How long should you test a strategy? | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

How do you test if the pair is in a medium/long term trend? 2 replies

How long should I test this system? 20 replies

Why EA test is forward test better than back test?? 1 reply

How long should you test a system for? 17 replies

How long do you forward test a discretionary system? 7 replies

Attachments: How long should you test a strategy?
Exit Attachments

How long should you test a strategy?

  • #1
  • First Post: Dec 15, 2011 3:56pm Dec 15, 2011 3:56pm
Hi guys, how long do you test a strategy in order to ensure it is consistent and profitable?

3 months, 6 months, 1 year...etc?
  • #2
  • Dec 15, 2011 4:15pm Dec 15, 2011 4:15pm
I wouldnt worry about that. If its a static system then its doomed, dynamic strategies would work better but they are the ones that can't be tested with info from price action alone.
  • #3
  • Dec 15, 2011 4:40pm Dec 15, 2011 4:40pm
If you give a test a long enough period (lets say 1 year) to prove consitent, you will impliment it just as conditions change.

Then back to the drawing board we go
  • #4
  • Dec 15, 2011 4:44pm Dec 15, 2011 4:44pm
Quoting dvc11
Hi guys, how long do you test a strategy in order to ensure it is consistent and profitable?

3 months, 6 months, 1 year...etc?
Test it as much as possible. Like more than 500 trades. Because I recently was testing using Forex tester and the first 3 months I made about 25-30% consistently each month with about 200 trades but the next 3 months, the equity curve is in a range where it goes up and down 20% and i just want to abandon the system. I thought 200 trades is enough but i found out its not with my own experience.
  • #5
  • Dec 15, 2011 4:45pm Dec 15, 2011 4:45pm
Quoting EmeraldEyes
I wouldnt worry about that. If its a static system then its doomed, dynamic strategies would work better but they are the ones that can't be tested with info from price action alone.
What's a static system? One where the rules are the same all the time?
  • #6
  • Dec 15, 2011 5:25pm Dec 15, 2011 5:25pm

Testing systems :nerd:

When you create a dynamic entry and dynamic exit...
You then can imagine to put that system in the worst case scenario...
You then will know, that if your system can survive the worst and unimaginable scenario...
you then will know you have a winning system.

When you understand market structure...
there is no such thing as testing it for months and years.
If you need a computer to test your system...
More likely, you don't have an intimate understanding of your system and the market.
You are like a confused car driver that couldn't start your car, not knowing that the gear stick is on drive.




  • #7
  • Dec 15, 2011 5:37pm Dec 15, 2011 5:37pm
Quoting dvc11
Hi guys, how long do you test a strategy in order to ensure it is consistent and profitable?

3 months, 6 months, 1 year...etc?

it all depends on the type of inefficiency that the strategy exploits in the market.
  • #8
  • Dec 15, 2011 5:50pm Dec 15, 2011 5:50pm
Quoting misterboy
What's a static system? One where the rules are the same all the time?
Yep.

Agreed CP. Good post.
  • #9
  • Dec 15, 2011 6:27pm Dec 15, 2011 6:27pm
Quoting misterboy
Test it as much as possible. Like more than 500 trades.
I have had the same experience as misterboy....
200-300 trades over a 3 month period is not enough...500 trades spanning more time is better.

Though, I do not backtest so much as review my own live trade results and make adjustments to market conditions.
(I scalp short-term.)
  • #10
  • Dec 15, 2011 7:48pm Dec 15, 2011 7:48pm
5-10 years of data across 3-5 different markets.
Don't wish it were easier, wish you were better. ~ Jim Rohn
  • #11
  • Dec 15, 2011 7:58pm Dec 15, 2011 7:58pm
Quoting LasVahGoose
5-10 years of data across 3-5 different markets.
In the past centuries, any market looked like this...

Attached Image


Today's market looks like this...

Attached Image


In the future, centuries from now, the market will look like this...

Attached Image



Market never changes, never have, never will.

  • #12
  • Dec 15, 2011 8:35pm Dec 15, 2011 8:35pm
CRUCIAL POINT

i want to thank you, your ideas have inspired me to dig deeper, you along with FTI are the 2 more influential people in my trading.

thanks to you, i have reached a degree of consistency.


P(X=1)=1/2 you are completely right.
F=(m)(v)
and you can actually back test your system but people back test the wrong part of their system. The analysis part, you have to have an understanding of the underlying forces that force the influx of money and understand how to recognize changing conditions.

however the back testing through the worst possible scenarios can be applied by a random chart generator and that is the MONEY MANAGEMENT strategy.

if you can deal with completely random variables and not be broke then the market being an information driven system.

its a little easier to survive.
AVT INVENIAM VIAM AVT FACIAM
  • #13
  • Dec 16, 2011 3:42am Dec 16, 2011 3:42am
Quoting dvc11
Hi guys, how long do you test a strategy in order to ensure it is consistent and profitable?

3 months, 6 months, 1 year...etc?
Hello there,
First of all there is no precise time to test and also word test measn nothing.You need to practice a lot 3-5 years all year one strategy to see how will your strategy react on current financial moment,every year we got some new finance moments.The more you practice the results will be better (improving skills)...

And for the end,
There is no guraantne that if you finished one year in plus that second year will be also plus,remember that,this is Forex and big volotility.Risk small and trade with your rules...

Cheers
Petar
"Buy on fear, sell on greed !!!
  • #14
  • Dec 16, 2011 3:49am Dec 16, 2011 3:49am
Quoting dvc11
Hi guys, how long do you test a strategy in order to ensure it is consistent and profitable?

3 months, 6 months, 1 year...etc?
Optimization it takes place in 3 faces
Up-trend, Down-trend, side- ways
By looking back in your charts on deferent assets
and locate these years then you will back test each of them separately
For one year in Daily or hourly time frames
  • #15
  • Dec 16, 2011 3:58am Dec 16, 2011 3:58am
Quoting CrucialPoint
In the past centuries, any market looked like this...

Attachment



Today's market looks like this...

Attachment



In the future, centuries from now, the market will look like this...

Attachment



Market never changes, never have, never will.

Is not correct volatility it has change on our days
Because of the computers and because the markets is widely available
An example Richard Dennis and his turtle the made huge amount of money
Back in 80’s using Donchians channels (20 and 50 periods break) the same strategy had been back tested by us from1970-2010 the results from 1970-2000 was profitable(for each decade) from 2000-2010 was loosing 30% of the equity
  • #16
  • Dec 16, 2011 4:09am Dec 16, 2011 4:09am
As forex market is controlled by variable factors,logic and interpretations it can not be controlled by a single strategy
  • #17
  • Dec 16, 2011 4:15am Dec 16, 2011 4:15am
Quoting fxpatt
As forex market is controlled by variable factors,logic and interpretations it can not be controlled by a single strategy
Exactly the old strategies don’t work any more because the market had change and will change more
Especially now with high frequency trading super computers that are used from large cooperation like Goldman Sachs known as high frequency trading producing such a volatility that the floor starts to complain
  • #18
  • Dec 16, 2011 6:32am Dec 16, 2011 6:32am
Quoting tranco
Is not correct volatility it has change on our days
Because of the computers and because the markets is widely available
An example Richard Dennis and his turtle the made huge amount of money
Back in 80’s using Donchians channels (20 and 50 periods break) the same strategy had been back tested by us from1970-2010 the results from 1970-2000 was profitable(for each decade) from 2000-2010 was loosing 30% of the equity

That's because you guys don't know what you are doing


Quoting tranco
Exactly the old strategies don’t work any more because the market had change and will change more. Especially now with high frequency trading super computers that are used from large cooperation like Goldman Sachs known as high frequency trading producing such a volatility that the floor starts to complain

The ones who complain are the losers, period.
Suck it up Princess or throw in the towel

  • #19
  • Dec 16, 2011 6:37am Dec 16, 2011 6:37am
Many people who state that static systems cannot adjust to changing circumstances fail to understand that a static system CAN take into account dynamic market circumstances.

Maybe I define static differently than the community, but for me, a static trading system is one in which the parameters are predefined and remain constant over time.

As an example, lets say that there are 2 types of markets: trending and ranging. Suppose my system only has an edge in trending markets, then we can program the system so that it detects in which state the market is (using a predefined rule), and to only allow trading when the market is trending.

On topic: the length of time doesn't matter, all that theoretically counts is whether the edge is statistically significant, which therefore is a function of the # of trades, not time (though the 2 are naturally related). The question then becomes: what level of significance are you looking for. You might increase the statistical significance of your edge by increasing the sample size by including a more distant past. There is one drawback of a greater sample: market circumstances may have changed!

Edit: haha, I realize that this sounds as a contradiction in terms. It should be noted that there are two types of dynamic forces in markets: recurring states and fundamental transitions. One should continuously test whether the system's edge hasn't faded due to such a fundamental transition (note that this too can be done within the system, and with fixed and predefined rules).
The nail that sticks out gets hammered back in
  • #20
  • Dec 16, 2011 6:45am Dec 16, 2011 6:45am
Quoting CrucialPoint

That's because you guys don't know what you are doing





The ones who complain are the losers, period.

Suck it up Princess or throw in the towel

You are right back in 1986 when i started i dint knew what i was doing
But I lean through obstacles
0 traders viewing now