Questions and Answers about
Design and Implementation Objectives
Describe your
investment philosophy in detail:
Our philosophy is that market conditions change on short timescales, and
the best opportunity to reduce risk and retain profits is to react as
quickly as the investment format allows. With mutual funds and variable
annuities, this means applying our technical and statistical analysis
software tools to execute trades at the close of each business day using
the most current market data available. With this approach, technical
and statistical analysis allows a more rapid reaction to changing market
conditions than fundamental analysis. All asset classes fall out of
favor at times, damaging a long-term portfolio’s chance of success. Our
priorities are first to limit downside volatility, and second to select
those investments for which a strong demand can be identified. For risk
management, we rely on diversification of both management system and
asset class. Our programs incorporate broad market stocks, bonds, real
estate, precious metals, industry sector equities, and international
stocks, so that at any one time, a portfolio may be invested in 10-15
different funds. Each asset class and equity sector is traded according
to one or more independently customized management systems. Our primary
objectives are to monitor an investor’s portfolio for downside
volatility to guard against catastrophic losses, to maintain
profitability, and to produce returns that exceed the broad market
indexes over time. We employ tactical asset allocation strategies that
include sector rotation, market timing, and the dynamic movement of
monies from one asset class to another, as opposed to always going to
money market funds when we exit a position.
Which mutual funds
are eligible for inclusion in the program series?
Our programs are
designed to be inclusive of all mutual funds available publicly in
tradable formats, whether at Rydex, ProFunds, or Potomac. Either
directly at the fund family or in variable annuity formats, we focus on
the broad market indexes, selected sectors, the U.S. bond market, and
international funds, as well as the inverse S&P 500 and inverse U.S.
bond funds.
How much money do
you have under management?
Currently, we manage
about $35 million in these programs, including all five of our risk
levels.
On a scale of 1 to 5
(1 being conservative and 5 being aggressive), where would you rank
these programs in terms of aggressiveness?
This program has
been developed with five risk levels, which we call aggressive (5),
capital appreciation (4), moderate (3), income & growth (2), and
conservative (1).
How are your
programs constructed?
Each of our programs
includes from 8 to 15 systems designed to allow dynamic investment of
money among funds in a variety of asset classes. Each system is composed
of multiple models, and each model may be conditioned with one or more
algorithms for buy or sell signals.
Would you describe
the models in these programs as systems models, discretionary models, or
combination?
The models are a
combination, although the frequency with which we override the
individual systems in the program (that is, use discretion) is low. We
have automated management systems that determine trades into, out of,
and between each asset class and sector, that ordinarily operate
completely autonomously. Our software generates daily reports that
allow us to monitor the daily drawdown (decrease in value from the
previous peak) of each system independently, and when the daily drawdown
approaches its historical maximum, we determine whether or not to
override that system. A review of the system and the investment
environment is conducted for alternative models, and the system may be
modified, or reinstated if conditions improve. The other systems in the
program continue running, unaffected by the override of a poorly
performing system.
Would you describe
these programs as technical, fundamental, monetary, or a combination?
Our programs rely heavily on technical analysis. We do not use
fundamental analysis to determine daily trades. Our technical analysis
does incorporate yield curves of interest rates and foreign currency
exchange rates, but no other monetary analysis.
What types of
indicators do you use to make buy and sell decisions, and which of these
indicators have the most influence over those decisions?
Pattern recognition, using probability systems and seasonality, accounts
for about half of our systems. The other half falls into the technical
analysis category, using countertrend tools to look at
overbought/oversold readings for some contrarian strategies. We also
use trend-following systems, such as moving averages and momentum
analysis, as part of our system design. This multiple approach is
applied, in combination, to each asset class in our program.
Approximately 1% of our trades are discretionary, following the
discontinuation of a system based on its daily drawdown.
How do you
differentiate between high and low confidence in an indicator, and how
does this influence the size of the allocation within a given fund?
The graduations of indicator confidence are implemented through our
diversification strategy. Our program is established with a number of
systems, operating on different funds using different indicators, with
fixed allocations. For a given fund, such as a small-cap fund or a
government bond fund, there are multiple systems in our program that may
invest in that fund when the indicators in those systems favor that
fund. This means that at a given time, if the indicators in several
systems all favor one fund (such as a government bond fund), that fund
will have a large overall allocation.
That fund’s
allocation will then decrease as those indicators independently begin to
favor other funds. From a global perspective, our allocations are
defined statically at the level of individual management systems (of
which there may be 10-15 in each program), and coincidences of those
systems selecting any one fund provide a variable allocation to that
fund.
Do your
indicators and influence of the indicators within each overall program
ever change? If yes, what would trigger the change? Our selections
of indicators and their allocations change for one of two reasons. We
are constantly testing alternative systems to search for better
performance, which we generally evaluate on non-client tracking accounts
before replacing our existing indicators. These changes may be made on
the timescale of a few months or longer. The other change comes about
through our discretionary overrides of systems that are performing
poorly, where after a review, we may attempt to modify those systems or
replace them entirely. This generally arises only every few months.
What types of
market environments are favorable for your programs, and what types of
market environments are unfavorable for your model? Our programs
are designed with multiple models to accommodate all market
environments. Many of our systems include both long and short funds
(short S & P 500 and short government bond funds) to take advantage of
any identifiable trends. The most difficult market for equities is a
shallow trading-range market, in which our systems generally prefer
other asset classes or individual sectors, like real estate or precious
metals.
How often are you
trading or rebalancing on an annual basis? Our dynamic asset
allocation programs are very active, with a trade occurring perhaps 4
out of every 5 days in one or more systems.
Do you have
capacity limitations for these programs? What are they? At this
point we don’t anticipate any capacity limitations for our programs.
However, sector funds (used in many of our systems) may be invested in a
limited number of individual stocks. Specifically, we’ve had verbal
assurance from both Rydex and ProFunds that we will be able to trade
large amounts of precious metals (greater than $10 million), and that we
will not affect the market sector on which the fund bases its investment
objective. It’s possible that capacity limitations could occur in the
future, with any sector fund, given a large enough investment.
Do you have a
targeted rate of return for these programs on an annual basis? A
targeted rate of return is not one of our objectives. Our performance
for the two years that the capital appreciation program has been in
existence has outperformed the benchmark used currently. With recent
changes, we are targeting an improved return with lower drawdown over a
longer measuring period such as 5 years.
What is your
performance benchmark? We have been using the S & P 500 as a
benchmark for these programs, primarily because it’s a widely accepted
benchmark for a capital appreciation equity-type exposure, and it’s the
most widely understood. Hybrid “buy-and-hold” benchmarks we have
developed, incorporating all of the asset classes and sectors we use,
are complicated by the fact that our instantaneous exposure to asset
classes varies from day to day. (That is, our systems do not simply time
individual funds, alternating between one fund and a money market
fund.) Hybrid benchmarks we have constructed underperform the S & P 500
because precious metals and other asset classes have not done as well
over the last 20 years. Our overall objective is to outperform the S &
P 500 on a total return basis and do it with less risk, as measured by
daily drawdown.
How has your
portfolio performed relative to its benchmark? Can you provide some
insight regarding any over- or underperformance? The capital
appreciation program started with the primary systems currently in place
December 31, 2002. Over that timeframe through December 31, 2004, this
program has outperformed the S & P 500 on an absolute and a
risk-adjusted basis. The S & P 500 Index over the same period had an
annualized return of 17.34%, maximum drawdown (MDD) of 14.05%, for a
drawdown index (DDI) of 1.234. The drawdown index is a ratio of the risk
taken to get the return. It is computed by dividing the annual return by
the maximum daily drawdown.
How long have you
been utilizing this strategy or similar strategy employed by this
model? Our program comprises a number of systems, with the oldest
dating back 10 years and the newest less than a year.
How do you avoid
the problem of over-fitting in your model design? One of our
primary objectives in designing a model is to avoid curve fitting, which
is an easy trap to fall into because it’s clear what would have worked
well historically. Our approach is to use standard technical analysis
parameters on our indicators. Where we are using moving averages, we
use widely accepted numbers like 200 days. Indicators showing funds are
overbought or oversold are widely recognized. Where we have designed
mathematical formulas to identify short-term high probability
opportunities, our approach is to use a large number of data points in
order to provide validity for our system design. Sector rotation and
other relative strength-type indicators are all standard in our systems,
based on the same parameters for each system.
Will you ever
override your models, and if so, why? Yes, we may override an
individual system when its daily drawdown approaches that system’s
historical maximum daily drawdown. This alerts us to the need to
monitor that system for recovery, modify that system, or replace it.
Do you analyze
past trades to see why they worked or didn’t work in this model? What
did you find? We continuously monitor the individual trading
performance of each of our systems. While we try to analyze why a
series of trades didn’t work, we recognize that even the best models
provide only a slight statistical improvement when examined over a small
number of trades. Our intention is to develop systems that, when
examined over many years of historical data, perform well statistically,
and question them only when they deteriorate relative to their
historical performance.
Do your programs
evolve? How? By the nature of our dynamic system architecture,
where different asset classes and sectors are favored under different
market conditions, the systems in each program keep pace with the
market. We constantly research new systems and new philosophies that
may offer improvement, and once we have tested them and can demonstrate
that those new systems do perform well, they replace underperforming
systems.
What software
application do you use to test your models and do research? We have
proprietary software that has been custom designed for research and
testing, allowing the flexibility to incorporate any novel technique
that improves our systems’ performance.
Return to
top of page
Risk Considerations
How do you
construct your programs to be conservative, moderate or aggressive?
We use money management techniques and allocation limits among asset
classes for designing five strategies from aggressive to conservative,
with all risk levels incorporating many of the same systems but with
different allocations.
How do you define
risk? Our primary measure of risk is maximum daily drawdown over
the entire history of the program. This is somewhat different than
using standard deviation or other measures, primarily because we’re
focused on the fact that an investor can handle small downside
volatility, but if the decline gets too big, the time to recover is much
longer, and the total return is impaired.
How do you
control risk in the portfolio? We use a number of approaches for
risk management, including diversification among asset classes and
diversification among management systems. Our systems are composed of a
number of separate models that look at different price data to build a
consensus, or to assess the longer-term environment for a specific
system, with the objective of picking periods for investment that have
the highest probability for success. We don’t have any single system
that dominates the program. To avoid curve fitting, we use similar
parameters for a variety of strategies. We also monitor all systems
daily with our automated software for problems with downside volatility.
Do you have
specific money management and risk control rules? Will you ever
deviate from those rules? Why? Each of our systems has a specific
maximum drawdown to control the risk. Our rule is that if a system is
at its maximum historical drawdown, the system is placed on hold until
we assess the cause of the problem, and what fixes are in order for
improvement. The discontinuation of one system does not affect the
other systems in a program, and the program can continue its automated
operation. Other rules we have include maximum exposure to any one
asset class or system, and using only mutual funds that allow exchanges
without limitations.
What do you feel
is the worst-case scenario in terms of maximum drawdowns for this model
going forward? Our maximum drawdowns are based on the risk level
for each of five programs that use the systems. The maximum drawdown
parameters have been: 10% for aggressive, 8% for capital appreciation,
6% for moderate, 5% for income & growth, and 4% for conservative.
Do you have any
optimization techniques in your model? To the extent that we’ve
designed each system so that it has performed well in the past, and it
continues to perform well in real time, we are optimizing our
strategies. By contrast, one of our primary objectives in developing
models is to avoid optimization in setting parameters for technical
analysis. Our experience shows that optimizing can give you a false
sense of expectation in real time.
To what extent do
you use cash? What is the minimum or maximum amount of cash you would
hold? Under what conditions would this position fluctuate? In our
approach using multiple asset class dynamic asset allocation, cash (or
money market funds) is the asset class of “last resort,” used when the
environment for all of the other asset classes in a system is
unfavorable. At any one time, the entire model may be anywhere from 0%
to 100% in cash, with the amount fluctuating daily as other asset
classes come into and out of favor.
Any new key
personnel? Key personnel have remained the same since Brian Kern,
PhD, started providing technical and analytical support in computer
software design, over 2 years ago. Last year we added two new people,
one in the electronic trading department and one in the customer service
and compliance department.
Any key personnel
depart the firm? No.
Has the business
structure changed? If so, do you anticipate this affecting operations?
Our company is now registered with the SEC, a change that went into
effect on January 1, 2004. Prior to that, our Registered Investment
Advisory filing was with the State of Indiana. There have been no other
changes in the business structure since the business was established in
1987.
How has your
investment philosophy changed relative to your management of this
model? In general? Our overall investment philosophy has not
changed since 1990; however, our application of the philosophy and
research have provided improvement with more data and better ideas, with
an accelerating improvement over the last 2 years. New indicators have
been added and the old ones have been refined. The definition of risk
has evolved from the industry’s use of standard deviation to the maximum
drawdown concept over the last 2 years. We have an on-going program to
review and research programs, and to add improvements when they become
apparent.
