To All: I noticed this exchange over on Lee's site!
Forum: Market thoughts picks observations
Author: John G
Date: 01/28/2000
I used to love lurking the debates between Don and LG
on SI over whether the near term or the long term is more
easily predicted.
Clearly, Don was right! If I see a bike rider going 30 mph,
it is clear that he is more likely to be moving forward one
half second from now as he is to be moving forward one
hour from now. Same with the market.
TA is ultimately about tracking money flows that produce
price changes.
First you look for divergences - evidence that flows are
beginning to turn before price turns - examples are ad lines,
accumulation/distribution indicators, obv; as well as sentiment
indicators such as put-call relationships, the Rydex Ratio, and
sentiment surveys. These are most useful for the position trade
of a couple weeks to a couple months.
Second, you look for signifcant price movements to produce
a change in money flows, especially around extremes of
advancing over declining volume, new highs to new lows,
and advancing over declining issues. These are what Don
uses. A significant price movement will typically do
exactly what an economist would expect: It changes the
willingness of the players at the table to deploy
their resources, and produces a reaction or counter
movement. This method of TA is ideal for the swing
trade of a few days.
After lurking over on SI, I can say that Don does this with
more discipline and much less emotion than most.
My TA tells me that controling the emotions and sticking
with the discipline is going to be very difficult over the
next few weeks because of the increasing volatility.
Forum: Market thoughts picks observations
Title: None
Author: DonaldSew
Date: 01/29/2000
John,
>>>> I used to love lurking the debates between Don and LG
on SI over whether the near term or the long term is more
easily predicted. <<<<
I cant remember if I had mentioned it before. I have stated in the past
the highest level of probability that can be achieved in predicting long term
using TA is 62%, which is only slightly better than flipping a coin.
I have heard of mathematical studies arriving at that figure, but I also have a friend who
is a genetic scientist at Hopkins. He attempted to do a probability study on using genetics
to cure diseases. On his 1st attempt he obtained the probability of 90+%, which he knew
was impossible, since that high of a probability implies that most diseases could be cure
tomorrow. gggggggg So what did he do wrong? He noticed that he limited the number of
variables during his 1st study. So he then opened his studies to all variables and the ran it
several times. The highest probabilty he could then obtain was no higher than the 63-64%
range, which is very close to the 62%(FIBINOCCHI #(61.8)).
So based on that and other stuff, I strongly believe that the shorter the time frame the
higher the probability since there are less variables.
I also enjoyed the discussion with LG on this topic. I believe that he was referring to
predicting the intraday pivot points as being harder to determining than longer term
prediction. What he failed to realise was that he wasnt comparing apples to apples and
that he was using different criterium. If one is looking for the pivot point to the minute
and price, what is harder - to look for it during the current day or 1 month later, meaning
the exact pivot point to the minute at a specific price 30 days from now. Hope I made
some sense.
Seeya
-------------------------------------------------
To JohnG and Donald Sew:
First I enjoyed my discussions with Donald as well. But I was not referring to picking the exact pivot points any more than Don's system does. Don's system tries to signal a window of time in which a pivot point will occur, but gives no real hint at which level that reversal will occur. Actually after some of the initial Class signals the market continues strongly in the direction of the trend before finally putting in a true reversal. Actually I have seen Donald's system give repeated sell signals in strong up trends with out giving a buy. This would get you out and then not get you back in early on in a extended rally. Same thing visa versa. Now I know Donald uses other analysis to get him back in or out depending on the trend, I just wanted to point that out about his guitar system.
Actually I believe in using several time intervals to make reads, from weekly, daily to various intraday time intervals. All can be dependant on the other.
I think time intervals at all levels can be difficult. But I would never purport to say one is statistically more viable. However, supposed government studies seem to say that trading very short-term or short-term is less successful than investing longer-term. This statistically would seem to counter Don's premise. At least as far as wallet thickness is concerned...after all that is what counts the most...<g>
No offense John, but your motorcycle rider comparison is ridiculous.
Let's put this direction calling into perspective.
Donald Sew's short-term cycle is 9 to 12 days as I remember. Let's go with 12 as I remember his last public post stated. (I could be wrong or he may have changed it since, either way it is good enough to make my point.)
Class 2 Sell on Monday: Direction is to Sell at Tuesday's High (you have to determine the high yourself) or you may need to sell as soon as the signal sounded today. Then it may morph into a Class 1 Sell signal on Tuesday.
Class 2 Sell morphs into a Class 1 on Tuesday: Direction is to sell at Wednesday high (you have to determine the high yourself) or you may need to sell immediately today. Then it could be late suggesting you sell at Thursday high.
That is four out of twelve days to take action assuming the signal is not negated. You have to decide exactly when to take that action if you're going to catch the top or at least be near the top. Assuming your good enough to pick the top on your own. This system can take up to a full one third of the 12-day cycle or 33%.
Let me put that into perspective. If I told you I had a medium-term system which was based on a 12-month cycle. And I told you...
Class 2 Sell in January means to sell at February's high (you have to determine the high yourself) or you may need to sell as soon as the signal sounded this month. Then it may morph into a Class 1 signal in February meaning you should sell at the highs in March (you have to determine the high yourself) or you may need to sell as soon as the signal sounded this month. Then it may be late meaning you may have to sell at April's high (you have to determine the high yourself). Of course, on occasion my signals are negated.
That means that if I make an intermediate market call at the first of January, it means from now until the end of April sell at the high sell and I still get to claim accuracy, right? I think not...
This comparison is not meant to put down anyone's system, but to point out the fallacy in saying one time frame can be more accurately called than the other. It really depends on what your investment time frame is and what your tolerance is. If you are trading or investing Indexes, actually once you project your time frame out far enough you might as well just buy and hold, really!
There are ripples in all time frames. If one can accurately trade the smallest time interval ripples one can stand to make the most money. However, government statistics indicate that tick to tick traders (scalpers/day traders) overall fair the worst. That is the shortest of all trading time frames. While buy and hold investors fair far better. Why do you think that is if calling shorter-term market moves is statistically more viable?
Food for thought...
Frankly, I think anyone that can come up with a system that works for them and makes them money on a regular basis should be commended. So kudos to Donald if his systems makes him rich and kudos to all others with styles and or systems that work for them. To each his own...<g>
Regards,
LG |
