See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/333664423

Factor Integration Based on Neural Networks for Factor Investing

Chapter · June 2019

DOI: 10.1007/978-3-030-22744-9_22

CITATIONS READS

0 364

4 authors, including:

Zhichen Lu Wen Long

Chinese Academy of Sciences China University of Geosciences (Beijing)
3 PUBLICATIONS 350 CITATIONS 35 PUBLICATIONS 652 CITATIONS

SEE PROFILE SEE PROFILE

Yingjie Tian
Chinese Academy of Sciences
79 PUBLICATIONS 1,437 CITATIONS

SEE PROFILE

All content following this page was uploaded by Zhichen Lu on 05 May 2020.

The user has requested enhancement of the downloaded file.

 Factor Integration Based on Neural
Networks for Factor Investing

Zhichen Lu1,2 , Wen Long1,2(B) , Jiashuai Zhang2,3 , and Yingjie Tian1,2,3

1
School of Economics and Management,
University of Chinese Academy of Sciences,
Beijing 100190, People’s Republic of China
longwen@ucas.ac.cn
2
Research Center on Fictitious Economy & Data Science,
Chinese Academy of Sciences, Beijing 100190, People’s Republic of China
3
School of Mathematical Sciences, University of Chinese Academy of Sciences,
Beijing 100190, People’s Republic of China

Abstract. Factor investing is one kind of quantitative investing

methodologies for portfolio construction based on factors. Factors with
different style are extracted from multiple sources such as market data,
fundamental information from financial statements, sentimental infor-
mation from the Internet, etc. Numerous style factors are defined by
Barra model proposed by Morgan Stanley Capital International(MSCI)
to explain the return of a portfolio. Multiple factors are usually inte-
grated linearly when being put to use, which ensures the stability of the
process of integration and enhances the effectiveness of integrated fac-
tors. In this work, we integrate factors by machine learning and deep
learning methodologies to explore deeper information among multiple
style factors defined by MSCI Barra model. Multi-factors indexes are
compiled using Smart Beta Index methodology proposed by MSCI. The
results show non-linear integration by deep neural network can enhance
the profitability and stability of the index compiled according to the
integrated factor.

Keywords: Neural networks · Deep learning · Factor investing

1 Introduction

The definition of factors of factor investing originates from “Arbitrage pricing

theory” proposed by Ross [10], which holds that the expected return of a finan-
cial asset can be modeled as a function of various macroeconomic factors or
theoretical market indexes. And then researchers have tried to use specific fac-
tors to model the return of stocks. Three-factors model [4] was the primary one
which modeled excess return of stock by book value, earning. Further researches
verified a series of factors can be used to explain the return of investing in
stocks, factors can be summarised into three main categories: macroeconomic,
c Springer Nature Switzerland AG 2019
J. M. F. Rodrigues et al. (Eds.): ICCS 2019, LNCS 11538, pp. 286–292, 2019.
https://doi.org/10.1007/978-3-030-22744-9_22
 Factor Integration Based on Neural Networks for Factor Investing 287

statistical, and fundamental. In risk model developed by Barra team from MSCI
company, factor returns are estimated through cross-sectional regression [8]. Fac-
tor portfolios were built according to target factors to construct factor returns in
Fama-French approach [1,4]. Similarly, Smart Beta Index from MSCI company
[2,3] is compiled according to target factors to reflect the style and performance
of specific factors under the different market situation. When being put to use,
multiple factors usually need to be integrated, a common way to integrate fac-
tors is a linearly weighted sum, and weights of each factor are calculated by
solving an optimization with subjectively defined target [3]. In recent years,
non-linear methods such Support Vector Machine, Logistic Regression, Random
Forest, Neural Networks and deep learning methodologies are well used in finan-
cial time series modeling, yet most existing works focus on stock price prediction.
They learn parameters of models by fitting training samples and presume that
the distribution of the training set and test set in the feature space are identical
[9,13–15]. In the aspect of cross-section modeling and feature integration, only
several works exist [5,6].
In our works, we introduce neural networks into the task of cross-section
factor integration, and we extract factors according to the definition from Barra
[8]. We use Smart Beta Index methodology to compile factor indexes to reflect
performance and style of them on the Chinese market. Experimental results show
the index that compiled based on factors integrated by neural networks results
in better profitability and stability.

2 Factors and Factor Indexes

The changes of the stock price are not just a result of historical market behavior,
but also affected by information from multiple sources such as macroeconomy
and financial situation of the corresponding listed company. Indicators can be
selected and defined to capture this information for usage on investment prac-
tice, and they are called factors. Factors are extracted from three main sources:
technical indicators from market samples, fundamental indicators from financial
statements and macroeconomic indicators.
When used in market practices, stocks are ranked and selected according to
scores calculated by one or multiple factors. Factors that proven to be robust
through a long time period are summarized by Barra risk model. Table 1 present
the definition of factors. Original indicators are extracted from market data of
stocks and financial statement of their corresponding listed companies. Factors
are usually sampled in monthly frequency when being used.
To reflect performances of factors on market practices, factor indexes are
compiled according to methodologies proposed by MSCI company. At beginning
of each season component stocks of benchmark CSI 800 are sorted by factor score,
and top 100 are selected as component of factor index and weighted according
to their market value. For single factor indexes, component stocks are sorted
by single target factor, for multi-factors indexes, weights of component stocks
are calculated by solving optimization whose objective are maximizing multiple
target factors:
288 Z. Lu et al.

Table 1. BARRA style factors

Factors Meaning Indicators

Size Size of listed company Market Value of listed company
Momentum Degrees of trend Risk adjusted returns of recent
mean(r )
20 days: std(r 20)
20
Non-linear size Middle level of size Residual of the regression
between size and third power of
size
Volatility Uncertainty of bias from Standard deviation of active
market return
The cumulative sum of the
active return
Standard deviation of daily
return
Value (BTOP) Book value to market Price earnings ratio (PE)
value
Market-to-book ratio (PBR)
Price-to-sale ratio (PS)
Liquidity Volume and frequency of Monthly logarithm turnover
trading rate
Mean value of monthly
logarithm turnover rate in
recent 3 month
Mean value of monthly
logarithm turnover rate in
recent 12 month
Growth Growth of listed company Net profit (YoY)
Total asset (YoY)
Operating revenue (YoY)
Dividend (Earning Yield) Profitability of listed Dividend yield
company
Dividend per share
Dividend to market value
Quality Quality of listed company Debt to equity
ROE
Leverage Leverage situation of Market leverage
listed company
Debt to asset
Book leverage

K 
 n
target
max ωi Xik
k=1 i=1

n
non−target 
n
non−target
s.t. ωi Xik ≥ ωibenchmark Xik − 0.25 ∗ std(Xknon−target ),
i=1 i=1
k = 1, 2, 3 . . . , K̃

n
non−target 
n
non−target
ωi Xik ≤ ωibenchmark Xik + 0.25 ∗ std(Xknon−target ),.
i=1 i=1
k = 1, 2, 3 . . . , K̃
max(0, ωibenchmark − 2%) ≤ ωi ≤ max(10ωibenchmark , ωibenchmark + 2%),
i = 1, 2, 3 . . . , n
 Factor Integration Based on Neural Networks for Factor Investing 289

According to this methodology we compile single factor indexes and multi-factor

indexes with target on Momentum, Size, Value, Dividend, which follows docu-
ment from MSCI. Figure 1 is back-test results of factor indexes during 2010 to
2017. Factors present different style among different market situation. Profitabil-
ity and risk of each factors are evaluated by indicators listed in Table 2, from
which we can see that factor indexes reach higher returns and Sharpe ratio than
benchmark, which verified the effectiveness of these factors on Chinese market.
Moreover, subjectively setting the objective of optimization for factor integra-
tion may lead to unsatisfied result on profitability and risk, since factors show
different performance in different market.

Fig. 1. Smart Beta factor indexes based on CSI 800.

Table 2. Smart Beta Index simulation results based on CSI 800

Return Annual Return Volatility Downside Beta VaR Alpha Beta Sharpe Sortino Loss rate MDD Active Return

Dividend 105.755% 10.425% 26.987% 1.0207 -2.739% 0.0860 1.0466 0.3612 0.2107 43.820% -48.622% 94.791%
Growth 63.471% 6.988% 27.255% 1.0594 -2.779% 0.0548 1.0627 0.2437 0.0977 43.820% -47.873% 52.507%
Vol 113.650% 10.998% 20.561% 0.7441 -2.082% 0.0838 0.7400 0.4244 0.3072 43.820% -33.409% 102.686%
Value 26.687% 3.304% 23.491% 0.8553 -2.410% 0.0146 0.8512 0.0921 -0.0242 48.315% -37.260% 15.722%
Quality 35.915% 4.308% 26.189% 0.9965 -2.678% 0.0275 0.9840 0.1454 0.0127 38.202% -45.286% 24.950%
Momentum 76.557% 8.126% 26.901% 1.0443 -2.737% 0.0644 1.0230 0.2831 0.1357 46.067% -47.984% 65.592%
Reversal 62.528% 6.903% 26.239% 1.0080 -2.676% 0.0524 0.9953 0.2397 0.1005 46.067% -43.197% 51.564%
Multi Factors 44.600% 5.199% 28.772% 1.1226 -2.934% 0.0387 1.0962 0.1869 0.0375 41.573% -54.206% 33.636%
Benchmark 10.964% 1.440% 24.670% 1.0011 -2.533% 0.0000 1.0000 0.0249 -0.0901 47.191% -48.984% 0.000%

3 Neural Networks for Factor Integration

Deep learning methodology is explored on stock price prediction [7,11,12], and
deep neural networks are designed to extract features from time series samples
for prediction. Portfolio construction is another kind of market practice which
provides cross-section level samples. In this work, we introduce Multi-layer Per-
ceptron (MLP) to deal with cross-section factors. Traditional machine learning
and linear regression are also applied in the experiment for comparison.
290 Z. Lu et al.

We use factors of each component stock of CSI 800 index from 2008 to 2017
for the experiment. Models are trained at the start of every year using monthly
samples {χit , yti } from previous 3 years, where χit denotes factors listed in Table 1
of stock i, and yti denotes return of from t to t + 1. At the start of each month,
factors of each stock are integrated by models trained at the start of that year,
and stocks are sorted according to integrated factors, and top 100 stocks are
used for index compilation and weighted according to their market size.

Fig. 2. Model integrated factor indexes based on CSI 800.

Results of indexes compiled based on integrated factors are performed in

Fig. 2, from which we can see that the net value of most models based integrated
factor indexes outperform benchmarks during most part of the back-test period.
We further evaluate each index by the same performance indicators listed in
Table 3. From the results of performance indicators, we can conclude that: (1)
Factors integrated by neural networks and linear regression show better perfor-
mance on profitability and stability than the multi-factors index. It implies that
the model based integration can potentially mine the relationship between fac-
tors of stocks and their future performances. On the one hand, neural networks
and linear regression based indexes show higher return than multi-factor indexes,
on the other hand, volatility of multi-factor index is higher which means higher
risk. Moreover, the higher Sharpe ratio still implies higher stability. (2) Neural
networks show better performance than linear regression, which means the non-
linear relationship between factors can be used to enhance the performance of
integrated factors.
 Factor Integration Based on Neural Networks for Factor Investing 291

Table 3. Integrated factor indexes simulation results based on CSI 800

Return Annual Return Volatility Downside Beta VaR Alpha Beta Sharpe Sortino Loss rate MDD Active Return

Multi Factors 44.600% 5.199% 28.772% 1.1226 -2.934% 0.0387 1.0962 0.1869 0.0375 41.573% -54.206% 33.636%
Benchmark 10.964% 1.440% 24.670% 1.0011 -2.533% 0.0000 1.0000 0.0249 -0.0901 47.191% -48.984% 0.000%
SVR -5.746% -0.810% 25.275% 0.9734 -2.606% -0.0228 0.9867 -0.0589 -0.1696 48.315% -52.491% -16.710%
Random Forest 25.914% 3.218% 25.691% 1.0176 -2.630% 0.0173 0.9980 0.1020 -0.0240 44.944% -51.071% 14.950%
Neural Network 68.440% 7.429% 25.615% 0.9938 -2.602% 0.0563 0.9537 0.2590 0.1145 44.944% -52.198% 57.476%
Linear Regression 60.386% 6.708% 24.650% 0.9513 -2.508% 0.0487 0.9185 0.2316 0.0953 43.820% -51.341% 49.421%

4 Conclusion
Factor indexes reflect performances of factors for factor investing so that robust
factors can be filtered. Filtered factors need to be further integrated, our work
introduces deep neural networks and other supervised models to integrate factors
supervised by future return. And indexes are compiled according to integrated
factors to evaluate their performance. Experimental results show that supervised
integration by the model can enhance the effectiveness of integrated factors com-
pared to integration by optimization with a subjectively defined objective. And
Neural network is verified to be more effective since it is able to mine deep
non-linear relationship between factors and future performance of stock price.

Acknowledgement. This research was partly supported by the grants from National
Natural Science Foundation of China (No. 71771204, 71331005, 91546201).

References
1. Ang, A.: A five-factor asset pricing model. Fama-Miller Working Paper (2014)
2. Bender, J., Briand, R., Melas, D., Subramanian, R.: Foundations of factor investing
(2013)
3. Bender, J., Briand, R., Melas, D., Subramanian, R.A., Subramanian, M.: Deploying
multi-factor index allocations in institutional portfolios. In: Risk-Based and Factor
Investing, pp. 339–363. Elsevier (2015)
4. Fama, E.F., French, K.R.: The cross-section of expected stock returns. J. Finance
47(2), 427–465 (1992)
5. Gu, S., Kelly, B.T., Xiu, D.: Empirical asset pricing via machine learning. SSRN
(2018). https://doi.org/10.2139/ssrn.3159577
6. Krauss, C., Do, X.A., Huck, N.: Deep neural networks, gradient-boosted trees,
random forests: statistical arbitrage on the S&P 500. Eur. J. Oper. Res. 259(2),
689–702 (2017)
7. Long, W., Lu, Z., Cui, L.: Deep learning-based feature engineering for stock
price movement prediction. Knowl.-Based Syst. 164, 163–173 (2019). http://www.
sciencedirect.com/science/article/pii/S0950705118305264
8. Menchero, J., Orr, D., Wang, J.: The Barra US equity model (USE4) methodology
notes. MSCI Model Insight (2011)
9. Rivest, R.L.: Learning decision lists. Mach. Learn. 2(3), 229–246 (1987)
10. Ross, S.A.: The arbitrage theory of capital asset pricing. In: Handbook of the
Fundamentals of Financial Decision Making: Part I, pp. 11–30. World Scientific
(2013)
292 Z. Lu et al.

11. Shen, F., Chao, J., Zhao, J.: Forecasting exchange rate using deep belief networks
and conjugate gradient method. Neurocomputing 167, 243–253 (2015)
12. Singh, R., Srivastava, S.: Stock prediction using deep learning. Multimed. Tools
Appl. 76(18), 18569–18584 (2017)
13. Valiant, L.G.: A theory of the learnable. Commun. ACM 27(11), 1134–1142 (1984)
14. Xiong, T., Li, C., Bao, Y., Hu, Z., Zhang, L.: A combination method for interval
forecasting of agricultural commodity futures prices. Knowl.-Based Syst. 77(C),
92–102 (2015)
15. Zhou, T., Gao, S., Wang, J., Chu, C., Todo, Y., Tang, Z.: Financial time series
prediction using a dendritic neuron model. Knowl.-Based Syst. 105(C), 214–224
(2016)

View publication stats