QuanTek is an Econometrics program created for
the purpose of designing and testing optimal technical indicators and trading
rules, as well as conducting general studies of the dynamics of financial
markets. This is accomplished by making use of state-of-the-art techniques
of Adaptive Filter
Theory as well as a variety of sophisticated statistical tests.
QuanTek makes use of a variety of different Linear Prediction
filters as well as the Savitzky-Golay smoothing filter. These digital filters make it possible to design a
whole new category of technical indicators that are much more
sophisticated than the traditional ones. The sophisticated statistical
tests make it possible to design custom technical
indicators and then test their effectiveness over a range of
past data. In this way different indicators can be tested
and optimized and their effectiveness for trading and investing
compared.
QuanTek works by defining various functions of past returns
(prices) using the different Linear Prediction and Savitzky-Golay
smoothing filters, along with the more usual Moving Averages.
These functions are the technical indicators and are designed to display
maximum correlation with future returns. This correlation
can be tested using the statistical tests in QuanTek.
This correlation is not so easy to find, because it is usually masked by
high-frequency stochastic noise. But it can be uncovered using the
appropriate filtering techniques. Then once a technical indicator
is found that displays positive correlation with future returns,
this indicator can serve as the basis of a set of trading rules.
Due to the verified correlation, the trading rules are then
effective and their effectiveness can be verified in a variety of realistic
trading scenarios by means of the Diagnostic Test. Having verified
the correlation with future returns of the technical indicator
and positive gains from the Diagnostic Test over the past data set of the
trading rules, this set of trading rules may then be expected to
yield positive results at least a short time into the future. In the
upcoming version 3.3 there will also be incorporated an automatic
optimization routine, which optimizes the trading rules
automatically for each day, then tests this optimization for the next day's
data. In this way the persistence of the optimization can be
tested, rather than just being assumed based on the optimization
over the past data set.
QuanTek
is also a portfolio management program, making use of the Markowitz Model
to construct an optimal portfolio. You can use this
optimization routine to adjust your portfolio to maximize
return and minimize risk in the overall portfolio. This risk
control is very important in connection with short-term trading. The
novel feature of this portfolio optimization is that the expected
returns are estimated by means of the Linear Prediction filter and Savitzky-Golay
smoothing filter.
QuanTek now works with either TeleChart or MetaStock
data, or directly with ASCII files in a variety of formats.
Price Projection
 | Price Projection: A Price Projection is computed using one of a choice of five Linear
Prediction filters, and displayed on the Main Graph. This Price
Projection is also used as part of the computation of the Momentum
indicators and Trading Rules. One
of the LP filters
uses the Wavelet spectrum, one uses the Periodogram
spectrum, and one computes the autocorrelation
directly. At present, all of these filters assume the returns
data to be a stationary
stochastic process over the past 1024 days, which seems a
reasonable approximation to the true non-stationary stochastic process
underlying the financial markets. You can experiment with the results
of using the different LP filters. |
| Hybrid LP
Filter: The Hybrid LP
Filter dialog is used to set the Linear
Prediction filter for each security separately. The type of
filter may be chosen, and two parameters which are called
the Order of Approximation and the Fractal Dimension can be
set. The Order of Approximation selects the degree of smoothing
of the filter spectrum, by setting the length of the series of LP
coefficients. The Fractal Dimension sets the low-frequency
response of the filter, corresponding to modeling the time series as a fractionally
differenced (FD) or long-memory process. (The Wavelet
filter works well with the default values of these settings.) The Hybrid
LP Filter dialog also has attached to it a
dialog box which displays the correlation between the filter output
(returns) , and
the future returns. Using this Correlation Test -
Filters dialog, the predictive power of each filter can be tested
directly for any settings of the filter type and parameters. |
Technical Indicators
| Technical Indicators: The main feature of QuanTek is
that you can design and test yourself a wide variety of technical
indicators based on the Linear Prediction filter and the Savitzky-Golay
smoothing filter. These consist of practically every conceivable
oscillator-type indicator that can be made out of the past price data.
These indicators can then be tested for correlation with future
returns using the statistical tests in QuanTek.
After a set of three of these indicators is designed, the Trading Rules
indicator is formed by a weighted sum of these three indicators, with
various filter rules applied. The effectiveness of the final Trading
Rules indicator can be tested in a variety of realistic trading
scenarios by means of the Diagnostic Test. |
| Harmonic Oscillator: A set of technical indicators is displayed in a splitter window,
with three panes, and is called the Harmonic Oscillator
display. This set of three indicators is called the Relative Price,
Velocity, and Acceleration indicators. They are based on
the Savitzky-Golay smoothing of the past prices, with zero, one, and two derivatives, respectively.
This filter is set to be acausal, so there is no time lag (zero-phase
filter). This is so the features will all line up properly. Then
the past buy/sell signals and buy/sell points are derived from the Harmonic
Oscillator graphs and displayed on the Main Graph as green/red
rectangles and arrows, respectively, and on the splitter windows the buy/sell
points are displayed as green/red vertical lines. (The buy/sell
points are the first of a series of buy/sell signals.)
These buy/sell
points serve to line up features on all the graphs, and mark the actual
optimum buy/sell
points for a given time horizon, for comparison to the Momentum
indicators that you design and test yourself. |
| Momentum Indicators: Another set of technical indicators is displayed in a second
splitter window, with three panes. These are the three Momentum indicators that you
design yourself, using the Technical Indicators dialog
box. You can design a technical indicator of one of the three
types listed under the Harmonic Oscillator indicator, and you can also
set the time scale for smoothing of the indicator for any value from
0 to 512 days. You can also vary the lead time, or phase
of the technical indicators, for maximum correlation with future
returns, and you can reverse their sign. These technical
indicators are causal in the sense that they depend only
on past data, with no future data included. To compute
these, you need to compute 1024 Price Projections, one for each day
in the past going back 1024 days, which is why they are much more
computation-intensive than the Harmonic Oscillator indicators, which
are acausal. You can then vary the proportion of these three Momentum
indicators to form the Trading Rules indicator, along with
specifying a set of filter rules, which gives the set of rules for short-term trading. The effectiveness of the three Momentum indicators
can be
measured by means of the Correlation Test - Indicators, and the effectiveness of the
Trading Rules in a variety of realistic trading scenarios can be tested directly by means
of a back-testing routine called the Diagnostic Test. |
| Forward MA Indicators: The third set of technical indicators is displayed in a third
splitter window, also with three panes. In the bottom pane is the Weighted
Sum of Momentum Indicators just described, with weights that you can set (for each stock). This
indicator forms the basis for the N-day trading rules, after
applying the filter rules, which
are listed on the Short-Term Trades dialog box and in the Portfolio
Report. The other two indicators shown in this splitter window are the N-day
future returns, where N is the time horizon, which
you set for each security individually and can range from 1 to 40 days. This is
the N-day future moving average of the returns, which is
placed next to the Weighted Sum of Momentum Indicators, because the two are supposed to
be correlated. Also shown is an N-day future volatility,
which may be used in a future version of QuanTek which
incorporates GARCH. |
Statistical Tests
| Correlation Tests: The most important statistical tests in QuanTek
are the tests for correlation with future returns. There
are two of these tests, called the Correlation Test - Filters and the
Correlation Test - Indicators. The Correlation Test -
Filters tests the output of the Linear Prediction filters, the Price
Projection (returns), directly for correlation with future
returns, which is what they are supposed to predict. This test
verifies that the LP filters are working and compares the performance
of the different filters. It turns out that several of the LP
filters, in particular the Wavelet filter, show nice correlation with
future returns, but the time horizon must be set to something
like 10 days or greater, otherwise the correlation gets masked by
high-frequency stochastic noise. The Correlation Test -
Indicators tests for correlation between the Momentum
indicators and future returns. This enables the optimum
indicators to be chosen for the most effective Trading Rules for a
given time horizon. The time horizon can also be set
from either of these dialogs. |
| Spectrum Tests: Two of the statistical tests are the standard Periodogram
spectrum and the Wavelet spectrum, which are two different
methods for computing the power spectrum of the (log) price
returns. According to the theory of stationary time series, if
the (smoothed) spectrum of the returns series is flat, then the returns are
a white noise series, and the log price returns form a stationary
stochastic process which when integrated is known as a Random Walk. However, in many
cases there is a significant deviation from a white noise spectrum, and these cases imply a non-zero autocorrelation
sequence (the Fourier transform of the power spectrum),
implying (partial) predictability of the future returns.
The LP filter coefficients may also be computed directly from the power
spectrum. |
| Scatter Graph: One more statistical test is a scatter graph which measures correlation between the returns of two different securities or the same
security. This scatter graph has attached to it a
display of the measured autocorrelation sequence or cross-correlation
sequence, between the returns of one security or of two different securities,
respectively. This test is useful for choosing securities in the optimal
portfolio, as well as general statistical studies of the market. |
Portfolio Optimization
| Markowitz Method: The Markowitz Method of portfolio optimization
is familiar from Modern Portfolio Theory. This method makes
use of an estimated expected return for each security in the
portfolio, and also computes the variance matrix of the returns
between all the securities in the portfolio. Then the goal is to find
the proportion of each security that maximizes returns for the
overall portfolio, and which also minimizes risk. The risk
is defined as the standard deviation (square root of variance)
for the portfolio return. This portfolio variance is in turn
computed from the variance matrix, obtained by measuring the volatility
of each security and the correlation of the returns between all the
securities in the portfolio. The computation is a complex one in
linear programming in general, but in QuanTek a slight
modification to this procedure is made to simplify the calculation.
Instead of letting the sum of the amounts invested in each security remain
constant, the square root of the sum of the squares (root mean square value)
remains constant. This has the effect that the total amount invested
will be decreased somewhat unless the amount of equity allocated to each
security happens to be the same. But this is actually an advantage,
because it helps to control risk. You can set the risk tolerance,
the opposite of risk aversion, in QuanTek to adjust the
optimization of the portfolio for different desired ratios of overall
portfolio return to risk. |
Other Features
| Main Graph: The scrolling graph has four different scales that cover the whole data set.
There is no limit to the time span of historical data that
can be displayed. The graphs are now available in either black or white
background. The black background looks great! These four graph
scales display prices together with various smoothings, the Price Projection, various buy/sell signals, Bollinger Bands, and on the
highest scale is a Candlestick Graph. |
| S-T Trades Dialog: The Short-Term Trades Dialog is a modeless
dialog box that may be viewed from anywhere in the program simply by
right-clicking the mouse. It shows a summary of the securities in the
portfolio including the current values of the Trading Rules, and also
other information such as the recommended and actual positions in the optimal
portfolio, as well as the expected return of each security.
Of course the current price information is also displayed.
There is also a Portfolio Report which displays the same as well as
additional information, which is in the form of a text file that can be
printed or saved. These displays are convenient for short-term
trading as well as maintaining the optimal portfolio. |
|
