I've develpoed a simple system for mecanicaly trading the QQQ's. Here it is: This is based on the opening price of the NASDAQ 100. Buy at the opening when the Standard Deviation is below its 8 day historical average and when the opening will be positive. Sell at the opening when the Standard Deviation is above its 8 day historical average and when the opening will be negative. From Oct 2,1985 to Feb 6,2002 This system made 357% on buy side and 99% on the short side. Does anyone have a simple mechanical system to beat this or something that can be added to make it better? JJ
1. QQQ exists since 3/10/99 so a backtest ist not possible back to 1985. 2. A backtest at the index is not identical to a backtest on the QQQ. a) the most chart programs show false open gaps for the index. b) the QQQ trades more agressive than the index. c) the index stops trading at 16:00 PM and the QQQ stops trading at 16:15. 3. What kind of stop rules do you use? 4. How did you calculate slippage? 5. How looks the average win/loss ratio and please tell us the average trade in % of the investement. 6. Yes I have a better performing tradingsystem for the QQQ but it uses intradaydatas.
Recently I've seen 'standard deviation' used in several publications and on my TS software. Could someone explain what it means, how it is calculated, and how it relates to the mechanical QQQ system previously mentioned? Thanks.
I think it will work great. Get all your friends together and raise a lot of money and put it to the test.
Pretzel: In this strategy one is always long or short the market. ufo12: 1. That's correct; but only going back a couple years is not long enough to base a theory on so I used the NASDAQ 100 and assumed that the QQQ's would act similar. 2. Since this is more of a long term position type of program, I again assumed that both the QQQ's and the nasdaq 100 would act similar, some times better some times worse. 3. NO stops. 4. No slippage or commissions included in the results. This was just a trial to see how it works. 5. On the BUY side there was 72 trades and a average holding time of about 2 months. The average return was 5.0% and was 76% profitable. On the sell side there was also 72 trades an average holding period of 1 month. The average return was 1.4% and was profitable 51% of the time. 6. Great as I have fully disclosed this program, would you care to share yours? I do not feel a mechanical program that deals with an index would not be giving away the holy grail. Thanks. Light.. Standard deviation is a statistical formula that is in all spreadsheet programs and can be found in many texts. It is the same as volatility, and is used to calculate the properties of a normal distribution. Easyrider; I don't think it's that killer of a program. I'm just looking for something to use for my IRA account. If you like it this much, as it's fully disclosed, I'll give you permission to get rich off of it. Rigel; Yes the sample size of the Std Dev is 8 days. In using this in IRA program I would get into an interest bearing account rather than a SHORT, the risk would also be much lower. Any ideas to incorporate into this program?
"I've develpoed a simple system for mecanicaly trading the QQQ's. Here it is: This is based on the opening price of the NASDAQ 100. Buy at the opening when the Standard Deviation is below its 8 day historical average and when the opening will be positive. Sell at the opening when the Standard Deviation is above its 8 day historical average and when the opening will be negative. From Oct 2,1985 to Feb 6,2002 This system made 357% on buy side and 99% on the short side. Does anyone have a simple mechanical system to beat this or something that can be added to make it better" i have a trading system that is better than that.i back tested it and it made 600% trading the qqq since 1991.here it is.buy the qqq in 1991(thats all the further back qqq goes) at 5 and sell now at 34. as you see system backtesting isnt all it seems.if it doesnt outperform buy and hold it isnt worth much.
Exactly...How can anyone perform a backtest on data from the mid-late 1990's and consider this a valid sample...The real problem is that all systems are still using trending data...Anybody look at the market NOW...It does not trend it goes flat and sideways in tight, volatile ranges... That is why systems will always be somewhat dubious...You cannot qualify a system solely based on what is has ALREADY done, just look at the difference in the market of the 1990's and the markets of 2002...
Sorry; I should have mentioned that the returns were the natural log returns. Vhehn and Easyrider; the return of buying at 5 and selling at 34 is: Ln(34/5)x100=192%. In developing a position trading program, you have to use the natural log of returns. Example: first trade I buy at 5 and sell at 10, I made 100%. 2nd trade I bought at 10 and sold at 5, I lost 50%. Total return (100-50)/2 or 25% per trade. Wrong!!!! I assume you are day traders and that this would mean nothing to you as you probably make or loose less than 1% per trade. Vulture; If you don't use the past to reflect the future, what else can you use for research. Maybe you are correct, that the market is a random walk, but I would like to think I have a better chance than 50-50. Also, if you looked at the program you will notice it is not trend following. Instead of trying to be so negative, maybe with all our experience and expertise we can develop a program within Elite Trader for all our use. I doubt if many or any mutual fund has a record to match what I have presented here. Would anybody like to make a positive statement? Best regards: JJ