Developing a Trading System Step by Step
- Thread starter Murray Ruggiero
- Start date
If you are using CSI as your data provider, you can do this pretty easily. There is a setting that will output the data with the contract month used in the continuous series to replace either the volume or the open interest. Now, you have a field you can reference that will tell you when the roll happened. The easiest thing to do is to just have your system exit and re-enter when this field changes. That will simulate a roll for purposes of bro and slip. If you want to get fancy, you can set the raw data (rolls from contract to contract, but doesn't adjust the values) as your primary tradeable and the continuous contract as your Data2 (in TradeStation) or Independent1 (in TradersStudio) then check to see when the roll happened and exit/reenter on the raw data.
X
Xephen Merchant Partners
swingman, please send me a wire transfer for $50,000. That is a very cheap price for what I am about to tell you... Don't read further until you send the wire...
Stop trying to make the long side and the short side work the same. Separate them into different entries and different exits. If you have to, for simplicity, run them as completely different systems. Then start testing your exits. You will find that the best(optimized) entries and exits for the long side vs. the short side are RADICALLY different. Some who consider themselves purists may think this is heresy, but I challenge you to think about it: There are very good, very logical reasons for why downward moving markets might behave differently than upward moving markets. Especially some kinds of markets. If you are testing across long periods of time and a diverse basket of markets, you will get a big enough sample to minimize the chance that this is random curve fitting. There's a lot more that could be said, but this should be a great start. Please let me know if you find anything cool, I am always looking for better ideas myself. I don't know everything, this just happens to be an area I have explored in detail.
There are some other general market characteristics that separate the good trades from the so-so trades. But this is an excellent place to start, I think.
Seriously, now, I have given you something of value. Please tell me some trick or useful idea of your own.
Murray Ruggiero
Sponsor
I started this thread a while ago. We discuss that I would walk you though the good and bad of a given idea. Our first attempt was to use a triple moving average crossover as a core to our system. Some of you question why I started with a triple moving average crossover. The reason I did was that many people, when you say moving average systems think of the dual moving average system and think of a triple moving average system as just a slower version of the dual. This is not the case. The triple MA system solves a major problem of the dual MA system. Here is the core problem with the dual system and it can be shown as follows:
Assume the current dominant system is 30 days. If we use a half-length cycle as the short-term system and a full-length cycle as the long term, we create a system which is 180% out of phase with the market when the market is not trending. We buy every top and sell every bottom. The triple MA system adds an additional shorter moving average, which we compare to the shorter moving average from the classic dual system.
If we assume our dual system is using a half and full cycle length MA, then our new shorter MA can be compared to the half cycle one and be used to confirms that the market is not topping at the point the half cycle crosses the full cycle one. If we are in cycle mode then this new shortest term average will turn direction before the half cycle crosses the full cycle, so a trade never happens because the shortest term average changes direction before the other two.We only get the three to line up when we are not in cycle mode. This protects against the major problem with the dual moving average crossover system.
Now I have a question for you. Do you want me to go from start to finish with developing a system , using the triple MA or would you like to study some more core methodologies. I can go though the theory as well as analyzing the optimizations and results of several more methods. We can then finally pick one to use as the core of our system and then finish it up. Let me know what you think.
Assume the current dominant system is 30 days. If we use a half-length cycle as the short-term system and a full-length cycle as the long term, we create a system which is 180% out of phase with the market when the market is not trending. We buy every top and sell every bottom. The triple MA system adds an additional shorter moving average, which we compare to the shorter moving average from the classic dual system.
If we assume our dual system is using a half and full cycle length MA, then our new shorter MA can be compared to the half cycle one and be used to confirms that the market is not topping at the point the half cycle crosses the full cycle one. If we are in cycle mode then this new shortest term average will turn direction before the half cycle crosses the full cycle, so a trade never happens because the shortest term average changes direction before the other two.We only get the three to line up when we are not in cycle mode. This protects against the major problem with the dual moving average crossover system.
Now I have a question for you. Do you want me to go from start to finish with developing a system , using the triple MA or would you like to study some more core methodologies. I can go though the theory as well as analyzing the optimizations and results of several more methods. We can then finally pick one to use as the core of our system and then finish it up. Let me know what you think.
A couple points - I'm kinda into a "finish what you started" view on life.
If the idea is to give an example of the methodoly on creating a system - then the core is less important than the how to create idea.
Perhaps after the creation of one idea you could speak to various core ideas.
That and I am kinda biased as I have something simular that I am currently watching.
Thanks for your efforts here.
Chris
lWell, some of us like myself are probably more concerned with certain specific core concepts and this might generate more ad hoc excitement and participation. I myself am interested in exits mainly, strategy evaluation and the intricacies of system optimization.
HOWEVER, since we're at different points of development, perhaps it might indeed be better to continue as before, step by step, with system development, though some of us may fall asleep at the wheel.
Still, I suppose we're all beginners here in one form or another, or at least curious, else we wouldn't be hear and would be making money off the markets or system design as you have done, Murray. So, proceed as you had intended, and we'll be patient until you get to our respective areas of interest...and bombard you with questions. Thank you.
HOWEVER, since we're at different points of development, perhaps it might indeed be better to continue as before, step by step, with system development, though some of us may fall asleep at the wheel.
Still, I suppose we're all beginners here in one form or another, or at least curious, else we wouldn't be hear and would be making money off the markets or system design as you have done, Murray. So, proceed as you had intended, and we'll be patient until you get to our respective areas of interest...and bombard you with questions.
Thank you.
Murray Ruggiero
Sponsor
Feel free to ask questions. , If I can answer them without violiating any NDA's or agreements with clients I will answer them.
Murray Ruggiero
Sponsor
I have decided how I am going to handle this , I am going to cover one more core method on this round so we can do a compare and contrast. Then we will build our system.
We can address many other core methodologies later, but we need at least one more to make it work the way I have decided to finish this.
One question I have is since this thread is 18 pages I was wondering if I should start a new thread and just copy a few important entries of mine so the thread is not so long to follow for new people or continue on this thread ?
I also have an idea for a new poll, which will ask about core methodolgies used by the 60%+ of people who say they build their own systems. Since this is a basket system thread , I will limit the core methods to basket type trading.
Let me know what you think.
We can address many other core methodologies later, but we need at least one more to make it work the way I have decided to finish this.
One question I have is since this thread is 18 pages I was wondering if I should start a new thread and just copy a few important entries of mine so the thread is not so long to follow for new people or continue on this thread ?
I also have an idea for a new poll, which will ask about core methodolgies used by the 60%+ of people who say they build their own systems. Since this is a basket system thread , I will limit the core methods to basket type trading.
Let me know what you think.