Preston's question is very relevant. Tested in what regard?
I've tested Williams and a lot of other guru systems over a variety of
time frames and stocks.
1. On the long side, no TA system or guru system works very well
unless the market is in an uptrend. When it is in an uptrend, a monkey
with darts and the wall street journal (random walk of sorts) does
just as well. However, TA lets you look at charts so you think you're
doing better.
2. Here's a simple way to review your system. Create a simple index
that tells you how many stocks are higher in price in the last 3 days,
5 days, 10 days and 20 days. Measure those percentages on the overall
market and compare them to the stocks picks from the guru system of
your choice. You can see if the guru system is beating the market in
terms of better picks than the market overall. Don't just look at
return, look at individual trade performance.
3. Consistency is the real issue. New traders, want-to-be's and other
lookie lou's want a system that generates 100% plus returns a year. It
ain't possible. If it were being done more than once in somebody's
lifetime there would be a lot of really rich traders. Their ain't.
Consistency wins. Measure it, study it, try to achieve it.
4. TA works the best when it is applied to another prescreening
methodology that has a long consistent track record for picking
winning stocks that beat the market. For example, if Valueline has a
method that results in a group of 100 stocks outperforming the market
based on fundamental and quantitative analysis, apply a TA system to
that group. Learn what kind of market conditions are necessary for the
Valueline (or other FA or QA picking method with a long successful,
consistent track record) to perform well and put a constraint on the
TA system so it only picks trades from the group of 100 stocks when
those market conditions exist. Now you'll have a chance to make money
consistently.
5. Yes, I've tested all of this. Some of the results I wrote up for
Roy's newsletter, the vast majority of my test results I haven't ever
published any where. Interesting stuff. You would be surprised at how
poorly TA perform on the general market. And that includes every kind
of TA you can think of. Basically, it's the market conditions that
dictate performance, not the indicators, etc. Yeah, I know money
management is important and all the other stuff in the guru books.
I've read almost all of them, and I've tested a lot of money
management systems also.
6. Specifically I've tested the Williams stuff. It's okay like every
thing else. His offset Wilder's median price MA's that he names
alligators (cute ) are okay. I've tested just those against all of the
other MA's that I could find. Every thing works well in an uptrend (on
the long side) and none of them perform very well in other market
conditions.
7. If you find a long side system that works in uptrends, simply
reversing it and using it during downtrends does not necessarily
produce good results. It has to be tested in downtrend conditions.
Trading the short side in a downtrend is very different than trading
the long side in an up trend.
Most guru systems have so many if, ands and buts, it makes testing
them hard, if not impossible. The guru's like it that way. It helps
them sell their products and books. Testing is not good for business
unless the results come from very limited duration tests over
carefully selected market periods when their whatever system was
performing well. Even a blind squirrel can find an acorn sometimes.
Things like Elliot waves, Murray math and other chart voodoo reading
methods are impossible to test. However, I did run across one Elliot
wave stock picking service that's been around for several years. I got
a look at it's real track record and it sucked. This is a professional
Elliot Wave reader! Ouch that hurt--well only the Elliot wave users.
(Is the current market in it's 3rd wave, leg or is in a flapping arm
wave? I can't tell!)
Even though I'm not going to share my Williams or other guru test
results with the group, all the other stuff I gave you is FREE! You'll
learn it all over the years anyway, if you last long enough.
Super
PS Sharing test results is very complex. It takes many, many pages to
define and explain the constraints that went into a test, why the test
was done the way it was, how bias was handled and another dozen issues
that are involved in every test that's worth anything. Testing ain't
cheap and easy, and remember, while it may look safe, it isn't
something you should attempt at home due to the risk of injury to your
wallet and your brain.
--- In
equismetastock@yahoogroups.com, pumrysh <no_reply@xx
.> wrote:
>
> Ticiano,
>
> Helps if we know which Williams System you are looking at.
>
>
> Preston
>
>
>
> --- In equismetastock@yahoogroups.com, "Ticiano Rêgo"
> <ticianorego@> wrote:
> >
> > I´m trying to develop a System Tester to verify the Bill Williams
> System.
> >
> > There are 12 breakout conditions. ¨6 on the long side and 6 on the
> short side.
> >
> >
> > Long:
> >
> > up1=(H>Ref(H,-1) AND H>Ref(H,-2) AND H>Ref(H,1) AND H>Ref(H,2));
> >
> > up2=(H>Ref(H,-1) AND H>Ref(H,-2) AND H=Ref(H,1) AND H>Ref(H,2) AND
> H>Ref(H,3));
> >
> > up3=(H>Ref(H,-1) AND H>Ref(H,-2) AND H>Ref(H,1) AND H=Ref(H,2) AND
> H>Ref(H,3));
> >
> > up4=(H>Ref(H,-1) AND H>Ref(H,-2) AND H=Ref(H,1) AND H=Ref(H,2) AND
> > H>Ref(H,3) AND H>Ref(H,4));
> >
> > up5=(H>Ref(H,-1) AND H>Ref(H,-2) AND H=Ref(H,1) AND H>Ref(H,2) AND
> > H=Ref(H,3) AND H>Ref(H,4));
> >
> > up6=(H>Ref(H,-1) AND H>Ref(H,-2) AND H>Ref(H,1) AND H=Ref(H,2) AND
> > H=Ref(H,3) AND H>Ref(H,4));
> >
> >
> > Short:
> >
> > dn1=(L<Ref(L,-1) AND L<Ref(L,-2) AND L<Ref(L,1) AND L<Ref(L,2));
> >
> > dn2=(L<Ref(L,-1) AND L<Ref(L,-2) AND L=Ref(L,1) AND L<Ref(L,2) AND
> L<Ref(L,3));
> >
> > dn3=(L<Ref(L,-1) AND L<Ref(L,-2) AND L<Ref(L,1) AND L=Ref(L,2) AND
> L<Ref(L,3));
> >
> > dn4=(L<Ref(L,-1) AND L<Ref(L,-2) AND L=Ref(L,1) AND L=Ref(L,2) AND
> > L<Ref(L,3) AND L<Ref(L,4));
> >
> > dn5=(L<Ref(L,-1) AND L<Ref(L,-2) AND L=Ref(L,1) AND L<Ref(L,2) AND
> > L=Ref(L,3) AND L<Ref(L,4));
> >
> > dn6=(L<Ref(L,-1) AND L<Ref(L,-2) AND L<Ref(L,1) AND L=Ref(L,2) AND
> > L=Ref(L,3) AND L<Ref(L,4));
> >
> >
> >
> >
> > Does anyone already have done this ?
> >
> >
> > Thanks,
> >
> > Ticiano
> >
>