Wanting to start a systematic trading approach

Discussion in 'Professional Trading' started by c.chugani, Aug 20, 2006.

  1. Ok.

    I am willing to put in the hard work and learn before jumping in and doing something stupid.

    I don't want to do anything impulsive, just to find out that I made a big mistake, start regretting my actions, and feeling that I should quit this field.

    Where can I learn the BASICS of market behaviour and the fundamentals of price action?

    Could you recommend only one starter book, which could explain me the very basics on how the stock market works?

    Like how is the "bid price" and "ask price" of a particular underlying determined?

    I need answers to basic questions, I am willing to put in the effort and learn, cos I really don't want to approach the market as though its a casino- for that I rather go to Vegas and i'll have a better time.

    So please, could someone enlighten me from where should I START?
     
    #11     Aug 21, 2006
  2. Hello C. Chugani,

    I think this is a good way to start. I remember doing similar studies about year 2003.

    1) Download about 40 years daily price data for General Electric stock symbol GE from yahoo.com.

    2) Copy and paste the GE price data into a Microsoft Excel or Works spreadsheet.

    3) Calculate the average price for 300 trading days using the data adjusted for splits and dividends. I mean, calculate a 300 day moving average.

    4) In another spreadsheet column calculate 3 percent of the moving average value.

    5) In another column add the 3 percent value to the 300 day moving average value.

    6) Now we want to see if it works. When the stock price first moves above the value of the moving average plus 3 percent then the system buys.

    7) When the price declines below the value of the moving average this is a sell signal.

    8) Test long positions only. (Just to keep things simple)

    9) Assume you start with US $ 10,000 and risk 1 percent on each trade. One percent of $ 10,000 is $ 100. Pick a number, say 10 percent. Calculate the position size so that $ 100 = 10 percent of the closing stock price. Now you know how many shares to buy.

    10) Write down the results of the trades. Total the profits and losses of each trade and see if the system makes any money.

    11) Later you might want to subtract commission costs and slippage but it is not necessary for a first test of a system.

    12) Change the numbers and keep track of the total profit. Maybe a 400 day moving average works better than a 300 day moving average. Maybe 4 percent added to the moving average works better than 3 percent.

    You might want to keep track of the results in another spreadsheet so you can copy, paste and perform calculations on the values.

    You can perform lots of variations to this simple trend following system and discover things that work and things that do not work.

    You can use the system to back test other price histories also such as currency exchange rates, crude oil prices or Coca Cola stock price.

    You can make the system as short term or long term as you wish. Test weekly and monthly data also.

    I believe this is a variation of a system developed by Chester Keltner about year 1950. You might want to google Keltner and learn more about the system and the man.

    I wrote a computer program to perform the calculations described above and report the results. Following is an edited copy of the results. The results are long, there are many trades so I will only paste the last few trades.

    If anyone is interested I could start a new thread and we can, all together, try to develop an improved version of this system.

    ====

    7-Jun-04 OHLC:[ 29.87 29.87 29.87 29.87 ] buy 29.87 size 33
    Dollar Position Cost is 985.71
    5-Oct-05 OHLC:[ 31.97 31.97 31.97 31.97 ] sell 31.97
    Position Net Gain Or Loss is 69
    Subtotal profit $ 11494

    7-Oct-05 OHLC:[ 33.47 33.47 33.47 33.47 ] buy 33.47 size 29
    Dollar Position Cost is 970.63
    24-Jan-06 OHLC:[ 32.47 32.47 32.47 32.47 ] sell 32.47
    Position Net Gain Or Loss is -29
    Subtotal profit $ 11465

    14-Mar-06 OHLC:[ 33.53 33.53 33.53 33.53 ] buy 33.53 size 29
    Dollar Position Cost is 972.37
    13-Jul-06 OHLC:[ 32.67 32.67 32.67 32.67 ] sell 32.67
    Position Net Gain Or Loss is -25
    Subtotal profit $ 11440

    Number of trades 43
    Total profit $ 11440
    Profit after subtracting $ 100 commission & slippage per trade: $ 7140
    Heat is $ 100 per trade
    Drawdown is 0.0677
    Initial capital is $ 10000
     
    #12     Aug 21, 2006
  3. Excuse my ignorance. Could you please elaborate on how I calculate how many shares one should buy?

    I understand where you are going with the 1% risk per trade (hence $100). That is your determination of the risk you want to take per operation.

    Now what I don't understand is where you get / or on what basis you determine the 10% figure.

    BTW, thanx for helping me out by this sample model layout.

    I guess yahoo provides historical price data for a monthly fee?
     
    #13     Aug 21, 2006
  4. MTE

    MTE

    My guess that 10% is arbitrary, just used as an example. Basically, it implies a 10% stop loss. E.g. stock is trading at $50, so 10% is $5. So if you are risking $100 per trade then you can buy or sell 100/5=20 shares.

    Historical price data on Yahoo is free.
     
    #14     Aug 21, 2006
  5. In your example, if the stock trades at $50, then I can only buy 2 shares per trade ($100 risk per trade).. am I right?

    From where do you get 20 then?
     
    #15     Aug 21, 2006
  6. Sorry!!!

    My BAD!!

    you mean I can risk $100 in LOSSES per trade..

    Excuse me.
     
    #16     Aug 21, 2006
  7. Yes I just picked 10 % of the closing price as a number to use for calculation purposes. Some traders call the 10 % value the "heat" parameter. In this case if the stock price falls 10 % it represents a $ 100 loss. We could "turn up the heat" by selecting 5 % instead of 10 %. In the case of 5 %, if the stock price falls 5 % then the position lost $ 100.

    Heat is another parameter that a system developer can experiment with.
     
    #17     Aug 21, 2006
  8.  
    #18     Aug 21, 2006
  9.  
    #19     Aug 21, 2006
  10.  
    #20     Aug 21, 2006