2018 - 8 Future Basket Trading Journal

Discussion in 'Journals' started by MidwesternTrader, Jan 6, 2018.

  1. southall

    southall

    Could you share some more back test statistics?

    eg.

    number of traders per year (im estimating about about 1500+ from your first few weeks)
    % profitable trades
    % profitable days
    % profitable months

    longest losing streak in days
    largest losing day
    largest winning day

    profit factor

    Or anything else you can share
     
    #31     Jan 7, 2018
  2. themickey

    themickey

    Imo, these type requests are stupid requests and annoy the shit out of me.
    Type of questions noobs are always asking, then comes the demand of "prove it!"
    The markets are fluid, a backtest result will give a general idea, so what does 'largest losing day' and all the rest of the blah blah "....% profitable days" have to do with it?
    Grates on me this sort of thing.
    Back testing is hypothetical, has nothing to do with real life other than giving an oveall picture if you are heading in the right direction.
    Midwestern's spreadsheets on results of page1 should give you enough of a clue on viability. Sheesh.
     
    #32     Jan 7, 2018
  3. i960

    i960

    If you're doing analysis before the open you can use eSignal delayed data and not really have to worry about ICE data fees. Do you really need live data to set exits if you've done your analysis before the open?

    With regards to the timeframe question I had though, you mentioned daily trading however given that you're in and out by close I presume this isn't daily bars, right?

    My own experience is that expectancy rises as timeframe goes up - that is I find daily and weekly timeframes to constitute less bullshit.
     
    #33     Jan 8, 2018
    benwm likes this.
  4. Although I generally share themickey's take that these can devolve into the "you can only prove it if you share everything publicly" requests, I don't mind sharing some basic stats as it will provide context. The point of the journal is to run a live example of how back test results never equate to live trade results and illicit comments from other traders as a learning exercise.

    Stats from 2016 and 2017 Back Test with slippage and commissions included:

    Total Trades: 4,144
    Profitable Trades: 2,041 (49.25%)
    Unprofitable Trades: 2,103 (50.75%)

    Profitable Days: 312 (60.23%)
    Unprofitable Days: 206 (39.77%)
    Profitable Months: 24
    Unprofitable Months: 0

    Longest Losing Streak in Days: 6
    Longest Winning Streak in Days: 10
    Largest Losing Day: -$4,670 (One instance occurred where all 8 trades hit full stop loss)
    Largest Winning Day: $8,090
    Average Losing Day: -$1,203.06
    Average Winning Day: $1,986.89

    Profit Factor: 2.501

    If I am able to run this live for 6 months, that will give us enough meaningful live trade data to do a comparison against the back test. But it will need to avoid my -$25,000 maximum draw down at any given time criteria. Any such draw will close the account.

    The biggest outlier I see in these back test results is 24 profitable months against 0 unprofitable months. That isn't going to hold in live trades for sure. I'm not running the Goldman Sachs trading desk here.
     
    #34     Jan 8, 2018
    shatteredx likes this.
  5. Correct, I really don't need the live data, but it is useful to ensure my stop loss exits are always precise in case my entry encounters 1 tick slip.

    I think the best way to describe timeframe is I am using a modified time period to create a slightly unique daily bar for each instrument to run my analysis tool sets on. Hope I described that well enough to makes sense.
     
    #35     Jan 8, 2018
  6. southall

    southall

    You are so easily triggered

    The OP in his opening post wrote:

    "Guesses on the cause of its eventual destruction are welcome."

    Had he not written this line i would not have asked him for more information.

    Your stats look good, although the win rate does look high.
    As you are only trading one lot you cant be averaging down on losing trades, this is also good.

    The best way to see what could go wrong is to test out of sample and that will show you the real drawdowns you can expect going forward.

    However if you do this then you could lose confidence in your system if it continues to work well over the next year or two.
    Ignorance can be bliss,at least in the short term when trading. But eventually it will catch up with you.

    You seem to be betting you can make a fortune before conditions change.

    But you never know, perhaps your system worked well in the past as well.
     
    Last edited: Jan 8, 2018
    #36     Jan 8, 2018
  7. southall

    southall

    Is your stop loss based on your actual filled entry price or theoretical perfect entry price?
     
    #37     Jan 8, 2018
  8. i960

    i960

    You shouldn't need live data for this though. The exchange will report back the fill as an explicit message to the broker and you can go from there. Quote data isn't necessarily needed for this. Also you should remember the market doesn't care what your entry is - why set stops offset by a fixed amount if you have slippage or not? Either price hits your stop or it doesn't.

    BTW with regard to stops make sure you're using a stop limit with the limit price set such they the stop becomes a marketable limit order (e.g. Sell stop at 800, limit at 798). This ensures that your stop sits on the exchange as the futures exchanges you trade don't support stop market orders natively (which means your broker holds the stop). The tick range should be chosen so that it'll get you out on a fast move but not violate the allowable range the exchange sets (which is per instrument).
     
    #38     Jan 8, 2018
  9. Bobbybax

    Bobbybax

    MidwesternTrader sounds like the present day Ed Seykota. Good job.
     
    #39     Jan 8, 2018
  10. I was referring to setting the stop loss order up correctly here, not necessarily meaning the fill would be perfect. Obviously the back testing data is only theoretical and trading live for a period of time will show real stop loss fill results.
     
    #40     Jan 8, 2018