Developing trading systems

Quote from jedwards:

Question: How do you determine when a system that used to work, no longer works and should not be used again? What criteria besides number of losses should you use to determine that it is no longer effective?

Basically, how do you differentiate between a drawdown and a broken system?
More...

You should discard it if the drawdown reaches more than 2x historical drawdown, and the drawdown was for a period of several months. Some people say 3x historical drawdown but I say 2.
 
Quote from Murray Ruggiero:

For example my intermarket divergence systems start as four line systems. I will show you an example soon which has worked since the 1990's and can be written in four lines.
More...

If you can make money with it then why are you working at Traders Studio? Or are you the founder of Traders Studio?
 
Quote from sneakoner:

If you can make money with it then why are you working at Traders Studio? Or are you the founder of Traders Studio?
More...

I built TradersStudio as the system development platform to base my models for my money management business. I developed and designed the product to meet my needs as well as responding to the request of our customers.

We sell TradersStudio and our other products to offset the cost of development and research.

I am chief system designer and analyst at Tuttle Wealth Management where we manage about one hundred million dollars currently. We use TradersStudio to implement our trading models.
 
Quote from Murray Ruggiero:

I built TradersStudio as the system development platform to base my models for my money management business. I developed and designed the product to meet my needs as well as responding to the request of our customers.

We sell TradersStudio and our other products to offset the cost of development and research.

I am chief system designer and analyst at Tuttle Wealth Management where we manage about one hundred million dollars currently. We use TradersStudio to implement our trading models.
More...

Nice, I just read your bio on your firm's website. Here it is for anyone else interested:

http://www.matthewtuttle.com/custom.php


Interested in learning more about what you have to say about system development.
 
Quote from jedwards:

Question: How do you determine when a system that used to work, no longer works and should not be used again? What criteria besides number of losses should you use to determine that it is no longer effective?

Basically, how do you differentiate between a drawdown and a broken system?
More...

You need to ask a more important question. Not if the system no longer works but was it ever robust. Things like turning a system off because the drawdown doubles/Triples over a relatively short timeframe is a function of curve fitting.

First drawdown of a system should increase with the square root of time. I could expect a system which has not been reoptimized to have double or triple it's original drawdown over 10 years after release.

If a system is robust and you understand it's premise then you stop trading it when the premise changes. For example Index systems buying the dips worked from the 1980's though 1999, except for a few periods that increase drawdown.

Let's look at this this buy the dip premise. During this period this premise was valid , the market remained above the 200 day MA most of the time, except for nine months in 1987 , we never spend more than a month or two below the 200 day moving average. Starting from 9/2000 to 4/2003 we were only above the 200 day moving average for a few weeks. this to me means that we can just stop trading a buy the dip system when we remain under the 200 day moving average for more than several months because the premise has changed. This would of turn the system off in 1987-1988 crash and from 1/2001 till 4/2003. We also would of turned the system off from 3/2008 until October 2009.

I am not saying this is a great system but an example of a system based on a premise which we can turn on and off and know when our premise is no longer valid, either right now or longer term. After the 2000-2002 lessons we might even added a 200 day moving average turnoff to this strategy.
 
Quote from Murray Ruggiero:

You need to ask a more important question. Not if the system no longer works but was it ever robust. Things like turning a system off because the drawdown doubles/Triples over a relatively short timeframe is a function of curve fitting.

First drawdown of a system should increase with the square root of time. I could expect a system which has not been reoptimized to have double or triple it's original drawdown over 10 years after release.

If a system is robust and you understand it's premise then you stop trading it when the premise changes. For example Index systems buying the dips worked from the 1980's though 1999, except for a few periods that increase drawdown.

Let's look at this this buy the dip premise. During this period this premise was valid , the market remained above the 200 day MA most of the time, except for nine months in 1987 , we never spend more than a month or two below the 200 day moving average. Starting from 9/2000 to 4/2003 we were only above the 200 day moving average for a few weeks. this to me means that we can just stop trading a buy the dip system when we remain under the 200 day moving average for more than several months because the premise has changed. This would of turn the system off in 1987-1988 crash and from 1/2001 till 4/2003. We also would of turned the system off from 3/2008 until October 2009.

I am not saying this is a great system but an example of a system based on a premise which we can turn on and off and know when our premise is no longer valid, either right now or longer term. After the 2000-2002 lessons we might even added a 200 day moving average turnoff to this strategy.
More...

Very good answer to this question Murray.

The only thing I would add is that the premise, or fundamental system concept is key here.

For example, if you're trading an mean reversion system, volatility will impact your drawdowns significantly. Hence, your system will display some element of correlation to the VIX, or, some other general volatility measure. If, in the last two weeks for example, you saw a much larger than average drawdown, it may not necessarily be a problem - its the nature of the trade.

Determining the root cause of drawdown can be done by solid research. Did your drawdowns correlate to volatility or being under the 200 MA (as Murray mentioned), in the past? Or, do you not know what factor was responsible for the drawdown during that historical period? In the later case, I would say that you re-evaluate your fundamental premise and make sure its sound.

Again, developing a good system starts with a good fundamental trading idea...
 
Quote from Mike805:

Very good answer to this question Murray.

The only thing I would add is that the premise, or fundamental system concept is key here.

For example, if you're trading an mean reversion system, volatility will impact your drawdowns significantly. Hence, your system will display some element of correlation to the VIX, or, some other general volatility measure. If, in the last two weeks for example, you saw a much larger than average drawdown, it may not necessarily be a problem - its the nature of the trade.

Determining the root cause of drawdown can be done by solid research. Did your drawdowns correlate to volatility or being under the 200 MA (as Murray mentioned), in the past? Or, do you not know what factor was responsible for the drawdown during that historical period? In the later case, I would say that you re-evaluate your fundamental premise and make sure its sound.
More...

See, it might not be the cause of the drawdown per say, but the fact underlying market footprint has changed is a stop sign.
Another example was when they change the big point value of the S&P500 from $500.00 to $250.00 , if a system drawdown increased greatly, starting right after this and the premise was related to risk then it might not work anymore. This happened to a day trading S&P500 system I developed based on Taylor book method in 1996, which broke after they cut the big point value. Market reversals just past the previous high or low did not create tradable turning points because cutting the point value reduced risk.
 
Quote from Murray Ruggiero:

... Another example was when they change the big point value of the S&P500 from $500.00 to $250.00... Market reversals just past the previous high or low did not create tradable turning points because cutting the point value reduced risk...
More...
How did it reduce risk?
 
Quote from abattia:

How did it reduce risk?
More...

Simple technical stop, like 60% of 3 day average range is now half the dollar value. Daily range in dollars 1/2
 
Murray, I think TS is adding machine generated trading system capability to their software. See, most people don't have the skills to go through the manual development steps. Integrating machine generation of trading systems, even something simpler and similar to price action lab functionality could add value to a trading platform. It could increase audience dramatically because people who cannot program will be able to use the platform to develop trading systems. I think out of 100 traders only 1 or maybe 2 at most can effectively develop systems. The potential is great when this is done internally and it is offered as a feature.
 
You must log in or register to reply here.
Back
Top