Investing
Process
Active management consists of selecting strategies or
screens for a group of stocks, selling the group or individual
positions when they appear vulnerable, and then using the cash
generated from the sale to reinvest and continually balance the
portfolio. The goal of active management is to achieve
significantly better returns than the market; my task is to make the
chart lines go up.
Every position is purchased within the context of a
specific strategy and set of criteria. Almost always stocks are
purchased in a set or to maintain a set of between seven and twelve
positions to execute a specific strategy and portfolio within a specific account.
For diversification, my client and I prefer at least three such
portfolios. Some households have up to a dozen
strategies with seven to twelve positions in each strategy.
Passive management at Wenzel Analytics is a strategy
to buy exchange traded funds and occasionally mutual funds for
smaller accounts or sometimes for a minor portion of larger
accounts. The goal of passive management is to achieve
diversification with minimal risk of underperforming the market.
For active or passive management, the process is to
buy, review, sell and buy again. Turnover is greater with active
management. Timing strategies are deployed cautiously at the point
of suspected major market turns or as an influence upon the exact
timing of a purchase or sale planned for other reasons.
This report is a fairly detailed snapshot of the
processes for buying and managing investments; the
processes will continue to change and evolve.
Buying
To start with the buying process, every position fits
within one of three different strategies. Some
strategies come from an outside tested source with data making it
credible and promising. Some strategies are derived from my
statistical and data mining work. And some strategies come from a
rationale or logic that seems to make enough sense to put money on
it.
Sources
Tested Source Strategies
Tested source strategies derive from a variety of
sources. Some have come from investment newsletters with strong
long-term returns according to Hulbert’s Financial Digest
which independently tracks about four hundred such newsletters. Sometimes I will take a mutual fund that
has exceptional and stable returns with low turnover. Mairs
and Power Growth Fund is an example that has only about thirty-five
positions. I pick and choose from the stocks it holds, paying
attention to the mutual fund's report each quarter as to what they
are buying and selling.
Data Mining Strategies
Data mining screens are the result of my ongoing
statistical work. Some of these start strictly from weekly or
monthly database, and some are refinements of AAII screens.
As introduction, let me give a brief description of
data mining. In traditional scientific and academic research, the
researcher begins with the hypothesis and then assembles data to
prove that the hypothesis cannot be false and therefore must be
true. In stock selection, the typical format is to take a logical
selection of criteria such as certain values for price/earnings
(PE), growth (PEG), debt, revenue, etc. and then back-test the
screen to see if it worked in the past. With data mining software
one searches through thousands of variable combinations and their
respective values looking for statistical significance. The
researcher does not ask the specific hypothesis before scanning for
relationships. After strong relationships are discovered, the screen of
variable combinations may or may not make sense. Those with some
kind of logical explanation are more likely to persist than those
which are accidental or “trained”, but not necessarily so. I typically
work with about eighty variables taken from ten or more years of
monthly data comprised of about four thousand stocks each month,
yielding a database of if 500,000 records or more. Sometimes the
relationships are linear, meaning the higher returns correspond with
variables going up or down. Often the highest returns are
at both extremes of a variable, and sometimes are true only for a
specific midrange. Typically four to six variables are found to
best define a productive screen. For data mining software, I have been using KnowledgeSEEKER for over
twenty years.
The Shadow strategy started with the variables and
respective values for a screen of that name which is frequently
published in the AAII Journal by James Cloonan, chairman of
AAII. Using the data mining process, I refined the screen, dropping
some variables, changing the screening values of others, and adding
still others according to my findings. Similarly, the Piotroski
Relative Strength strategy uses a Piotroski scale published by AAII
for the Piotroski screen, but uses different values on the scale and
combines the scale with variables other than called for by the AAII
Piotroski screen.
The latest statistically derived screen is called
Playing Defense. It
was created from data in down-trending markets, and thus has better
resilience in down markets. The number of stocks meeting the
criteria in any given month is incredibly predictive of future
returns one year later.
Strong Rationale Strategies
The final source category is those strategies with a
strong rationale or logic, but not necessarily an easily verifiable
history. The High Income strategy has had extraordinary returns for
several years. The Resource Scarcity strategy consists of stocks
selected on the assumption that energy, metals, timber, water and
other resources are going to be in increasing short supply and their
price will go up. Some of the stocks are major holders of such
resources, and some stand to benefit by providing specialty services
in response to increased demand. There are occasions when I will
break the rules and buy a single position on a tip or other reason.
Usually these are in my personal account unless a client requests
buying a specific position.
Our overall performance for each of these strategies,
both recently and since inception, can be found under the Stock
Performance heading in the navigation panel.
Strategy Selection Review
If the potential strategy comes from a trusted source
or from data mining, rather than merely a strong rationale, the next
step is to subject the strategy or screen to a rigorous analytical
and quantitative review. The first test is to see if the criteria
consistently produce an adequate number of stocks. Screening criteria that produce five stocks one month and a hundred
and five another month are hard to implement. The next review is consistently
high rates of return. Screens are dropped if the returns vary
considerably, show a pattern of exaggerating market down trends, or
if the average high returns are because of a small number of stocks
within the screen rather than dispersed throughout all selected
stocks. Since a higher return will necessarily create a higher
standard deviation, screens are compared using a coefficient of
variation obtained by dividing the standard deviation by the average
return. Returns rates going forward from the date of the data are
compared for four weeks, thirteen weeks, twenty six weeks and fifty
two weeks, plus using the average of these time periods. Returns
are often compared going back a longer time period, such as four years,
then for the last year and sometimes for the last quarter using
weekly data. A major change in market direction will often change
the performance dynamics of how well a screen works, prompting new
data mining research and new average return comparisons for
selecting screen strategies. Sometimes a pattern will work for
forty-five years, and suddenly not work. It may resume working at
some point, or it may not.
Please ask if in the process of selecting a strategy
for your account you are interested in the coefficient of variation
or other comparison numbers.
Technical Analysis to Select Individual
Positions
Once a screen has been selected using historical data
and the steps described, the filters for the strategy are run on a
current stock database. The thirty or so candidates are imported
into Worden’s TeleChart software and each position is given a
technical rating. Some are a "buy now" and some are
"interesting, but not right now." Some stocks are excluded
on principle, such as tobacco and gaming stocks.
The technical analysis is a study of the balance of
supply and demand for a security, charting price and volume for
various time periods. Like in real estate, a stock is worth today
what a buyer will pay for it, and worth tomorrow what a buyer will
pay for it. As explained and illustrated below in the section on
selling, I mostly use trend lines and horizontal shelf lines of
previous resistance or support, looking at daily charts and then
multiple time frames usually going back ten or fifteen years. I
also review the preponderance of evidence from several other
indicators. I tighten or loosen my judgment until I get a list of
ten or so stocks from diverse industries.
Occasionally I do a blind test on technical analysis
predictability by taking from fifty to a hundred stocks and, not
showing the later history, make my selections. I then unfold the
history and evaluate my choices. Sometimes I’m able to do much
better than a random selection, and sometimes the results are not
significantly different from a random selection. It depends on the
market and on the screen. Some screens will get high returns from
stocks that technically look very scary. At any rate, the process
does not do worse than a random selection, and provides a way to
winnow a screen down from thirty to ten.
The whole selection process is a way of implementing
the laws of large numbers such as used by an actuary. An actuary
can predict fairly precisely what the death rate will be for 100,000
people sharing common characteristics, but cannot predict who will
die next year. To benefit from the predictability offered by the
laws of large numbers, implementing a strategy needs to include
enough stocks to represent the whole but few enough to minimize
trading costs and management complexities.
Trading Desk
Once I have ten or so stocks current for a given
strategy, I paste the list into my list of buys for all current
strategies. When to actually buy stocks depends upon what the
market is doing. If the market is heading down, I will wait
until it turns up. (All purchases are long; I do not buy
short. Most accounts are in IRAs which do not permit shorting.
A minimum amount of margin may be used temporarily in non-IRA accounts.)
When I’m ready to fill a strategy for one or more accounts, I review
intraday charts for each position and determine a likely price and
whether based on trading volume it will be a market or a limit
order. I then paste the buying list for the screen into the
Excel database that is used as a trading desk tool as well as for
various filtering and for creating a pivot table. Once in the
database, I paste from a previous row to the current entries about
five fields identifying the account, and then divide the amount of
money available in this account for investment in this strategy
between the different positions such that there are roughly equal
dollars in each position, rounding out lots when possible. I
then upload orders to Scottrade. After each
transaction executes, price and cost are entered back into the Excel
database. The next morning when the brokerage has updated
their records, the transaction data are retrieved into Fund Manager
(www.FundManagerSoftware.com),
the portfolio management program used for performance calculations
and performance charts.
Review and Management
While a chart for each position in each account is
maintained on Fund Manager, usually my regular reviews are done on TeleChart which has watchlists of all unique positions in each
strategy. Stocks that are precarious are kept on a separate list
and reviewed more often. The TeleChart
watchlists are imported from pivot table unduplicated lists of
positions in each strategy.
Almost all selling is based on a technical chart
review. Stocks are harvested when they become parabolic, meaning
that the prices on a logarithmic chart are going up steeper and
steeper, and volume or other indicators are pulling back. I will
get out if there is a frenzy and most of the float is trading in a
short period of time. Stocks are also harvested when the chart
lines are rolling over and breaking through an upward trendline.
Sometimes I will replace stocks after a prescribed period such as a
year if the supporting research was based on that timeframe.
The best time to buy a stock is when it is priced
higher than it was, and the best time to sell a stock is when it is
priced lower than it was. To buy at the absolute low or sell at the
absolute high is unrealistic and usually means one is buying as it
continues to fall or selling as it continues to rise. Instead of stop orders or goal prices, I prefer to
place expectation trend lines on each chart.
I do not believe in selling when a stock goes down a
given percent from the purchase price. Contrary to popular belief,
what we paid for a stock has nothing to do with whether a stock will
go up or down. The reasons to sell a stock are if it threatens to
go down, regardless of the purchase price, or does not continue up
at a satisfactory rate. Maintaining account value and avoiding
losses are important at the account level, but not a major
consideration for each position.
Occasionally I or a client will want to move from one
strategy to another or into cash. Even if I decide to abandon a
portfolio, the selling process is usually over time and is based on a
more aggressive stance towards selling individual positions and not
replenishing with new purchases. Rarely will a position be sold
merely because it is no longer a buy on the source screen. If the
market as a whole drops below key support lines, I will become more
aggressive in selling positions and accumulating cash. Any passive
investments (exchange traded funds) are the first to be sold in a
dropping market.
As cash accumulates from the sale of positions, and
if the market appears strong, I then buy more positions. This
requires a pivot table review of the strategy allocation for each
household, which may apply to multiple accounts if each spouse has
an IRA, a Roth IRA, and they jointly have an after-tax account. If
a household has thirty or forty thousand dollars in cash, I consider
adding a new portfolio and may call to review it with the client
depending upon our understanding. Or I may replenish an existing
portfolio using the most promising stocks from the current buy list.
So,
Continuous Improvement
The most tedious parts of the investment process are
in transferring and cleaning the research data, and in keeping
records on Excel and Fund Manager reconciled with the official
brokerage records.
The transferring and cleaning process involves
downloading the research data, setting up return calculations for
different time periods, and accommodating stocks that had splits,
dividends, were acquired, had symbol changes or went out of
business. Stocks are deleted if priced less than one dollar and
trading fewer than five thousand shares a day over the last ten
days. Outliers with exceptionally high or low returns are sometimes
either deleted or the returns limited to a maximum or minimum amount
so as to not overly distort the findings. The
availability of cheap external storage has enabled storing Stock
Investor Pro data for each month just as it was originally
published. This has made historical data more accessible.
Scottrade and the Fund Manager portfolio
management software continue to improve their interfaces and the
capabilities for retrieving data and easily reconciling records.
I continue to work with them to assist them in design that
facilitates my business processes. The more investors and
businesses there are that have the same needs as Wenzel Analytics,
and the more assertive and transparent we are as to what we need,
the more accommodating will be the software.
Conclusion
So there you have the investment process in more
detail than any sane person would probably want to know, unless you
are interested in starting a similar business, or would want to use
the process for your private investing. In either case, I would be
happy to answer questions and share my experience. When I take my
car to a mechanic, I don’t want to know the details of what he or
she knows, I just want to know that he or she can fix it. The
detail is here in case you think your investments are worth more
than my car — which I hope they are — and deserve a thorough review
of how they will be selected and managed. |