Signal Processing - May 2017 - 34

unreasonably they may seem. Excess economic benefits are
associated with risks.
Time-series models, system models, and statistical models are certainly useful tools to analyze economic data. SP
researchers can try to formulate the economic problems into
filtering, array SP, impulse responses, finite-state machine,
Markov models, and state-space models. Whenever a lagged
transform seems to be useful, researchers can consider the Z
transform, Laplace transforms, and spectral analysis, which
may provide additional insights into economic systems and
problem solutions.
Caution also needs to be exercised when formulating
input-output relationships and objective functions for an
economic system. In many cases, humans are only observers of the system and have little power to change the input,
e.g., consumer price indices, or the output, e.g., sales or GDP.
Meanwhile, humans are oftentimes also part of an economic
system, and any human actions or predictions may change the
system or system equilibrium. For example, a buying order
for a stock may inevitably change the price of the stock. There
are times when humans are indeed actively responsible for
the input of the system, such as an investment budget decision
on R&D or marketing for a company. The different nature
of inputs and outputs shapes not only problems and models
but also the methodology to the solution and interpretation
of results.
In terms of mathematical formulations and modeling tools,
many similarities exist between SP and econometrics. However, the weaknesses of SP methodology come from what is
often taken for granted in SP, such as the knowledge of physical systems, ground-truth data, controlled experiments, or
the confidence in simulated data. Mean-squared errors or
variances are often good statistics in SP performance evaluation but are not sufficient in economic and business studies.
Indeed, when making investment suggestions and decisions,
i.e., putting one's money to bet on a model, researchers need to
understand more of the probability and statistical significance
of their predictions, estimation, data samples, model validity,
etc. They need to understand causal relationships and examine
all possible alternative models and hypotheses about the data
because an accurate mechanism of a social or economic system is always unknown.
In the big data era, there is a wealth of data analysis and
processing work for which SP tools are useful, e.g., denoising
and data cleaning, feature extraction, pattern detection, tracking, and abnormality detection. Note that traditional economics are more about equilibrium, and SP can play a big role
in understanding the process and transition patterns to reach
equilibrium from data analysis.
Indeed, numerical data are at the core of SP. SP is promising in exploring economic big data to learn information better
and faster and in finding sophisticated subtle structures within
data. Researchers need to be careful not to be biased by the
data at hand.
As SP research and applications are expanding, SP researchers can broaden their vision and take advantage of the meth34

odologies and advancements in econometrics, economics, and
statistical methods for SP problems. For example, game theory
has been an active research tool in communication systems
and resource allocation. More cross-pollination between SP
and economics will produce fruitful results in research as well
as practical applications affecting people's daily lives.

Acknowledgments
We thank EidoSearch Inc. (www.eidosearch.com) for providing the empirical test results in the section "Big Data Analysis
of Financial Data Based on SP."

Authors
Xiao-Ping (Steven) Zhang (xzhang@ryerson.ca) received the
B.S. and Ph.D. degrees from Tsinghua University, Beijing,
China, in electronic engineering and the M.B.A. degree in
finance, economics, and entrepreneurship from the University
of Chicago, Illinois. He is a professor of electrical and computer engineering and is cross appointed to the Finance
Department at the Ted Rogers School of Management at
Ryerson University, Toronto, Canada. His research interests
include signal processing, electronic systems, machine learning, big data, finance, and marketing. He is the cofounder and
chief executive officer for EidoSearch, an Ontario-based company offering a content-based search and analysis engine for
financial big data.
Fang Wang (fwang@wlu.ca) received the M.B.A. degree
in finance from the University of Texas, San Antonio, and the
Ph.D. degree in management information systems from Michael
G. DeGroote School of Business, McMaster University, Hamilton,
Ontario, Canada. She is an associate professor of marketing at
the Lazaridis School of Business and Economics, Wilfrid
Laurier University, Waterloo, Ontario, Canada. Her research
interests include market response models, marketing data mining/analysis, long-term marketing productivity, information systems management, big data, and social media applications in
marketing and finance.

References

[1] IEEE Signal Process. Mag. (Special Issue on Signal Processing for Financial
Applications), vol. 28, no. 5, Sept. 2011.
[2] IEEE J. Sel. Topics Signal Process. (Special Issue on Methods in Finance and
Electronic Trading), vol. 6, no. 4, Aug. 2012.
[3] IEEE J. Sel. Topics Signal Process. (Special Issue on Financial Signal
Processing and Machine Learning for Electronic Trading), vol. 10, no. 6, Sept.
2016.
[4] E. F. Fama, "Efficient capital markets: A review of theory and empirical work,"
J. Finance, vol. 25, no. 2, pp. 383-417, May 1970.
[5] J. H. Cochrane, Asset Pricing, rev. ed. Princeton, NJ: Princeton Univ. Press,
2005.
[6] A. W. Lo, H. Mamaysky, and J. Wang, "Foundations of technical analysis:
Computational algorithms, statistical inference, and empirical implementation,"
J. Finance, vol. 55, no. 4, pp. 1705-1765, Aug. 2000.
[7] D. Kahneman and A. Tversky, "Prospect theory: An analysis of decision under
risk," Econometrica, vol. 47, no. 2, pp. 263-291, Mar. 1979.
[8] R. J. Shiller, "Do stock prices move too much to be justified by subsequent
changes in dividends?" Am. Econ. Rev., vol. 73, no. 1, pp. 236-237, Mar.
1983.
[9] E. F. Fama, "Market efficiency, long-term returns, and behavioral finance,"
J. Finance Econ., vol. 49, no. 3, pp. 283-306, Sept. 1998.

IEEE Signal Processing Magazine

May 2017

http://www.eidosearch.com

Table of Contents for the Digital Edition of Signal Processing - May 2017