|
PureBytes Links
Trading Reference Links
|
In this case initial equity but it could be anything. To clarify the
mechanics of this particular simulation for anyone interested, it
started with a basket of trades from AB output (% profit/trade). It
randomly drew trades from the basket w/replacement and made a sequence
of 100 (100 was arbitrary) trades, an artificial equity line if you
will. Then it calculated max % dd *for that sequence* by taking the
largest percentage distance between peak equity value and following
trough values. It did this 999 more times then made the histogram and
cumulative distribution graph based on the 1000 max % dd results.
What you're in effect doing is coaxing all you can out of your data.
--- In amibroker@xxxxxxxxxxxxxxx, "Fred" <fctonetti@xxxx> wrote:
> So the estimated MaxDD is based on what ? an indivdiual trade ?
> initial equity ? the account balance at any given point in time ?
>
> --- In amibroker@xxxxxxxxxxxxxxx, "quanttrader714"
> <quanttrader714@xxxx> wrote:
> > It's late and I've had too much scotch, so one very quick example
> > which I'll explain the basics of but would like to have someone
else
> > please take a stab at interpreting.
> >
> > To recap the Robustness Criteria, Condensed Version 1-5:
> >
> > 1. Test on small, mid & large cap stocks in bull, bear & sideways
> > markets.
> > 2. Evaluate performance on top 20% most actively traded small,
mid &
> > large cap stocks.
> > 3. Graph and evaluate system performance consistency
(%profit/trade
> > and % profit/bar) on select stocks.
> > 4. Perform simulation to estimate probability of profit in 10
trades
> > (for select stocks).
> > 5. Perform simulation to estimate future drawdown (for select
> stocks).
> >
> > For this example I picked a stock, any stock. I think everyone
gets
> > what I mean by criteria 1 and 2 (whether they agree or not),
correct
> > me if I'm wrong. I've posted the output of criteria 3-5 in the
> > example folder in the photos section. Criterion 3 output is
photos
> 1
> > and 2, criterion 4 output is photos 3 and 4, and criterion 5
output
> is
> > photos 5 and 6. I think the criterion 3 graphs are self
> explanatory.
> > On criterion 4, forget how it's calculated for now. It
estimates
> the
> > probability of profit (and how much) at the end of 10 trades.
Unit
> of
> > measure is % of starting equity. Looking at the histogram, the
> > highest bin (the mode of the distribution) is 19.16 -- 29.63 which
> > means approx 15.5% of the time (y axis) the profit at the end of
10
> > trades fell in this bin, between 19.16% and 29.63% of initial
> equity.
> > The cumulative distribution graph is the histogram in cumulative
> > form and shows the likelihood that a result falls below the value
on
> > the x axis. For example, 20% of the simulations (of the sum of 10
> > trades) lost money so you can *estimate* there's an 80% chance
> you'll
> > be profitable after 10 trades with this. Same unit of measure
for
> max
> > dd and those graphs are read the same way. P.S. Each simulation
was
> > 1000 runs, so the graphs of criterion 3 show one actual pass
through
> > the data by AB, while the others depict the collective results of
> 1000
> > simulated runs (and include my adjustment factor).
Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
(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/
|