Optimization of a trading system with avoidance of curve fitting.

Quote from albertly:

But what do you do not to fall into curve fitting trap?

You can measure predictive power of optimization with walk forward test.

Actually I think that only technology can make drastic changes in basic market behaviour. For example internet in 1999-2000 has very big impact on all trading industry - new crowd involved, new method of trading, discount brokers etc.
Computers earlier made big impact on involving more people in trading.

Besides of that - nothing new - up trend, down trend, side walk - period of big volatility, low volatility- nothing new under the sun.

I talk about stock market only as I have no judgment about Forex, futures, etc because it is out of my interest.
More...

fwiw- I second the use of walk forward testing. I like to have at least 50 trades in the base optimized period and 25 in the walk forward period. Jumping from a data set to trading without a test in a period outside the optimized data set is a invitation to losing your capital.

Seneca
 
Quote from intradaybill:
If your system has even one parameter which you adjust based on some criteria, then this is a fitted system and it will fail. [/B]
More...

I'd be curious to know as well as to why you disagree with optimization? If you are a technical trader, I dont see a difference between handpicking your parameters and timeframes versus optimizing them.

Yes, curvefitting is a problem - in particular if your degrees of freedom are very large, but there are a couple of feasible remedies IMHO and I pretty much agree with Joman's list.

On top of that it also depends on the type of data you are using. The coarser the granularity the more of a problem you have. Roundturns are also a good indicator - the less the more likely that you have a curvefit and big draws in your forward tests or live trading.

Another issue is what assumptions you make in regards to entry and exit points which also is strongly related to the data you are using. The finer the granularity, I find, the better your assumed entries and exits match reality.

I don't know how others do it, but I do not use the price curve alone to calibrate a candidate system.
 
Quote from intradaybill:

If you already know, why are you asking?

If your system has even one parameter which you adjust based on some criteria, then this is a fitted system and it will fail.
More...


Cool, thanks for telling me this! My systems have multiple parameters and produce double digit returns with real money.
Perhaps you should tell the same thing to many 1B+ funds, making 10-20%/yr who use parameters?

the truth is, any single strategy with or without parameters can fail.
for some reason, there are folks who claim their approach to trading will not fail, while others methods are doomed.
 
Quote from intradaybill:

Buy if C[0] > H[1]
More...

I might be wrong but I consider that your system has few parameters.

- C is a parameter you can adjust: you can also use O, H, L.

- 1 period ago is a parameter you can adjust: you can use x period.

- ">" is a parameter you can adjust: you can use "<" or "=".

Price patterns can also have "optimizable" parameters !
 
Quote from Joman:

I might be wrong but I consider that your system has few parameters.

- C is a parameter you can adjust: you can also use O, H, L.

- 1 period ago is a parameter you can adjust: you can use x period.

- ">" is a parameter you can adjust: you can use "<" or "=".

Price patterns can also have "optimizable" parameters !
More...

Joman does not know or understand the following:

1 - the difference between a parameter and a math operation

2. The difference between a non-adjustable and an adjustable parameter

3. The difference between a qualifier and a parameter
 
Quote from Joman:

I might be wrong but I consider that your system has few parameters.

- C is a parameter you can adjust: you can also use O, H, L.

- 1 period ago is a parameter you can adjust: you can use x period.

- ">" is a parameter you can adjust: you can use "<" or "=".

Price patterns can also have "optimizable" parameters !
More...

You are completely right, there are few parameters in this decission rule and all they can be optimized. And are optimized in the so called price action based trading systems. Price action based trading is now a new hype after the magic indicators ceased to be magic.

A really funny part of the story is that the guy who's posting this has no clue what he is talking about.
 
Quote from intradaybill:

Joman does not know or understand the following:

1 - the difference between a parameter and a math operation

2. The difference between a non-adjustable and an adjustable parameter

3. The difference between a qualifier and a parameter
More...

Joman didn't understand this, thanks for the clarification.

I now understand how intradaybill avoids curve fitting. :)
 
You must log in or register to reply here.
Back
Top