|
PureBytes Links
Trading Reference Links
|
I think a distinction needs to be made between optimization and
curve-fitting. Optimization is valid and necessary if you use indicators of
any sort. Most indicators require a period parameter and a value of 2 or 50
or 300 will give very different results. I don't think that anyone would say
any arbitrary value is good. You test and pick a value that works well for
what you are trying to accomplish. The key distinction is that if you decide
on a value of 50 then 48 or 52 should work just as well. If you test a
system and find that a value of 38 works significantly better than 36 or 40
then you are curve-fitting and you won't get the same results in the future
-----Original Message-----
From: Paul Brittian <tradeblt@xxxxxxx>
To: RealTraders Discussion Group <realtraders@xxxxxxxxxxxxxx>
Date: Tuesday, July 07, 1998 11:11 PM
Subject: Re: GEN: Optimization IS a four letter word!
>Optimization should be banned, even though past trading patterns can be
>used as a barometer of the future, optimization trading systems will not
>and cannot repeat their past performance in real time.
>Paul Brittain
>
>GREHERT@xxxxxxx wrote:
>>
>> Let's stop condemning optimization!
>>
>> I'd like to come to the defense of optimization. I use optimization in
>> developing trading systems that trade stocks and stock market indexes.
>> Optimization let's you know what parameters worked the best historically.
And
>> if you believe history is the best evidence for the future, then
optimization
>> gives you the future's best parameters.
>>
>> What optimization doesn't give you is a valid prediction of future
returns.
>> The crime of optimization is when hucksters claim future profits should
>> approach optimized profits. In reality, future profits will probably
fall
>> some where near the middle of your range of profits that result from your
>> parameter array.
>>
>> Best Regards,
>>
>> Jerry Rehert (grehert@xxxxxxx)
>> Atlanta, GA
>> @ 11:47 pm, July 7th, 1998
>
|