Has the strategy stopped working or is it to reaching a new max drawdown

Discussion in 'Trading' started by rvsw, Dec 1, 2019.

  1. rvsw


    I have a strategy which has been working well for the past about 1 1/2 years now it is producing a maximum drawdown that I have not seen in my testing in the past 5 years of back testing data. Added to it, it is also seeing maximum number of consecutive trades that are in a loss. How do we determine whether a strategy has stopped working? It may also be possible that this is just a new maximum drawdown and eventually the strategy would start performing.
    In general, how do we find out whether it is time to discard a strategy or to modify parameters?
    dennis86 and cornerstone like this.
  2. dozu888


    they all stop working eventually...
    all 'edges' are self defeating except they are not (e.g. you have some advantage that others can't get to... inside info, proximity to the exchanges etc).

    if your 'edge' is just some market conditions that others can discover also, they will.

    then what happens is this market condition is crowded with traders and you all become vulnerable.... another group of guys come in and kill you.

    better get busy and figure out some new edge
    jys78 likes this.
  3. like dozu said, most strategies stop working in some point. my strategies get confirmed by volume and I use Bookmap to evaluate if the amount of volume the strategies are needed will appear, so if the levels that important to the strategies doesn't have enough liquidity over and over again I presume that something in how the market works changed.
  4. tiddlywinks


    You probably should mention what instrument(s) or type of vehicle(s) your strategy is for.

    Five years of data for back-testing is not enough. You found a "something" that works in a specific market environment. Your job now is to differentiate the current environment from that of your test period.

    I would rather backtest with as much data as I can get (fyi, 25 years of data will include the dot-com bubble AND the 07-09 period along with subsequent breakouts, whereas 10 years of data won't even include the 07-09 period. Your 5 yrs worth, well...) You may find that the "something" you think you found is merely a face in the clouds.
    dennis86, inCom and jys78 like this.
  5. traider


    How often does your strategy trade?
  6. tomorton


    I don't believe that valid strategies work for a bit and then fail for ever.

    So if this one is failing now, it was never valid anyway - it was either only working through a unique market environment or it was producing positive results through a period of good luck (even roulette players get a run of wins from time to time).

    If you can't identify and understand the edge your strategy had when it was winning, its invalid, stop trading it.
  7. guru


    Backtest it for 12-13 years since 2007. Although even this wouldn’t be enough, so more the better. Also use Monte Carlo simulation.
    With 5 year backtest you can’t even cal it a “strategy”. It looks more like an experiment that requires 10 more years to see whether it may work.
    Volker Knapp likes this.
  8. imjohn


    You didn't indicate whether you're trading intraday or interday. If intraday, note that volatility & RTH ranges in the US indices has collapsed since October 12. Some other recent time periods with conditions similar to the present were April 2019, September 2018, and the latter-half of 2017.

    For me, intraday becomes frustrating in such time periods, but an interday "long-only" strategy with some type of trailing stop does well.

    As volatility/range expands (such as February 2018, October-December 2018, August/Late September 2019), conditions ripen and there's a ton of intraday movement and opportunity.

    Perhaps go back and overlay a filter to your strategy results to determine performance in periods of medium- or high-volatility, versus times similar to the current low- volatility environment. Were all of your worst drawdowns in times of low/high volatility, low/high range, etc ?

    To me, Backtesting a 5, 10, 20 year period etc, just applying the same parameters across all years/days, without noting different environments, is just a way to confirm the obvious, "nothing works good all of the time".

    Key is being able to recognize the current conditions and adjust your strat accordingly.
    tommcginnis likes this.
  9. no one can ever answer this.
  10. maxinger


    You didn't give sufficient information.

    So my following response is generalised.

    Do not blindly apply your strategy.
    Do not blindly apply TA to any financial instruments.

    Macro environment changes with time.

    many people persist to trade currencies and they find life very stressful.
    talk about currency like gbpusd.
    It was highly highly tradable in 2009.
    those who traded currency past few months might be gnawing & grinding their teeth in dispair.

    lately index futures, oil , gold , bonds are moving.
    so look at macro picture first.
    #10     Dec 2, 2019
    dennis86 likes this.