Longest time of success on a single system

Beginner Questions
1

Hi people,

What is the longest time of success you have had on a single system without changing any of it’s core variables. This could include both forward live testing, and back-testing over historical data. I don’t mind if you trade EA’s as any logical manual method is a mirror image to an EA - it’s just not programmed into a script.

I’m curious to see if trading systems really do break-down over time, or if it’s simply down to the variables which are implemented into the system to adapt to changes in market conditions.

The point i’m trying to arrive at is this:

In mid 2008 we saw cable slide a cool 6,300 pips, would your current trading method survive that financial crash… it’s all very well turning profits now while the markets are considerably more stable than in 2008, but surely if you cant cope with a market crash we’re all sitting on a ticking time bomb of a trading method.

The smart traders will of course react quicker than the novice trader when seeing a market sliding. In hindsight we can all look back on the charts to 2008 and say “it was an obvious crash”, but i bet your last dollar you would have kept taking buy signals all the way down to 1.3500 from an original high of 2.0000.

Just some food for thought i guess. The reason for this little random post - don’t focus inside the box and always test extremes because you can bet my last dollar that a crash will happen again, it’s just a matter of time.

Jezz.

2

[QUOTE=“Jezzode;486363”]Hi people,

What is the longest time of success you have had on a single system without changing any of it’s core variables. This could include both forward live testing, and back-testing over historical data. I don’t mind if you trade EA’s as any logical manual method is a mirror image to an EA - it’s just not programmed into a script.

I’m curious to see if trading systems really do break-down over time, or if it’s simply down to the variables which are implemented into the system to adapt to changes in market conditions.

The point i’m trying to arrive at is this:

In mid 2008 we saw cable slide a cool 6,300 pips, would your current trading method survive that financial crash… it’s all very well turning profits now while the markets are considerably more stable than in 2008, but surely if you cant cope with a market crash we’re all sitting on a ticking time bomb of a trading method.

The smart traders will of course react quicker than the novice trader when seeing a market sliding. In hindsight we can all look back on the charts to 2008 and say “it was an obvious crash”, but i bet your last dollar you would have kept taking buy signals all the way down to 1.3500 from an original high of 2.0000.

Just some food for thought i guess. The reason for this little random post - don’t focus inside the box and always test extremes because you can bet my last dollar that a crash will happen again, it’s just a matter of time.

Jezz.[/QUOTE]

I’ve done significant back testing of mechanical systems… The few of the best would be reasonably profitable from 2009-present but would be flat or negative on data pre 2007… There was a definite structural shift in the currency markets during 2007-2009 and I am firmly convinced that it is impossible to have a static, mechanical system that can be relied on to perform well through all market conditions.

In the case of the currency crashes such as the Gbp during 2008… A logical question is: do I want to be trading based on a solid understanding of the fundamental concepts behind market moves? Or do I want to rely on chart patterns and indicators…

There were huge fundamental reasons for the GBP fall of thousands of pips without much of a retracement… And these would be fairly obvious to a fundamental trader… The traders losing money during that time period were the ones using stochastics and RSI. They would be screaming at their computer screen as the gbp continue to fall despite the indicators saying they were “oversold”… The price action traders would be watching support line after support line get destroyed before they could even draw their magical fibo levels…

The longevity of a trading system seems to depend on the flexibility and degree of dynamical inputs that are used to generate a decision… Something that is inherently opposite to what a rigid mechanical system embodies.

3

The less rules the more robust a trading system becomes. By robust, it means the survivability among various timeframes, markets and market environments.

To determine the length of a backtest, you should consider the degrees of freedom.

4

[QUOTE=“ClarkFX;486367”]The less rules the more robust a trading system becomes. By robust, it means the survivability among various timeframes, markets and market environments.

To determine the length of a backtest, you should consider the degrees of freedom.[/QUOTE]

Yeh… In terms of rigid mechanical systems, the ones that I have tested that are most stable over the longest period of time are the ones with the fewest rules. One system in particular only had one rule for entry… But even that one could not be profitable in the time before 2008 AND after… Could only do one or the other.

It is of my opinion that every mechanical system has an expiration date.

5

Yes automated strategies rarely work over a long time frame which is why some of the EAs (that work), keep updating their scripts and of course, as to whether they work is another topic altogether.
The point being that market changes over any given period and for any strategy or working model to remain efficient, it has to evolve…

6

I am talking about any set of trading rules. Whether the system is profitable or not makes no difference. The fact is that the more robust the rules are the more “consistency” the model exhibits. Visually, you’re able to see that the equity curve does not have multiple “personalities”. What I mean by this is that the system will win/loss money consistently in its test period.

And yes, I agree, no mechanical system can last forever; even the Turtle’s Donchian system has failed. :wink:

7

[QUOTE=“ClarkFX;486375”]

I am talking about any set of trading rules. Whether the system is profitable or not makes no difference. The fact is that the more robust the rules are the more “consistency” the model exhibits. Visually, you’re able to see that the equity curve does not have multiple “personalities”. What I mean by this is that the system will win/loss money consistently in its test period.

And yes, I agree, no mechanical system can last forever; even the Turtle’s Donchian system has failed. ;)[/QUOTE]

What determines if the rules are “robust”?

Would almost be a chicken and egg thing where which comes first… You would need the stable equity curve to call the rules robust… Or would you be able to call the rules robust even if the equity curve was all over the place.

8

I wish I had an answer for you Jezz but in the few years I have been trading I have constantly tried to improve my system. By adding and taking away pairs that were hurting my win/loss ratio. My core system based on price action has pretty much stayed the same. But even that I have had to make some teaks to. I used to only trade the 50% retrcement of a pinbars wick. This year pinbars on my pars just have not setup like they used to. So now I had to adapt.

It is my opinion though that your system must be able to adapt to ever changing markets.

I will say though if I employed the same strategy of only shorting the AU to that crash I would be rich today :smiley:

9

Surly robust rules are rules which have a high degree of signification when assuming results follow a normal distribution pattern…therefore the probability can be calculated of a positive outcome

10

I would say there are no concrete rules to make a trading model robust or not. There is no either/or; I see it more as a spectrum. The degrees of freedom concept I mentioned in my earlier post is one of the closest ways we can try to quantify robustness.

At the end of it, it comes down to experience and personal judgement.

11

[QUOTE=“Jezzode;486378”]Surly robust rules are rules which have a high degree of signification when assuming results follow a normal distribution pattern…therefore the probability can be calculated of a positive outcome[/QUOTE]

So could we just exchange the term “robust” for “profitable”? It’s easier to quantify profitability then it is robustness…

12

Well I would assume one would only want robust rules which are profitable as that’s the aim of the game :wink:

13

[QUOTE=“Jezzode;486382”]Well I would assume one would only want robust rules which are profitable as that’s the aim of the game ;)[/QUOTE]

You would not want profitable rules that weren’t robust?

14

Robust does not necessarily mean profitable. Robust losing systems exist. :stuck_out_tongue:

15

No, of course not as that would equal luck, and luck runs out

16

Which adds the elements of personal skill, desire, and ability, that to me make this a really awesome endevour.

17

Ha, I feel we’re going around in circles lol

18

[QUOTE=“Jezzode;486386”]No, of course not as that would equal luck, and luck runs out[/QUOTE]

How do we know that the “robust” rules will not lose profitability though…

19

We don’t - but we do know that they are “robust” rules

20

[QUOTE=“Jezzode;486388”]Ha, I feel we’re going around in circles lol[/QUOTE]

Lol it seems like a circular reference error in Excel where the formula in one cell is using the value of another cell which is using the value of the original cell.