Manual strategy testing

Discussion in 'Trading' started by etrades, Oct 12, 2019 at 4:57 PM.

  1. etrades

    etrades

    Hi guys,

    Is it possible to test a manual strategy ? Or is testing only for automated strategies ?

    If it is possible, how to do it ?

    Thank you
     
  2. guru

    guru

    I just did one using "thinkBack" feature of TD Ameritrade software (Think or Swim) that you can get for free with a demo/paper account.
    Screenshot below, though I erased the specific positions I was testing.
    I don't have time to explain details, so you may just want to spend some time figuring this out, possibly Googling, etc.
    Generally you just select a stock and a past date, then click "Buy" or "Sell" on specific option or stock symbol, then keep doing this for different dates and see how the P&L chart looks as compared to the benchmark.

    upload_2019-10-12_15-9-47.png
     
    etrades likes this.
  3. CharlesS

    CharlesS

    I've seen several podcasters use and enthusiastically recommend "ForexTester"
    https://forextester.com/

    Haven't used it myself as I do the same directly off any chart and just remember (!); forextester records results of trades you make with it against a chart -- provides as much automation as is likely possible.

    Don't know if it can be used for non-fx.

    I have no affiliation w/ it.
     
    etrades likes this.
  4. tomorton

    tomorton

    Seems like a high risk of failure and much effort.

    Profitable trading does not demand a band new strategy which the trader just invented. Why not pick up a proven strategy from the public domain for the price of a seminar or a couple of second-hand books or a few hours browsing? Or even buy a strategy if you feel you must. You don't need to re-invent the wheel.
     
  5. qlai

    qlai

    For example?
     
  6. The fatal flaw in that reasoning is the inherent recursion in it. How and where does someone acquire the judgement to decide whether a strategy is "proven" without being an expert trader in advance?

    Lather, rinse, repeat...
     
    imjohn likes this.
  7. imjohn

    imjohn

    I trade manually. Though my entries are based off rules that are clear to me.

    For all of my trading "ideas", I manually recorded every occurrence of the "signal" in multiple timeframes over the prior 250 trading sessions. If the idea was not looking profitable in that period, it was dropped. If it was profitable, with similar results across multiple timeframes, I'd double down and look through multiple timeframes in another 250+ sessions (in different market conditions), to see if results still held true. If they did, I'd start live trading the idea, continuing to keep tabs on the results.

    For each sig occurrence, I recorded:

    -stop size (based on signal bar)
    -how much price moved in my favor before it reversed back to stop (or to next sig level).
    -several concurrent things that could potentially be used as filters to negate the trade (e.g. what's happening in other time frames, what are the MAs doing, which entry #)

    With this data, I determined how many trades would have won/lost when applying various reward:risk scenarios (1:1, 1.5:1, 2:1) and filters to all sigs.

    Note, testing ideas felt extremely demanding on my time and focus. Also, ideas didn't come until 1000s of hours from my initial starting point.

    EDIT add: This isn't a recommendation.. just process I came up which satisfies my comfort level to put on a trade.
     
    Last edited: Oct 12, 2019 at 11:25 PM
  8. etrades

    etrades

    Thank you for the replies guys !

    hi imjohn,

    why did you choose 250 trading sessions? also, is there any automated solution for this ?

    but then I guess if you can automate the test then the same automation would be for actual trading as well !
     
  9. imjohn

    imjohn

    There’s approximately 250 sessions per year. To me, 1 year seemed like a good minimum starting point.

    No experience with automation, so no comment there. Lucky for me, I’ve enjoyed manual trading since day 1.
     
  10. pstrusi

    pstrusi

    Possible certainly it is because all you need is reading the tick by tick data and applying your strategy's rules. But practical or even that you might finish a minimum timespan of data is other matter. Unless you have a strategy that trades a few times a day, perhaps. Still, it doesn't seem practical and even more, if you had your strategy ready, you'll miss the fact that a computer runs your strategy, cause in this time, markets are mostly driven by computers, and they go at a high speed, above all when news pops in. My suggestion: learn some basic and easy language or get professional trusty help.