Quantitative Finance

Discussion in 'Strategy Building' started by Spectre2007, Jan 9, 2007.

  1. There are a lot of people trying to apply systems to different markets. The core principles:

    Stop Loss (SL)
    Take Profit (TP)
    Equity (E)
    Lots (L)
    Leverage (ratio)

    Take the above, and plug them into a given market, any market. And you will realize, the volatility itself, will take most people out, with their stop losses being hit.

    Lot of people advocate TP/SL of atleast 2:1, but the odds are, that since SL is half, the volatility will most likely hit your SL before it hits your TP.

    The above two conditions occur most of the time in any given market, since 80 percent of the time, the price will have a minute to minute volatility great enough to take out most people's stop losses.

    On a rare occassion, price instability secondary to massive order flow or news events, will trend the market, in these instances even though the minute to minute volatility may be the same, the minute to minute volatility migrates the price unidirectionally.

    The urge to trade all the time, leads to failure for most people. Unless the prudent ones minimize leverage or lot size during 80% of the market periods.

    The key is to have price migrate away from your entry, and let your TP be greater then 5:1.

    The other technique is to use a very large stop loss and a small TP, to let the volatility keep hitting your TP in the range channel. When adopting this technique, recognizing when the periods of transition occur between price stability and instability.

    Trendfollowing should be the goal of most people. Apart from that we are just spinning our wheels in these markets. Massive wealth accumulation cannot occur if we dont make a effort to capture the trend.

    In summary:

    Trade your system very infrequently. Unless your system is geared for range trading, with the ability to recognize transitions from price stability to instability.

    Am I wrong?
  2. Heres a simple system:

    Linear regression # (number of bars on timeframe chart)

    If long TP hits enter next trade long if LR is upward sloping for defined period.

    If long SL hits stop trading, till LR becomes downward sloping. Enter next trade short.

    If Short TP hits enter next trade short if LR is downward sloping fo defined period.

    If Short SL hits stop trading, till LR becomes upward sloping. Enter next trade long.

    If the LR is too long it is not as responsive, if LR is too short it becomes overresponsive ensuing a lot of false signals.

    The LR period needs to be optomized for the derivative being traded. Its different for each market when you plug it in. The SL and TP also are different for each market.

    Anyone else doing this?
  3. Applying this system during periods of price instability, on the 1 minute chart, where the system calculates minute to minute variability in price movement (SL has to be greater then this). And setting a large TP, during a news events, 10 minutes after.

    Works very well during nonfarm payrolls first friday of each month in T-bonds and Forex(GBP/USD).

    GBP/USD since it has a great degree of follow through.
  4. You might try back testing your ideas to see how they work on historical data...
  5. Already did, had a system created for GBP/usd. Working well. It seems to adhere to the SL and TP's and LR periods. Meaning its just not me doing this.. Some large funds must be using the same techniques.
  6. Sounds good then...
  7. Bars in test 2834 Ticks modelled 5568 Modelling quality n/a

    Initial deposit 100000.00
    Total net profit 108610.00 Gross profit 368730.00 Gross loss -260120.00
    Profit factor 1.42 Expected payoff 728.93
    Absolute drawdown 0.00 Maximal drawdown 30040.00 (16.61%) Relative drawdown 17.77% (22030.00)

    Total trades 149 Short positions (won %) 68 (80.88%) Long positions (won %) 81 (83.95%)
    Profit trades (% of total) 123 (82.55%) Loss trades (% of total) 26 (17.45%)
    Largest profit trade 3000.00 loss trade -10040.00
    Average profit trade 2997.80 loss trade -10004.62
    Maximum consecutive wins (profit in money) 13 (38950.00) consecutive losses (loss in money) 3 (-30040.00)
    Maximal consecutive profit (count of wins) 38950.00 (13) consecutive loss (count of losses) -30040.00 (3)
    Average consecutive wins 5 consecutive losses 1
  8. kevinmr


    What is the bar duration? What about commissions and slippage?
  9. Looks to me like 5 consecutive losers from your starting capital would cause most people to throw in the towel. Here's a few questions before I could really comment.

    How does your performance differ from a random entry with the same profit targets and stops?

    What is the average time in trade?

    What % of equity are you committing per trade?

    Is it an intraday approach or will you take signals overnight?

    What's the Sharpe ratio?

    A PF of 1.4 is not that impressive in the larger system trading universe. The win loss ratio is what I would expect for having such large stops.

    I tried to set up your basic approach in WL but had some coding issues related to the description you posted.

    Does your system only enter on news or will it trade any change in the LR slope as conditioned by the previous trade's success/failure?
  10. It stays out trading asian session. There are time constraints, the system uses to stay out of slow periods historically.

    It seems from the price action of that derivative, large stops are needed. And incremental price progression occurs before another SL is hit, and then another series of incremental price progressions occur.

    Linear regression is one of the most powerful tools I've come across in trading.

    I dont have a quantitative finance background, I started this discussion to see what other quantitative finance techniques people are using.
    #10     Jan 9, 2007