/slippery portions of meat.) They are a major part of cartilage, tendon structure, the synovial sheath AND the synovial fluid [!!!!!]. Boost that bunch, and your whole body gets more *slippery* -- a good thing. In the meantime, you will *not* be queuing up for a lifetime of NSAID load on your liver. "Eeek!"

    3) THE ORIGINAL ISSUE!! From a (methodological) distance, I'm hearing a disconnect between algo and market orders -- I think the algo is only half-there: as accomplished as the algo must be, just to do the first half of the task [reading market history/current portfolio and identifying entrys/exits], when it comes to trading [the second half of the issue!], it's been left in α-mode. It needs to be finished.

    SO GO OLD-SCHOOL!.
    Write down the step-by-step process when trading by hand.
    Work it into a flow chart.
    Code it.

    I mean no disrespect by this, but from reading this thread, it's like you forgot the rigor of the set-up must follow into the execution, as well! Don't leave your diamond unpolished, or it's just a stone. :thumbsup:


    Hope that helps.
     
    #16     Feb 19, 2018
    fordewind, Sprout and vanzandt like this.
  7. Why do you experience this slippage? Is the bid/ask spread very large? Or is your analysis based on the last traded price and have bid/ask moved away from it by the time you place your order?

    I am trading futures full-automatic, but not intraday trading. In most cases is the bid/ask spread only one tick. In cases where the spread is two ticks (or more) I place initially the order halfway in between. This often gets filled.
     
    #17     Feb 19, 2018
    murray t turtle likes this.
  8. It`s not bid/ask slippage issue, but the indicator slippage issue.It`s a momentum strategy which relies on an immediate entry.Manually you recognize when it slips right away, but couldnt find an algorithm that can handle this.
     
    #18     Feb 19, 2018
  9. For better understanding, see the image attached.
    entries.png

    At some point you have three-four, etc such thing in a row, which diminishes overal perfromance.News impacts of course don`t count as it can be handled.The price always gets back to the entry and as the signals are quite strong is very forgivable.But as the volatility is not predicatable in any way it can`t be handled on the execution framework.The price can slipp 1 point or 10 points and always uncertain.
     
    #19     Feb 19, 2018
  10. I see. I don't have a solution for this. Maybe an analysis on multiple time frames is necessary to respond fast enough.
     
    #20     Feb 19, 2018
    fordewind likes this.