Feature
Understanding Market Type
by
Van
K. Tharp
This is the first of a
series of articles on
market type. For example, in the update for this month I
said that the market type for the S&P 500 index was SIDEWAYS
VOLATILE. That prompted someone to ask, “How can you say that we are in
a sideways market when the market has gone up so much in the last two
months?” And that’s all the more reason for a discussion of
market type. Market type depends upon how you define it.
Right now I’m actually looking at three different measures of
market type and they couldn’t be showing more different
results. Here are the results for 2009 as of May 15th.

View
Larger Charts
This is the perfect time
to do such an article because all three models are showing a different
market type and Methods 2 and 4 are as far apart as two models could be.
So let’s look at what each model does and why.
Understanding Method 2
Method 2 is my original market type that I have been publishing in our monthly updates in
Tharp’s Thoughts. It started with The Definitive Guide to Position
Sizing. In that book, I said that market type should be
determined quarterly. And that’s an important point: My time frame is
three months. That’s why I can say that the market is sideways over
three months even when it has been going up over the
past two months—especially if the first month down is much more than the last two months of up
movement.
Anyway, my idea was that
if you put a quarterly chart on the market, a bull and bear market
should be obvious. And if you couldn’t tell, then the market was
sideways. And in my book, that was my first way of doing
classifications. However, I wanted something I could do weekly and
mechanically and that required a revision. I couldn’t just look at a
chart and have it be mechanical. In addition, I had to do 13 week
rolling windows.
I then came up with the
following idea. What was the average 13 week change over many
13 week windows? That turned out to be useless because the
up and down periods cancelled themselves out. The actual change was
1.7% over a 58 year period from 1950 through April 2009. Thus, I went
with the absolute value of the 13 week change. This turned out to be
5.58% and I made the arbitrary decision that any change, positive or
negative, that was less than 5.58% was a sideways market. This means
that a positive change greater than 5.58% was a bull market. And a
negative change less than 5.58% was a bear market. And I could now do
this with 13 week rolling windows.
Next, I had to decide
how to distinguish volatile and quiet markets. My original thinking was
to look at the weekly changes as a measure of volatility. What was the
average weekly change? Were at least 7 weeks above average? If so, we
had a volatile market. If not, we had a quiet market. But the problem
with that measurement was that the weekly change could be very small
while the price range during the week could be huge. Thus, this idea
didn’t work some of the time.
To overcome this
limitation, I decided to look at the Average True Range—a much more
traditional measure of volatility instead. I’d look at the mean weekly
ATR over 13 weeks over many time periods. My first thought was that if
the ATR was above average, then we had a volatile market and if the ATR
was below average, then we had a quiet market. This was a great idea,
but it turned out that for the S&P 500 most markets before 1997 were
quiet and most markets from 1997 on were volatile. This was because
the price of the index had a major impact on the volatility. Thus, I
had to take the 13 week ATR as a percentage of the price of the S&P
500 index. Over 58 years this mean turned out to be 2.95 with a
standard deviation of 1.45. Thus a volatile market was defined as one
in which the ATR% over 13 weeks was greater than 2.95. A quiet market
was defined as one in which the ATR% over 13 weeks was less than 2.95.
The chances of getting one at the average was very slim because we went
out many decimal places.
So the following is a
distribution of 13 week periods falling into each market type with our
new definitions, defined as Method 2. The last 3,000 periods are
included in the table. Notice that 58.87% of the markets are defined as
sideways. Conventional wisdom says that markets trend about 30% to 40%
of the time, so my definition is pretty good. 12.37% of the markets are
bear and 28.77% are bull. Thus, there are about twice as many bull
than bear markets.
In addition, the table
shows that 39.87% of the markets are volatile, while 60.13% of the
markets are quiet. Again, this seems to fit conventional wisdom.

View
Larger Charts
Understanding Method 3
Method 3 probably
wouldn’t exist except for the fact that
Method 2 was developed and tested under market conditions in
which I could get weekly changes in the S&P 500 of 10% or more. My
biggest concern, since we were using rolling windows, was that a market
type could change, not because of what happened in the most recent
week, but because of what was dropped that occurred 14 weeks ago. For
example, if the market dropped 14% 14 weeks ago and that change was
dropped, then the market type could change from bear to bull even if the
latest week’s price change was down 1%—just because the 14% drop was no
longer included.
Method 3 made one minor
change to the method by substituting a 13 week exponential average for
the last week. In other words, the market type was still determined by
absolute change over 13 weeks, but the last week was an exponential
moving average rather than the exact change that happened 13 weeks ago.
Notice that Method 2
moved to volatile
bull on the week beginning May 4th. But Method 3 has stayed
at volatile sideways. Method 3 also produces smoother results. For
example, Method 2 had four weeks of volatile sideways in the first two
months of 2009 whereas Method 3 remained in a volatile bear mode. Let’s
see how our distributions change as a result of what happens in the
first week of our 13 week window. These are shown in the next table.

View
Larger Charts
Notice that while our
results are smoother, the distribution of market types is pretty
similar. We are 60% sideways and 60% quiet. Bull markets are still
twice as prevalent as bear markets.
Understanding Method 4
Now let’s look at a
totally different method that comes from Ken Long. Ken started working
with a traditional definition of bull and bear. A bull market is when
prices are above their 200 day moving average and a bear market is when
prices are below their 200 day moving average. Ken, however, also
wanted a sideways measure. As a result, he developed a 2% band around
the 200 day moving average. When prices are within that band, the
market
is sideways. When prices are more than 2% above the 200 day
moving average, the market is a bull. And, lastly, when prices are
more than 2% below the 200 day moving average, the market is a bear.
Notice that because
Method 4 uses a much longer time frame (i.e., 200 days) it still
classifies the market as bearish. Even with a two month rally, the
market is still more than 2% below the 200 day moving average. And
thus, Method 4 still says we are in a bear market. Notice that with
this definition, the 200 day moving average could be moving at a sharp
angle (up or down) and if price were staying close to it, the market
would be considered sideways.
So how does Ken measure
volatility? Ken again uses the ATR as a percentage of the price. He
looks at the 14 day ATR% windows over 100 days. He finds the mean ATR%
over 100 14 day windows and the standard deviation. When the ATR% is
within one standard deviation from the mean, the market is considered to
be normal. When the ATR% is more than one standard deviation above the
mean, he calls the market volatile. And lastly, when the ATR% is more
than one standard deviation below the mean, he calls the market quiet.
We used Ken’s model on
weekly changes to the S&P 500 (Ken looks at daily changes in the
SPY). We used the 40 week moving average instead of the 200 day—almost
no difference here. However, we had a major difference in how we
calculated the volatility. We used a 13 week ATR, computing the mean
and the standard deviation as rolling windows. If the price ATR% for
this week was less than one standard deviation below the mean of the
last 13 weeks, the market was considered to be quiet. If the price ATR%
for this was greater than one standard deviation above the mean of the
last 13 weeks, it was considered to be volatile. And if neither of
those statements were true, then the market was considered to be quiet.
We are actually taking
some strong liberties with Ken’s method by using the weekly data.
First, 13 weeks is only 65 days. Ken uses 100 days. He also uses 14
day ATR%, finding the mean and standard deviation over 100 days. We are
using 13 week ATR percentages finding both the mean and the standard
deviation. There is a huge potential difference here.
Thus, Ken has nine
market types. The next table shows the distribution of market types
over the last 58 years using Ken Long’s market types.

View
Larger Charts
I’m not at all happy
with these distributions, so we’ll need to recalculate those using daily
data. With over 3,000 13 week windows, we are getting 38%
normal markets, 29% volatile, and 33% quiet markets.
Even more amazing (and this should be accurate according to Ken’s
method) is that 58% of the markets are considered to be bullish, 25% are
bearish and only 18% are sideways. This is probably because prices are
only within the 2% 200 day moving average only about 18% of the time.
This does not fit the conventional wisdom that markets are sideways
60-70% of the time.
Conclusions
First, the time frame of
the market type makes a huge difference in your conclusion. Thus, a 13
week method can show the market as bullish whereas a method based upon
the 200 day moving average can show the market as
bearish.
Second, market type is
very individualized based
on how you trade. Day traders and swing traders will have
an entirely different view of market type than longer term traders or
investors.
Third, minor assumptions
in how you calculate market type can make a huge difference in the
conclusions you make. I plan to start using Method 3.
In my next article on
market type, I’ll include daily calculations so that we can accurately
represent market type and we’ll look at the daily, and weekly predictive
value of the various market types.
About
Van Tharp: Trading coach, and author, Dr. Van K. Tharp is
widely recognized for his best-selling books and his outstanding
Peak Performance Home Study program - a highly regarded classic
that is suitable for all levels of traders and investors. You can
learn more about Van Tharp at www.iitm.com.
|