DOI QR코드

DOI QR Code

Estimation of GARCH Models and Performance Analysis of Volatility Trading System using Support Vector Regression

Support Vector Regression을 이용한 GARCH 모형의 추정과 투자전략의 성과분석

  • 김선웅 (국민대학교 비즈니스IT전문대학원) ;
  • 최흥식 (국민대학교 비즈니스IT전문대학원)
  • Received : 2017.05.12
  • Accepted : 2017.06.16
  • Published : 2017.06.30

Abstract

Volatility in the stock market returns is a measure of investment risk. It plays a central role in portfolio optimization, asset pricing and risk management as well as most theoretical financial models. Engle(1982) presented a pioneering paper on the stock market volatility that explains the time-variant characteristics embedded in the stock market return volatility. His model, Autoregressive Conditional Heteroscedasticity (ARCH), was generalized by Bollerslev(1986) as GARCH models. Empirical studies have shown that GARCH models describes well the fat-tailed return distributions and volatility clustering phenomenon appearing in stock prices. The parameters of the GARCH models are generally estimated by the maximum likelihood estimation (MLE) based on the standard normal density. But, since 1987 Black Monday, the stock market prices have become very complex and shown a lot of noisy terms. Recent studies start to apply artificial intelligent approach in estimating the GARCH parameters as a substitute for the MLE. The paper presents SVR-based GARCH process and compares with MLE-based GARCH process to estimate the parameters of GARCH models which are known to well forecast stock market volatility. Kernel functions used in SVR estimation process are linear, polynomial and radial. We analyzed the suggested models with KOSPI 200 Index. This index is constituted by 200 blue chip stocks listed in the Korea Exchange. We sampled KOSPI 200 daily closing values from 2010 to 2015. Sample observations are 1487 days. We used 1187 days to train the suggested GARCH models and the remaining 300 days were used as testing data. First, symmetric and asymmetric GARCH models are estimated by MLE. We forecasted KOSPI 200 Index return volatility and the statistical metric MSE shows better results for the asymmetric GARCH models such as E-GARCH or GJR-GARCH. This is consistent with the documented non-normal return distribution characteristics with fat-tail and leptokurtosis. Compared with MLE estimation process, SVR-based GARCH models outperform the MLE methodology in KOSPI 200 Index return volatility forecasting. Polynomial kernel function shows exceptionally lower forecasting accuracy. We suggested Intelligent Volatility Trading System (IVTS) that utilizes the forecasted volatility results. IVTS entry rules are as follows. If forecasted tomorrow volatility will increase then buy volatility today. If forecasted tomorrow volatility will decrease then sell volatility today. If forecasted volatility direction does not change we hold the existing buy or sell positions. IVTS is assumed to buy and sell historical volatility values. This is somewhat unreal because we cannot trade historical volatility values themselves. But our simulation results are meaningful since the Korea Exchange introduced volatility futures contract that traders can trade since November 2014. The trading systems with SVR-based GARCH models show higher returns than MLE-based GARCH in the testing period. And trading profitable percentages of MLE-based GARCH IVTS models range from 47.5% to 50.0%, trading profitable percentages of SVR-based GARCH IVTS models range from 51.8% to 59.7%. MLE-based symmetric S-GARCH shows +150.2% return and SVR-based symmetric S-GARCH shows +526.4% return. MLE-based asymmetric E-GARCH shows -72% return and SVR-based asymmetric E-GARCH shows +245.6% return. MLE-based asymmetric GJR-GARCH shows -98.7% return and SVR-based asymmetric GJR-GARCH shows +126.3% return. Linear kernel function shows higher trading returns than radial kernel function. Best performance of SVR-based IVTS is +526.4% and that of MLE-based IVTS is +150.2%. SVR-based GARCH IVTS shows higher trading frequency. This study has some limitations. Our models are solely based on SVR. Other artificial intelligence models are needed to search for better performance. We do not consider costs incurred in the trading process including brokerage commissions and slippage costs. IVTS trading performance is unreal since we use historical volatility values as trading objects. The exact forecasting of stock market volatility is essential in the real trading as well as asset pricing models. Further studies on other machine learning-based GARCH models can give better information for the stock market investors.

주식시장의 주가 수익률에 나타나는 변동성은 투자 위험의 척도로서 재무관리의 이론적 모형에서뿐만 아니라 포트폴리오 최적화, 증권의 가격 평가 및 위험관리 등 투자 실무 영역에서도 매우 중요한 역할을 하고 있다. 변동성은 주가 수익률이 평균을 중심으로 얼마나 큰 폭의 움직임을 보이는가를 판단하는 지표로서 보통 수익률의 표준편차로 측정한다. 관찰 가능한 표준편차는 과거의 주가 움직임에서 측정되는 역사적 변동성(historical volatility)이다. 역사적 변동성이 미래의 주가 수익률의 변동성을 예측하려면 변동성이 시간 불변적(time-invariant)이어야 한다. 그러나 대부분의 변동성 연구들은 변동성이 시간 가변적(time-variant)임을 보여주고 있다. 이에 따라 시간 가변적 변동성을 예측하기 위한 여러 계량 모형들이 제안되었다. Engle(1982)은 변동성의 시간 가변적 특성을 잘 반영하는 변동성 모형인 Autoregressive Conditional Heteroscedasticity(ARCH)를 제안하였으며, Bollerslev(1986) 등은 일반화된 ARCH(GARCH) 모형으로 발전시켰다. GARCH 모형의 실증 분석 연구들은 실제 증권 수익률에 나타나는 두터운 꼬리 분포 특성과 변동성의 군집현상(clustering)을 잘 설명하고 있다. 일반적으로 GARCH 모형의 모수는 가우스분포로부터 추출된 자료에서 최적의 성과를 보이는 로그우도함수에 대한 최우도추정법에 의하여 추정되고 있다. 그러나 1987년 소위 블랙먼데이 이후 주식 시장은 점점 더 복잡해지고 시장 변수들이 많은 잡음(noise)을 띠게 됨에 따라 변수의 분포에 대한 엄격한 가정을 요구하는 최우도추정법의 대안으로 인공지능모형에 대한 관심이 커지고 있다. 본 연구에서는 주식 시장의 주가 수익률에 나타나는 변동성의 예측 모형인 GARCH 모형의 모수추정방법으로 지능형 시스템인 Support Vector Regression 방법을 제안한다. SVR은 Vapnik에 의해 제안된 Support Vector Machines와 같은 원리를 회귀분석으로 확장한 모형으로서 Vapnik의 e-insensitive loss function을 이용하여 비선형 회귀식의 추정이 가능해졌다. SVM을 이용한 회귀식 SVR은 두터운 꼬리 분포를 보이는 주식시장의 변동성과 같은 관찰치에서도 우수한 추정 성능을 보인다. 2차 손실함수를 사용하는 기존의 최소자승법은 부최적해로서 추정 오차가 확대될 수 있다. Vapnik의 손실함수에서는 입실론 범위내의 예측 오차는 무시하고 큰 예측 오차만 손실로 처리하기 때문에 구조적 위험의 최소화를 추구하게 된다. 금융 시계열 자료를 분석한 많은 연구들은 SVR의 우수성을 보여주고 있다. 본 연구에서는 주가 변동성의 분석 대상으로서 KOSPI 200 주가지수를 사용한다. KOSPI 200 주가지수는 한국거래소에 상장된 우량주 중 거래가 활발하고 업종을 대표하는 200 종목으로 구성된 업종 대표주들의 포트폴리오이다. 분석 기간은 2010년부터 2015년까지의 6년 동안이며, 거래일의 일별 주가지수 종가 자료를 사용하였고 수익률 계산은 주가지수의 로그 차분값으로 정의하였다. KOSPI 200 주가지수의 일별 수익률 자료의 실증분석을 통해 기존의 Maximum Likelihood Estimation 방법과 본 논문이 제안하는 지능형 변동성 예측 모형의 예측성과를 비교하였다. 주가지수 수익률의 일별 자료 중 학습구간에서 대칭 GARCH 모형과 E-GARCH, GJR-GARCH와 같은 비대칭 GARCH 모형에 대하여 모수를 추정하고, 검증 구간 데이터에서 변동성 예측의 성과를 비교하였다. 전체 분석기간 1,487일 중 학습 기간은 1,187일, 검증 기간은 300일 이다. MLE 추정 방법의 실증분석 결과는 기존의 많은 연구들과 비슷한 결과를 보여주고 있다. 잔차의 분포는 정규분포보다는 Student t분포의 경우 더 우수한 모형 추정 성과를 보여주고 있어, 주가 수익률의 비정규성이 잘 반영되고 있다고 할 수 있다. MSE 기준으로, SVR 추정의 변동성 예측에서는 polynomial 커널함수를 제외하고 linear, radial 커널함수에서 MLE 보다 우수한 예측 성과를 보여주었다. DA 지표에서는 radial 커널함수를 사용한 SVR 기반의 지능형 GARCH 모형이 가장 우수한 변동성의 변화 방향에 대한 방향성 예측력을 보여주었다. 추정된 지능형 변동성 모형을 이용하여 예측된 주식 시장의 변동성 정보가 경제적 의미를 갖는지를 검토하기 위하여 지능형 변동성 거래 전략을 도출하였다. 지능형 변동성 거래 전략 IVTS의 진입규칙은 내일의 변동성이 증가할 것으로 예측되면 변동성을 매수하고 반대로 변동성의 감소가 예상되면 변동성을 매도하는 전략이다. 만약 변동성의 변화 방향이 전일과 동일하다면 기존의 변동성 매수/매도 포지션을 유지한다. 전체적으로 SVR 기반의 GARCH 모형의 투자 성과가 MLE 기반의 GARCH 모형의 투자 성과보다 높게 나타나고 있다. E-GARCH, GJR-GARCH 모형의 경우는 MLE 기반의 GARCH 모형을 이용한 IVTS 전략은 손실이 나지만 SVR 기반의 GARCH 모형을 이용한 IVTS 전략은 수익으로 나타나고 있다. SVR 커널함수에서는 선형 커널함수가 더 좋은 투자 성과를 보여주고 있다. 선형 커널함수의 경우 투자 수익률이 +526.4%를 기록하고 있다. SVR 기반의 GARCH 모형을 이용하는 IVTS 전략의 경우 승률도 51.88%부터 59.7% 사이로 높게 나타나고 있다. 옵션을 이용하는 변동성 매도전략은 방향성 거래전략과 달리 하락할 것으로 예측된 변동성의 예측 방향이 틀려 변동성이 소폭 상승하거나 변동성이 하락하지 않고 제자리에 있더라도 옵션의 시간가치 요인 때문에 전체적으로 수익이 실현될 수도 있다. 정확한 변동성의 예측은 자산의 가격 결정뿐만 아니라 실제 투자에서도 높은 수익률을 얻을 수 있기 때문에 다양한 형태의 인공신경망을 활용하여 더 나은 예측성과를 보이는 변동성 예측 모형을 개발한다면 주식시장의 투자자들에게 좋은 투자 정보를 제공하게 될 것이다.

Keywords

References

  1. Bollslev, T., "Generalized Autoregressive Conditional Heteroskedasticity," Journal of Econometrics, Vol.31 (1986), 307-327. https://doi.org/10.1016/0304-4076(86)90063-1
  2. Brooks, C., "Predicting stock index volatility: Can market volume help?", Journal of Forecasting, Vol.17 (1998), 59-80. https://doi.org/10.1002/(SICI)1099-131X(199801)17:1<59::AID-FOR676>3.0.CO;2-H
  3. Cao, L. and F. Tay, "Financial forecasting using support vector machines," Neural Computing & Applications, Vol.10 (2001), 184-192. https://doi.org/10.1007/s005210170010
  4. Chen, S., W. Hardle, and K. Jeong, "Forecasting volatility with support vector machine-based GARCH model," Journal of Forecasting, Vol.29 (2010), 406-433.
  5. Chen, S., W. Hardle, and R. Moro, "Modeling default risk with support vector machines," Quantitative Finance, Vol.11 (2011), 135-154. https://doi.org/10.1080/14697680903410015
  6. Dao, T. and H. Ahn, "An optimized combination of $\pi$-fuzzy logic and support vector machine for stock market prediction", Journal of Intelligence and Information Systems, Vol. 20 (2014), 43-58. https://doi.org/10.13088/jiis.2014.20.4.43
  7. Engle, R.,"Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation," Econometrica, Vol.50 (1982), 987-1007. https://doi.org/10.2307/1912773
  8. Fiorentini, G., G. Calzolari, and L. Panattoni, "Analytic derivatives and the computation of GARCH estimates," Journal of Applied Econometrics, Vol.11 (1996), 399-417. https://doi.org/10.1002/(SICI)1099-1255(199607)11:4<399::AID-JAE401>3.0.CO;2-R
  9. Glosten, L.R., R. Jagannathan, and D.F. Runkle, "On the relation between the expected value and the volatility of the nominal excess return on stocks," Journal of Finance, Vol.48 (1993), 1779-1801. https://doi.org/10.1111/j.1540-6261.1993.tb05128.x
  10. Hamid, S. and Z. Iqbal, "Using neural networks for forecasting volatility of S&P 500 Index futures prices," Journal of Business Research, Vol.57 (2004), 1116-1125. https://doi.org/10.1016/S0148-2963(03)00043-2
  11. Kang, J. and D. Ryu, "A study on the empirical performance of GARCH-type models in the KOSPI 200 options market," Korean Journal of Financial Studies, Vol.38 (2009), 137-176.
  12. Kim, S.W., "Negative Asymmetric Relationship between VKOSPI and KOSPI 200," Journal of the Korean Data Analysis Society, Vol.12 (2010), 1761-1773.
  13. Kim, S.W. and H. Ahn, "Development of an intelligent trading system using support vector machines and genetic algorithms," Journal of Intelligence and Information Systems, Vol.16 (2010), 71-92.
  14. Mukherjee, S., E. Osuna, and F. Girosi, "Nonlinear prediction of chaotic time series using support vector machines," Proceedings of IEEE NNSP'97, 1997.
  15. Nelson, D.B., "Conditional heteroskedasticity in asset returns: A new approach," Econometrica, Vol.59 (1991), 347-370. https://doi.org/10.2307/2938260
  16. Park, S., S.W. Kim, and H.S. Choi, "Selection model of system trading strategies using SVM," Journal of Intelligence and Information Systems, Vol.20 (2014), 59-71.
  17. Perez-Cruz, F., A. Afonso-Rodriguez, and J. Giner, "Estimating GARCH models using support vector machines," Quantitative Finance, Vol.3 (2003), 1-10. https://doi.org/10.1080/713666155
  18. Ra, Y.S., H.S. Choi, and S.W. Kim, "VKOSPI forecasting and option trading application using SVM," Journal of Intelligence and Information Systems, Vol.22 (2016), 177-192. https://doi.org/10.13088/jiis.2016.22.4.177
  19. Roh, T.H., T. Lee, and I. Han, "Forecasting the volatility of KOSPI 200 using neural network-financial time series model," Korean Management Review, Vol.34 (2005), 683-713.
  20. Seo, B. and T. Lee, "A new algorithm for maximum likelihood estimation in normal scale-mixture generalized autoregressive conditional heteroskedastic models," Journal of Statistical Computation and Simulation, Vol.85 (2015), 202-215. https://doi.org/10.1080/00949655.2013.812092
  21. Tay, F. and L. Cao, "Application of support vector machines in financial time series forecasting," Omega, Vol.29 (2001), 309-317. https://doi.org/10.1016/S0305-0483(01)00026-3
  22. Vapnik, V., Estimation of Dependences based on Empirical Data, Springer Series in Statistics, 1982.
  23. Vapnik, V., The Nature of Statistical Learning Theory, Springer Verlag, 1995.
  24. Vapnik, V., Statistical Learning Theory, Wiley, 1998.
  25. Vapnik, V., S. Golowich, and A. Smola, "Support vector method for function approximation, regression estimation, and signal processing," Advances in Neural Information Processing Systems, 1996.