Extremely simple strategies with > 100% annual return

[d08]

Quote from Bob111:

9 lines of code and you done? then go for it!
just don't forget to come back after few weeks and tell us how you(and you account) are doing..nah...could be lucky strike..comeback after a year.

i have probably more lines of code than Tolstoy in his war and peace and still far away from 75% win(but enough to make a living off it)

The most robust strategies are simple, the fewer parameters you have, the more certainty you have that's it's not curve-fit.
This doesn't only apply to the trading world.

 

[ssrrkk]

Quote from d08:

The most robust strategies are simple, the fewer parameters you have, the more certainty you have that's it's not curve-fit.
This doesn't only apply to the trading world.

yes a profitable strategy should be simple but more importantly it must not be obvious.

 

[total_keops]

Quote from jcl: This strategy with only 9 lines of code returns about 75% with EUR/USD in a 4 years backtest.

Is this with daily bars?
If so, how do you deal with stops being hit or not intrabar (high, low)? You close stop at the stop price or the close of the bar?

Are stops fixed at the start or trailing?

Are you sure you are not using the current bar as info to trade on it?

I have experience/knowledge with track following algos (mainly Kalman) and I can tell that your results look somewhat too good to be true.
Or is it that you are using a lot of leverage and would run huge drawdowns and a large probability of blowing right of the bat?

 

[ronblack]

Quote from jcl:

This strategy with only 9 lines of code returns about 75% with EUR/USD in a 4 years backtest. However it is not really tradeable, as the Sharpe Ratio is only 0.72.

This isn't 9 lines of code in general. You are using a specific high level language of some specific platform. There are several functions used in the code you posted. It will run only on that platform so I don't understand the purpose of this.

 

[jcl]

Quote from total_keops:

Is this with daily bars?
If so, how do you deal with stops being hit or not intrabar (high, low)? You close stop at the stop price or the close of the bar?

Are stops fixed at the start or trailing?

Are you sure you are not using the current bar as info to trade on it?

I have experience/knowledge with track following algos (mainly Kalman) and I can tell that your results look somewhat too good to be true.
Or is it that you are using a lot of leverage and would run huge drawdowns and a large probability of blowing right of the bat?

It's 1-hour bars. The stop is neither at the stop price nor at the close of the bar, but is calculated from intra-bar ticks to be as realistic as possible. The stops are not trailing but fixed at the start.

I'm sure that I'm not using the current bar. The trade price is the Open of the next bar plus the slippage delay. The leverage is 100:1.

I understand that the filter strategy might look too good, but filters were the only method I found so far with those results. Polynomial regression and pattern detection also gave quite good results. All my attempts with standard indicators failed miserably.

 

[jcl]

Quote from ronblack:

This isn't 9 lines of code in general. You are using a specific high level language of some specific platform. There are several functions used in the code you posted.

This is true, but the platform is free and the functions are not so high level but rather simple, so anyone can reproduce the results. The code for the lowpass filter is from a standard engineering book and available in source code in the platform's header file.

 

[d08]

Quote from ssrrkk:

yes a profitable strategy should be simple but more importantly it must not be obvious.

"Obvious" is very subjective. Sometimes the obvious works too, admittedly it might not seem obvious for others.

 

[total_keops]

Quote from d08:

"Obvious" is very subjective. Sometimes the obvious works too, admittedly it might not seem obvious for others.

Keep in mind that filters are rocket science.

 

[Bob111]

Quote from d08:

The most robust strategies are simple, the fewer parameters you have, the more certainty you have that's it's not curve-fit.
This doesn't only apply to the trading world.

well..i can describe my "system"\core rules in one sentence. that's how simple it is. however..how to get there and how to handle the trade after you are in-is another story.

 

[ronblack]

Quote from jcl:

var *Price = series(price());
var *Trend = series(LowPass(Price,1000));
Stop = ATR(100);

if(valley(Trend)) {
  sellShort();
  buyLong();
} else if(peak(Trend)) {
  sellLong();
  buyShort();
}

How did you choose 100 for the ATR period and 1000 for the low pass filter? Did you optimize?