|
PureBytes Links
Trading Reference Links
|
Larry,
I don't think there is such an animal as a 'best optimization period.' I, too, would like to develop a completly mechanical trading system, but there has to be some human intervention and interpretation somewhere in the development. What it sounds like you want to do is optimize 2 variables over, say, 2000 bars (8 years using EOD data), then, after the program selects the 'best' optimization parameters, run a backtest on ONE NEW BAR, to be followed up with another optimization on the entire data set (now 2001 bars)? Maybe I'm missing something, but how can you do a backtest on one bar? You may be able to buy, but you cannot sell (unless you are daytrading). I really don't see the advantage of what you are trying to do. I don't think you should trust a computer to pick your parameter values for you. That task should be yours and yours alone, IMHO. Also, I think the optimization period should encompass both bullish and bearish periods, and then the user picks the most ROBUST parameter values, not necessarily the ones that give you the highest profitability or lowest drawdown or best PF or highest W/L ratio. The user has to interpret these factors and decide what is best, taking into consideration all those factors simultaneously. The best parameter values are those that give you good results over a broad range of parameter values. If you select a local maximum, you won't get consistent results in time, and eventually you'll lose big.
Al Venosa
>From: "lel4866"
>Reply-To: amibroker@xxxxxxxxxxxxxxx
>To: amibroker@xxxxxxxxxxxxxxx
>Subject: [amibroker] Re: Suggestion for better backtesting/optimization
>Date: Mon, 30 Sep 2002 16:33:45 -0000
>
>I'm missing something then. I want to do an optimization within an
>optimization. I don't see how I can programmatically (without using
>JScript or VB) select the "best" parameters from an optimization run
>and then use them in a subsequent backtest step, for one day only,
>then repeat that for all days available WITHOUT any manual
>intervention...
>
>What I want to do is (assuming we have, say, 2000 data points, and a
>system that has 2 variables that can be optimized),
>find the "best" opt_period:
>
>For opt_period=250 to 1000 by 50
> For begDate = 0 to 2000-opt_period by 1
> endDate = begDate + opt_period
>
> // Find best opt_var1, opt_var2 for begDate to endDate
> For opt_var1 = v1init to v1end by v1step
> For opt_var2 = v2init to v2end by v2step
> result1[] = backtest from begDate to endDate
> End for
> End For
>
> // Decide what to do on endDate+1
> Select opt_var1, opt_var2 from "best" result1[]
> Use chosen opt_var1 and opt_var2 for endDate+1
> End For
>
> Save result2[]
>End For
>list values from result2[]
>
>Larry Lewis
>
>
>--- In amibroker@xxxx, "Tomasz Janeczko" wrote:
> > Larry,
> >
> > As Al wrote, there is no obstacle in adding additional optimization
> > parameter that controls walk-forward test.
> >
> > Best regards,
> > Tomasz Janeczko
> > amibroker.com
> > ----- Original Message -----
> > From: "lel4866"
> > To:
> > Sent: Monday, September 30, 2002 5:30 PM
> > Subject: [amibroker] Re: Suggestion for better
>backtesting/optimization
> >
> >
> > > This is not what I meant -
> > >
> > > I would like to do all of this in an automated fashion where the
> > > look back period and stable parameter period are both
>optimization
> > > variables, and the parameter selection was automatic.
> > >
> > > Larry Lewis
> > >
> > > --- In amibroker@xxxx, "Avcinci" wrote:
> > > > You can do walk-forward testing now with AB. Simply optimize
>from,
> > > say, 1997 to 2000, then using the optimum parameter values from
>that
> > > optimization, backtest on the same stocks from 2000 to present
>(out
> > > of sample data). You can vary the time periods to do your
> > > optimizations and backtests. Of course, this is 2 steps, but it's
> > > not that hard. TJ has said that, later this year, he will be
> > > incorporating in his graphics engine 3-D graphics that enable the
> > > user to view the robustness of an optimization in a manner
>similar
> > > to but better than the Excel add-in that Herman developed and
> > > uploaded earlier. You would choose parameter values from a flat
>area
> > > of the optimization graph where the parameter values and equity
> > > curve would not change much from one set to another.
> > > >
> > > > Al V.
> > > > ----- Original Message -----
> > > > From: lel4866
> > > > To: amibroker@xxxx
> > > > Sent: Sunday, September 29, 2002 12:29 PM
> > > > Subject: [amibroker] Suggestion for better
> > > backtesting/optimization
> > > >
> > > >
> > > > My suggestion is for support for walk forward optimization.
>This
> > > is
> > > > what we all actually do in real life. Since we can't look
>into
> > > the
> > > > future, we test/optimize based on the past, then apply the
> > > results
> > > > to the following days trading. We then look at the results
>and
> > > make
> > > > additional changes.
> > > >
> > > > There are 2 simple variables: 1) The length of time we look
> > > back,
> > > > and the length of time (or other conditions) we wait until we
>re-
> > > > optimize.
> > > >
> > > > There's a 3rd, more complicated variable - how we choose the
> > > > optimium parameters from the last optimization run. My
> > > suggestion is
> > > > for a "score" formula that takes into account things like:
> > > maximum %
> > > > drawdown, best compound rate of return, highest minimum of
>the
> > > > running 1 year returns (or whatever period you like).
> > > >
> > > > I particularly like the last one - I always look at a graph
>of
> > > the 1
> > > > year returns (Equity() - Ref(Equity(), -253)). Ideally, it
> > > should be
> > > > as flat as possible (or maybe rising). I've been treating
>this a
> > > > little like analysis of variance - it should look like white
> > > noise
> > > > with a given mean and standard deviation - any periods that
> > > don't
> > > > bear investingation.
> > > >
> > > > Right now I use AmiBroker for experiments, but when I find an
> > > idea I
> > > > like, I must write my own program to do walk-forward testing.
> > > >
> > > > Larry Lewis
> > > >
> > > >
> > > >
> > > > Yahoo! Groups Sponsor
> > > > ADVERTISEMENT
> > > >
> > > >
> > > >
> > > >
> > > > Post AmiQuote-related messages ONLY to: amiquote@xxxx
> > > > (Web page: http://groups.yahoo.com/group/amiquote/messages/)
> > > >
> > > > Check group FAQ at:
> > > http://groups.yahoo.com/group/amibroker/files/groupfaq.html
> > > >
> > > > Your use of Yahoo! Groups is subject to the Yahoo! Terms of
> > > Service.
> > >
> > >
> > >
> > > Post AmiQuote-related messages ONLY to: amiquote@xxxx
> > > (Web page: http://groups.yahoo.com/group/amiquote/messages/)
> > >
> > > Check group FAQ at:
>http://groups.yahoo.com/group/amibroker/files/groupfaq.html
> > >
> > > Your use of Yahoo! Groups is subject to
>http://docs.yahoo.com/info/terms/
> > >
> > >
> > >
>
MSN Photos is the easiest way to share and print your photos: Click Here
|