Which Strat to use..

Discussion in 'Strategy Development' started by t0yland, Sep 1, 2003.

  1. t0yland

    t0yland

    Hi everyone,

    I have been developing different strategies for awhile now. So far I have two that I am thinking of doing some live testing on.
    When I live test them I will use probably 100 shares per trade or just an equal amount of money per trade.

    Anyway which do you guys think will be better to test first.


    Both of them use the same set up but the profit taking is different.

    1: 95% win rate, W/L ratio: .5
    I have backtested for the past ten years on about 300 different stocks. The results do vary but only by a slight amount. The above is an average.

    2. 70% win rate, W/L ratio 1.5
    Again I have backtested this for many years on a range of securities.


    Anyway heres a little background on what these do.

    I use daily data to take a 'guess' on what the next day will do. In the morning, if the price opens within a range, then it buys in. The 95% w/l strat sells at a .05% gain. The other one holds off untill the day closes and tries to sell at the close.

    ONe more question: How difficult will it be to buy in at the opening price. I will need to be able to see what the price has opened at before the decision is made to buy. Therefore I cant use MOO or LOO orders... :/

    Oh, I am automating it, using java though IB DDE...
     
  2. size

    size

     
  3. I am concerned about the need to enter at the opening price. Not so much that you have to wait to see what it is, but if you need that exact price. Last year I did some backtesting on some ideas that at first run were very profitable. But varying the entry and exit price by a nickle turned a winning method into a losing one, and that is my concern for you.

    Most days don't open and then go from there, they will cross the open at least once so having an opportunity to see the opening price before trying to get is in not the problem. But a method that needs to get a certain price.... be very careful realtime.
     
  4. tntneo

    tntneo Moderator

    Use slippage to see how your strategy is dependent on the exact price.

    in your back test, use a fixed amount of slippage (ie use a worse price than the actual open) do the same for the exit (although with limit prices, you can get away with it and still be realistic). it's very important to count slippage in your entry though. specially if you want to 'see' the opening print before you pull the trigger.

    regarding the ratios, you need to consider maybe more as a factor the drawdown, and study the drawdown patterns. win ratio and profit factors are seldom the most important ones.
     
  5. acrary

    acrary

    The second strategy has almost twice the expectancy of the first. Try that one first.

    1. e = (.95*.5) - (.05*1)
    e = .425

    2. e = (.7*1.5) -(.3*1)
    e = .75
     
  6. If you want to save yourself some money paper trade both strategies with live data, and if after 200 trades you are nowhere near the projected profit (which will most likely be the case), you will know what I am talking about.

    The most important thing is: During your paper trading, you must not modify the strategies in any way, otherwise you will never see what I mean until you lose money.
     
  7. jrkob

    jrkob

    If your proffit target for each trade is 0.05%, it seems pretty clear to me that what you're merely trying to do is market making.

    Market making is: I buy right now at market price and put simultaneously an offer 0.05% higher, hoping for the volatility on your index/stock/future to be high enough for your offer to be taken.

    I personally would bet that you have grossly underestimated slipage (and may be cost?).

    Also, in case you buy say at opening price, and if opening price is the highest of the day, the loss you gonna take is going to eat up quite a number of your "win".

    I think you have to test your system with real data quite seriously. I would be very suspicious that an individual like you and me could succeed at market making.
     
  8. size

    size

    Since you calculate the range in which you want buy, why can't you just enter a limit order(s) prior to the open?
     
  9. damir00

    damir00 Guest

    if you mean buy the opening price, at the open, then virtually impossible except when you get lucky. since you're talking about 0.05% gains - if i read you right - it sure doesn't sound like you have enough slop in the system to account for the realities of position entry.
     
  10. t0yland

    t0yland

    size,

    I was actually thinking about that. For instance I want to buy in at no more than 25.50. If I use a LOO order It will either buy in there or lower or the order will be canceled. Also this kind of forces large diversification. About 1/3 of the stocks in my universe will be within a price range that I want to buy/ss the next morning. So I would have to 'randomly' pick some and set LOO orders with say 5% of capital each.

    Anyone know if theres a certain number of sharees that must be purchased to do a LOO order? I heard 100 but not sure..
     
    #10     Sep 1, 2003