Need some help with first trend-following system

Discussion in 'Strategy Development' started by garchbrooks, Feb 21, 2010.

  1. Ok trend-followers, I could use your help.

    - Do I have to adjust my psychology for lots of small losses in order to make big gains when I'm trading trend?
    - What's a good psychological metric I can use for knowing the "stomachability" of a trade? What measures should I use?

    I've got a trend-following system on a stock that trades around 500k shares per day. This system doesn't use moving averages at all, and fades certain situations that tend to predict a movement to the other side -and- sustain the trend.

    I made the following assumptions in the backtest:

    - $1 per trade commissions (since I'm doing 100 shares at a time)
    - .05 cent slippage (intraday, the market depth book shows around a spread of 3-5 cents, and occasionally splits apart 30 cents)

    My first unoptimized backtest resulted in:

    - 39% of trades are successful
    - "Edge" per trade is about $9

    Tradestation tells me the long side has a profit factor of 1.34, the short side has a profit factor of 1.71, the total profit factor is 1.54.

    Average long-side winner is $50.95, average long-side loser is $23.34. Average short-side winner is $70.38, average short side loser is $27.80.

    RINA index is 2252.86, and return retracement ratio is 2.49. The max drawdown is $312.

    I usually trade market-neutral, but I stumbled upon this trade by accident and started working on a model to capture it. The equity curve goes up, then goes flat for a long time, then goes up again.

    I'm not sure I can stomach the flat zones -- I'm looking to get opinions from trend-following traders on techniques to address flat-periods in the equity curve. The bursts clearly more than make up for flat-zones, but the flat zones can tend to persist.

    I'm going to go back and optimize this system and smooth out the parameters, but would like to avoid pitfalls as this is the first trend-following system I have ever considered running live.
  2. These figures are well within statistical error or noise. For trend following IMO you should have profit factor >5.

    In other words, such low PF figures are a strong indication this system will turn unprofitable sooner than later.
  3. After optimization, I got avg win to avg loss ratio to be 5.95 and the tradestation profit factor to be 3.11. In order to get to something > 5, what do you suggest is the best approach -- go look at the losers and try to isolate another variable that can detect the losers more readily?
  4. I cannot suggest anything because I do not know your system details, neither I would anyway. Given your PF and avg win to avg loss ratio, your success rate is about 34%. Although some trendfollowers can live with such a low success rate, I find it very difficult to trade with 1 winner out of every three trades on the average. This could imply up to maybe 8 consecutive losers during certain periods of sideways markets.
  5. You are on the wrong path and will eventually grind to 0 if a streak of luck doesnt hit and then you quit.

    Look elsewhere for your edge.


  6. vikana

    vikana Moderator

    I just want to warn against trading systems with very low win ratios. You're basically dependent on getting ALL the winning trades for the numbers to work out.

    Try to remove the best 5-10% of trades and see if the system still is tradeable.

    If you factor in not getting filled, being out/sick/not paying attention, etc. the risk of the low win-ratio systems can be very high.
  7. Well said. This is what I was also trying to say and this is what experience has shown. Low win-ratio systems are often associated with high risk of ruin.
  8. Where are you getting this myth from? Some of the world's biggest, most successful traders operate with win % ratios well below 45%.
  9. This is the truth of trend following. Thats why it has held up longer then any strategy out there, the fact that you will be profitable on 40% of trades makes it difficult follow. It's not
    easy spending greater then 90% of your waking hours in drawdown. In recent years 2000-present, older methods of trend following (ie Turtle) have required tweaking to survive.
    Overall the more speculative participants in the market in question can cause evident trends to reverse violently (Too crowded).
  10. You trade first and then come to give us lessons.
    #10     Feb 22, 2010