Developing a mechanical system from scratch

[sinner]

You know, this could explain why i had 0 trades when i was just using price patterns and volume....

Heres some of my options....

You gotta be careful what you plug into it of course, many of the indicators you list suffer from the same 'un-normalised' problem which gives the optimality you encountered.

It really really helps to think in terms of genetics in the case of genetic algorithms (surprise).

If the environment which the species exists in is highly specific, then selection/adaptation will be fast to stabilise but even a small incremental change to the environment can prove catastrophic.

On the other hand species which have traits which allow generalised access to a range of environments can continue to exist in the face of incremental changes.

 

[CanOz]

You gotta be careful what you plug into it of course, many of the indicators you list suffer from the same 'un-normalised' problem which gives the optimality you encountered.

It really really helps to think in terms of genetics in the case of genetic algorithms (surprise).

If the environment which the species exists in is highly specific, then selection/adaptation will be fast to stabilise but even a small incremental change to the environment can prove catastrophic.

On the other hand species which have traits which allow generalised access to a range of environments can continue to exist in the face of incremental changes.

Well perhaps i should just work my way through the indicators using only a few at a time. I can also change the population size, the number of generations, Crossover and mutation, tree depth and tournament size.

CanOz

 

[CanOz]

Taking a break from developing a DAX system using the Builder today. There are several strategies that come with the builder and today i'm going to test them on MultiCharts, and if they're profitable and meet my OF then I'll trade them on IBs Paper Trader.

Zero strategies were found yesterday. It seems more difficult to find useful systems.

Getting a bit discouraged about using Builder. On a positive rote the EUR.USD stratgy seems quite profitable on the most recent data.

Its a daily system....recalling what Radge said about time frames....intra-day is very difficult.

CanOz

 

[motorway]

Taking a break from developing a DAX system using the Builder today. There are several strategies that come with the builder and today i'm going to test them on MultiCharts, and if they're profitable and meet my OF then I'll trade them on IBs Paper Trader.


Its a daily system....recalling what Radge said about time frames....intra-day is very difficult.

CanOz

Consider that maybe;

The market reality is in fact built from the momentary meeting of a quantity of demand and supply. A buyer and seller meeting across the Bid Ask Spread.. This does not occur as a linear function of time so your use of time frames can only introduce noise and distortion. At the smallest time scales this might will become a larger component in proportion..

This paper might interest you.


http://www.newton.ac.uk/programmes/DQF/seminars/070714501.pdf


From that smallest market unit.. Either Supply or Demand is Momentarily Exhausted to a larger or smaller degree... From this building Block are all the trends ( degrees ) composed.

But the largest active trend ( and it's momentum ) is the container from where the arrow of casulality flows.

So the smallest unit is the building block
But the causation flows from the large to the small in terms of trends

There is no contradiction here

It is obviously so when you do away with the time frames and see everything occurring in the moment.

At the bid/ask spread is the only place where buyer can meet seller.
This is a particular moment ... The same moment no matter what those particular buyer and seller individual time horizons may be.

Motorway

 

[CanOz]

thanks Motorway....love ya brother!

Here's a daily system for the ES. Nice win rate but it needs a MM stop, hehe, as you can see. It not going to hurt the system much either going by the excursions. I'd say about $5000 stop should do it. This is trading 5 contracts.

CanOz

 

[CanOz]

Here's a pattern that i found using a show me study. I was looking for another pattern at the time. Two of these instances where profitable on the 15 min DAX recently. Once i have it coded up I'll test it over a few years of data and eyeball the chart to see if its worth pursuing.

This is my practice for learning Easy Language.....nothing to do with code is easy for me!

CanOz

 

[CanOz]

I finally have my DAX system coded and i've been testing most of the day. It doesn't take allot of trades on the DAX, about 30 per year, but on the HSI and SPI it trades almost every day. Here are three markets and the results. In order, DAX, HSI, SPI. This is a very simple system with a money management stop, and the exit is end of session.

For being quite raw at the moment, it seems quite robust.

CanOz

 

[CanOz]

Here are the equity curves, same order DAX, HSI and SPI. 3, 5, and 2 years of data.

CanOz

 

[Joules MM1]

thumbs up......kudos to you for the work thus far, can

 

[CanOz]

thumbs up......kudos to you for the work thus far, can

Thanks! They're a bit raw right now, so the curves are a little rough, but trade-able IMHO.

CanOz

 

[Lone Wolf]

You said the DAX system trades just 30 times a year. A question I've been asking myself is how many trades per year do you need in order to have confidence that the system is robust? It seems to me that if the trade frequency is too low then it makes any back tested stats less reliable. Any thoughts on this?

I see your equity curve for the HSI has a couple of big dips in there. For example the $125,000 drop from $175,000 down to $50,000. Would you be able to keep trading the system through that kind of drawdown? Also, what would've happened if you had been unlucky enough to start trading at the time where the drawdown first started? Would the account have been blown or would the position size have been smaller and therefore less drawdown?

To me it makes a big difference when you encounter drawdown. I could easily sit through a 30% drawdown after making large gains. But if I were to go into max drawdown right off the bat then I may start to panic that the system is broken and not have the nerve to keep trading the system. So when I consider my drawdown tolerance I need to consider that it could happen right at the start.

 

[CanOz]

All of our points are valid LW. These systems are raw and as such they only have basic money management. With the application of more complex stops and targets the stats will change, but hopefully improve a little. Its all up to the person trading it whether or not they can take the draw-down or not.

One thing is for sure though, these returns cannot be realized unless you stick with the system, good with the bad. Same as buy and hold yeah? Except the draw-downs should be much much less.

That's why many agree its better to build your own system than to trust someone else's.

If you cannot understand how the system makes money then forget about trading it, you'll never make it through the first DD.

Also, many traders use an equity/system curve switch. The system shuts off and turns on based on the equity curve.

CanOz

 

[Lone Wolf]

I fully agree with the concept of understanding why your system works and sticking to it. That's why I find it important to have enough trades to support the results and to ensure the drawdown is something I can tolerate. I first started with a financial adviser in a buy and hold with leverage strategy. Max system drawdown is something I pay great attention to these days.

I've never tried the equity curve switch, my luck I'd always turn it off right before a recovery. But I do use a system filter to keep it out of undesirable conditions, so I guess the concept is the same.

 

[CanOz]

I've never tried the equity curve switch, my luck I'd always turn it off right before a recovery. But I do use a system filter to keep it out of undesirable conditions, so I guess the concept is the same.

A filter, such as an index filter, is there to take you to cash when the broader market turns out of trend. An equity curve switch shuts your system off before it can go underwater too far. There are many examples, one such is the MSA from Adaptrade.

Joe Krutsinger uses one on most of his highly acclaimed trading systems. As does David Bean from Capstone trading systems on his NY Scalper.

Here is a very basic one, a 100 period MA of the equity. You can see how it would be useful. The thing is, some equity curves are mean reverting and some are more trendy.

Cheers,


CanOz

 

[Lone Wolf]

Thanks for that CanOz.

I've been playing with some amibroker code for an equity switch this afternoon and it may be useful. But I'd rather design a system with an equity curve that doesn't need an off switch.:

 

[CanOz]

Thanks for that CanOz.

I've been playing with some amibroker code for an equity switch this afternoon and it may be useful. But I'd rather design a system with an equity curve that doesn't need an off switch.:

Yeah me too, but they might be handy anyway! Sort of like a condom i.e. - better to have one and not need it, than need one and not have it.

CanOz

 

[tech/a]

Yeah me too, but they might be handy anyway! Sort of like a condom i.e. - better to have one and not need it, than need one and not have it.

CanOz

Have a look at turning your system on and off using YOUR systems equity curve.

 

[sinner]

I use this technique. It used to be difficult to play along, now there is an ETF for almost everything, it is quite easy.

http://www.cxoadvisory.com/43/big-ideas/beat-the-market-with-hot-anomaly-switching/
http://www.cxoadvisory.com/54/big-ideas/diversifying-across-equity-anomalies/

All my systems have a switch, sometimes implicit, sometimes explicit which stops the system trading if the market regime is not appropriate for that sort of trading.

 

[tech/a]

I use this technique. It used to be difficult to play along, now there is an ETF for almost everything, it is quite easy.

http://www.cxoadvisory.com/43/big-ideas/beat-the-market-with-hot-anomaly-switching/
http://www.cxoadvisory.com/54/big-ideas/diversifying-across-equity-anomalies/

All my systems have a switch, sometimes implicit, sometimes explicit which stops the system trading if the market regime is not appropriate for that sort of trading.

So you follow anomolies in EFTs and use them as your filter?
Can you elaborate more on this.?

 

[sinner]

So you follow anomolies in EFTs and use them as your filter?
Can you elaborate more on this.?

tech, I am really not even sure what you mean? What you wrote sounds like gibberish.

The links are an example of tracking momentum for a set of trading strategies and trading the ones with high momentum while ignoring those with low momentum. The ETFs track the anomaly, I am not looking for new anomalies in them. Seems you are having trouble with the lingo, so I will make it clear: an anomaly is just a market phenomenon that is not explained by efficient market hypothesis. For example, some well known anomalies are momentum, value and liquidity.

Tracking the momentum of a set of ETFs which tracks a set of anomalies is exactly the same as "turning your system on and off using YOUR systems equity curve" except across multiple systems.

Instead of turning off the system and going to cash, I turn off one system and turn on another one which the new market regime is suited for. I don't choose, I just follow the momentum of the set of systems.