Data Snooping Research Articles

The high sensitivity of pairs trading returns

Pairs trading is a simple and popular relative value trading strategy. This article deals with the most common implementation of the method based on a distance criterion. It demonstrates the high sensitivity of the return to changes in the length of the formation period and shows that a reasonable modification of this parameter may lead to generate positive excess returns which are robust to data snooping. This empirical result underlines the difficulty in understanding pairs trading returns, dynamics and sources of profitability through time.

Detecting synchronous cycles in financial time series of unequal length

This paper proposes a modification of an optimal test for cycles in multiple time series and applies it to test the hypothesis that there is a relationship between stock returns and the phases of the moon. No significant relationship is found, which is in line with the evidence from descriptive statistics. The fact that previous studies have reached different conclusions is traced to the use of inappropriate statistical methods and data snooping. • A modification of an optimal test for cycles in multiple time series is proposed. • The new test does not find any evidence of a lunar effect on stock returns. • Contradictory evidence of previous studies is traced to data snooping effects.

Data-snooping procedure applied to errors-in-variables models

The theory of Baarda’s data snooping — normal and F tests respectively based on the known and unknown posteriori variance — is applied to detect blunders in errors-invariables (EIV) models, in which gross errors are in the vector of observations and/or in the coefficient matrix. This work is a follow-up to an earlier work in which we presented the formulation of the weighted total least squares (WTLS) based on the standard least squares theory. This method allows one to directly apply the existing body of knowledge of the least squares theory to the errors-in-variables models. Among those applications, data snooping methods in an EIV model are of particular interest, which is the subject of discussion in the present contribution. This paper generalizes the Baarda’s data snooping procedure of the standard least squares theory to an EIV model. Two empirical examples, a linear regression model and a 2-D affine transformation, using simulated and real data are presented to show the efficacy of the presented formulation. It is highlighted that the method presented is capable of detecting outlying equations (rather than outlying observations) in a straightforward manner. Further, the WTLS method can be used to handle different TLS problems. For example, the WTLS problem for the conditions and mixed models, the WTLS problem subject to constraints and variance component estimation for an EIV model can easily be established. These issues are in progress for future publications.

The intra-day performance of market timing strategies and trading systems based on Japanese candlesticks

We develop market timing strategies and trading systems to test the intra-day predictive power of Japanese candlesticks at the 5-minute interval on the 30 constituents of the DJIA index. Around a third of the candlestick rules outperform the buy-and-hold strategy at the conservative Bonferroni level. After adjusting for trading costs, however, just a few rules remain profitable. When we correct for data snooping by applying the SSPA test on double-or-out market timing strategies, no single candlestick rule beats the buy-and-hold strategy after transaction costs. We also design fully automated trading systems by combining the best-performing candlestick rules. No evidence of out-performance is found after transaction costs. Although Japanese candlesticks can somewhat predict intra-day returns on large US caps, we show that such predictive power is too limited for active portfolio management to outperform the buy-and-hold strategy when luck, risk, and trading costs are correctly measured.

Multiple comparisons problem: Recent advances applied to energy and emissions

Within the field of empirical finance, the econometric analysis of markets commonly suffers from the well-established problem of data-snooping bias, whereby there is a likelihood that statistically significant results may be identified by pure random chance alone. This is the multiple comparisons problem resulting from Multiple Hypothesis Testing (MHT). Recent advances in MHT techniques to control the multiple comparisons problem are uniquely showcased within a VAR and Granger causality testing of energy and emissions market interactions. Four generalized p-value-based MHT techniques show no evidence of interactions between European Union Allowance (EUA) prices and a range of energy prices – spanning key oil, gas, coal and electricity markets – over the first half or so (2008–2010) of Phase II of the EU Emissions Trading Scheme. The generalized familywise error rate procedures show evidence of regional and cross-regional interactions within European electricity markets. However, in contrast, the more conservative false discovery proportion procedures identify much fewer statistically significant relationship and, indeed, show little evidence of such cross-regional electricity market interactions.

Evidências empíricas: arbitragem no mercado brasileiro com fundos ETFs

De acordo com a literatura de gestão de risco, a diversificação contribui para a mitigação do risco. Neste sentido, os fundos de índice Exchange Traded Funds (ETF), recém-introduzidos no mercado brasileiro, permitem sua fácil execução. Dentro deste contexto, o presente artigo investiga a eficiência do processo de valuation das cotas do fundo iShare Ibovespa com relação ao seu valor justo. Para isto, primeiramente é empregada uma análise das séries temporais de alta frequência do ETF e Ibovespa, seguido de simulações de estratégias que contemplem ágios/deságios entre as séries dos ativos, sem e com custos de transação. A fim de evitar efeitos de Data-Snooping nos resultados das operações, foi aplicado um Bootstrap para séries temporais. No primeiro momento os resultados apontam para uma ineficiência do apreçamento das cotas, visto que a incorporação de ágios/deságios na estratégia produziu retornos de 172,5% acima de seu índice. No segundo, verifica-se que mesmo com a introdução dos custos operacionais, os ganhos ainda assim apresentam ineficiência. Entretanto, a partir da técnica de Bootstrap, os resultados não apontaram para retornos excedentes, o que pode ser atribuído meramente ao fenômeno de Data-Snooping. Os resultados evidenciam, portanto, a inviabilidade dos agentes em auferir rendimentos anormais a partir de divergências entre os valores da cota do ETF e seu respectivo índice, o que indica uma eficiência nas precificações das cotas do fundo iShare Ibovespa.

Outlier separability analysis with a multiple alternative hypotheses test

Outlier separability analysis is a fundamental component of modern geodetic measurement analysis, positioning, navigation, and many other applications. The current theory of outlier separability is based on using two alternative hypotheses—an assumption that may not necessarily be valid. In this paper, the current theory of outlier separability is statistically analysed and then extended to the general case, where there are multiple alternative hypotheses. Taking into consideration the complexity of the critical region and the probability density function of the outlier test, the bounds of the associated statistical decision probabilities are then developed. With this theory, the probabilities of committing type I, II, and III errors can be controlled so that the probability of successful identification of an outlier can be guaranteed when performing data snooping. The theoretical findings are then demonstrated using a simulated GPS point positioning example. Detailed analysis shows that the larger the correlation coefficient, between the outlier statistics, the smaller the probability of committing a type II error and the greater the probability of committing a type III error. When the correlation coefficient is greater than 0.8, there is a far greater chance of committing a type III error than committing a type II error. In addition, to guarantee successful identification of an outlier with a set probability, the minimal detectable size of the outlier (often called the Minimal Detectable Bias or MDB) should dramatically increase with the correlation coefficient.

Building Technical Trading System with Genetic Programming: A New Method to Test the Efficiency of Chinese Stock Markets

Testing whether technical trading rules can beat buy-and-hold strategy is a common approach to study the efficiency of stock markets. Noticing that the common approach of evaluating popular technical trading rules' profitability would result in the biases of data snooping and incomplete test, we build a technical trading system with genetic programming to test the efficiency of Chinese stock markets. This system takes historical prices and volumes as inputs, randomly generates treelike structured technical trading rules composed of basic functions, and optimizes the rules using genetic programming according to the inputs. Using daily prices and volumes of Shenzhen Stock Exchange 100 index from January 2, 2004 to March 12, 2010, we find out that the optimal technical trading rules generated by our technical trading system have statistically significant out-of-sample excess returns compared with buy-and-hold strategy considering realistic transaction costs. Therefore, we conclude that Chinese stock markets have not achieved weak-form efficiency.

A Timing Channel Spyware for the CSMA/CA Protocol

This paper presents the design and implementation of spyware communication circuits built into the widely used carrier sense multiple access with collision avoidance (CSMA/CA) protocol. The spyware components are embedded within the sequential and combinational communication circuit structure during synthesis, rendering the distinction or dissociation of the spyware from the original circuit impossible. We take advantage of the timing channel resulting from transmission of packets to implement a new practical coding scheme that covertly transfers the spied data. Our codes are robust against the CSMA/CA's random retransmission time for collision avoidance and in fact take advantage of it to disguise the covert communication. The data snooping may be sporadically triggered, either externally or internally. The occasional trigger and the real-time traffic's variability make the spyware timing covert channel detection a challenge. The spyware is implemented and tested on a widely used open-source wireless CSMA/CA radio platform. We identify the following performance metrics and evaluate them on our architecture: 1) efficiency of implementation of the encoder; 2) robustness of the communication scheme to heterogeneous CSMA/CA effects; and 3) difficulty of covert channel detection. We evaluate criterion 1) completely theoretically. Criterion 2) is evaluated by simulating a wireless CSMA/CA architecture and testing the robustness of the decoder in different heterogeneous wireless conditions. Criterion 3) is confirmed experimentally using the state-of-the-art covert timing channel detection methods.

Is momentum a self-fulfilling prophecy?

The author shows in a simple framework that momentum trading can exist in equilibrium and that momentum trading is profitable. A property of the model is that the relation between risk, reward, and the intensity of momentum trading provides a natural limit to the amount of momentum trading that will exist in equilibrium. Properties of the model fit the empirics well. First, the model captures in a parsimonious manner both short-term overreaction and long-term reversals. Second, it predicts that momentum and long-term reversals should be observed in any market where there is noise. Thus, the model gives theoretical support to the empirical evidence that these anomalies are not artifacts of data snooping and to the extant empirical evidence that these anomalies are pervasive. Momentum traders observe noise shocks and trade on it as information. This trading incorporates a predictive role to the noise. That is, if agents believe a past price change to be informative of future price changes and act on this belief, it will be true and trading on this belief will be profitable. Thus, momentum trading is a self-fulfilling action.

Noise, Beliefs, and Momentum

We show in a simple framework that momentum trading can exist in equilibrium and momentum trading is profitable. Properties of the model fit the empirics well. First, the model captures in a parsimonious manner both short-term overreaction and long-term reversals. Second, it predicts that momentum and long-term reversals should be observed in any market where there is noise. Thus, the model gives theoretical support to the empirical evidence that these anomalies are not artifacts of data snooping and to the extant empirical evidence that these anomalies are pervasive. Momentum traders observe noise shocks and trade on it as information. This trading incorporates a predictive role to the noise. That is, if agents believe a past price change to be informative of future price change and act on this belief, it will be true and trading on this belief will be profitable. Thus, momentum trading is a self-fulfilling action.

Are Monthly Seasonals Real? A Three Century Perspective

Over 300 years of UK stock returns reveal that well-known monthly seasonals are sample specific. For instance, the January effect only emerges around 1830. Most months have had their 50 years of fame, showing the importance of long time series to safeguard against sample selection bias, noise, and data snooping. The overall conclusion is that monthly seasonals might simply be in the eye of the beholder. Copyright 2013, Oxford University Press.

Testing linear factor models on individual stocks using the average F-test

In this paper, we propose the average F-statistic for testing linear asset pricing models. The average pricing error, captured in the statistic, is of more interest than the ex post maximum pricing error of the multivariate F-statistic that is associated with extreme long and short positions and excessively sensitive to small perturbations in the estimates of asset means and covariances. The average F-test can be applied to thousands of individual stocks and thus is free from the information loss or the data-snooping biases from grouping. This test is robust to ellipticity, and more importantly, our simulation and bootstrapping results show that the power of the average F-test continues to increase as the number of stocks increases. Empirical tests using individual stocks from 1967 to 2006 demonstrate that the popular four-factor model (i.e. Fama–French three factors and momentum) is rejected in two sub-periods from 1967 to 1971 and from 1982 to 1986.

The Intraday Performance of Market Timing Strategies and Trading Systems Based on Japanese Candlesticks

We develop market timing strategies and trading systems to test the intraday predictive power of Japanese candlesticks at the 5-minute interval on the 30 constituents of the DJIA index. Out of 83 Japanese candlestick rules, around a third outperforms the buy-and-hold strategy at the conservative Bonferroni level. After trading costs, just a few rules remain significant however. We also correct for data snooping by applying the SSPA test on double-or-out market timing strategies. No single candlestick rule beats the buy-and-hold strategy when trading costs are taken into account. Finally, we design fully automated trading systems by combining the best performing market timing rules. Out of the 24,232 rules and systems tested on average per stock, no evidence of outperformance is found after transaction costs. Although Japanese candlesticks can somewhat predict intraday returns, we show that such predictive power is not useful for active portfolio management. When luck, risk, and trading costs are correctly measured, we find that intraday trading activity on large US caps is not sufficiently inefficient for the buy-and-hold strategy to be beaten by Japanese candlestick trading rules.

CO-REGISTRATION BETWEEN MULTISOURCE REMOTE-SENSING IMAGES

Abstract. Image registration is essential for geospatial information systems analysis, which usually involves integrating multitemporal and multispectral datasets from remote optical and radar sensors. An algorithm that deals with feature extraction, keypoint matching, outlier detection and image warping is experimented in this study. The methods currently available in the literature rely on techniques, such as the scale-invariant feature transform, between-edge cost minimization, normalized cross correlation, leasts-quares image matching, random sample consensus, iterated data snooping and thin-plate splines. Their basics are highlighted and encoded into a computer program. The test images are excerpts from digital files created by the multispectral SPOT-5 and Formosat-2 sensors, and by the panchromatic IKONOS and QuickBird sensors. Suburban areas, housing rooftops, the countryside and hilly plantations are studied. The co-registered images are displayed with block subimages in a criss-cross pattern. Besides the imagery, the registration accuracy is expressed by the root mean square error. Toward the end, this paper also includes a few opinions on issues that are believed to hinder a correct correspondence between diverse images.

Technical trading revisited: False discoveries, persistence tests, and transaction costs

We revisit the apparent historical success of technical trading rules on daily prices of the Dow Jones Industrial Average index from 1897 to 2011, and we use the false discovery rate (FDR) as a new approach to data snooping. The advantage of the FDR over existing methods is that it selects more outperforming rules, which allows diversifying against model uncertainty. Persistence tests show that, even with the more powerful FDR technique, an investor would never have been able to select ex ante the future best-performing rules. Moreover, even in-sample, the performance is completely offset by the introduction of low transaction costs. Overall, our results seriously call into question the economic value of technical trading rules that has been reported for early periods.

Registration of Multisource Satellite Images by Thin-plate Splines with Highly Reliable Conjugate Points

Image registration is essential for geospatial information systems analysis, which usually involves integrating multi-temporal and multispectral datasets from remote optical and radar sensors. An algorithm that deals with feature extraction, keypoint matching, outlier detection, and image warping is experimented in this study. The methods currently available in the literature rely on techniques, such as the scale-invariant feature transform, between-edge cost minimization, normalized cross correlation, least-squares image matching, random sample consensus, iterated data snooping, and thin-plate splines. Their basics are highlighted and encoded into a computer program. The test images are excerpts from digital files created by the multispectral SPOT-5 and Formosat-2 sensors, side-looking ERS-2, and Envisat synthetic aperture radars, and by the panchromatic Ikonos and QuickBird sensors. Suburban areas, housing rooftops, the countryside and hilly plantations are studied. The co-registered images are displayed with block sub-images in a criss-cross pattern. Besides the imagery, the registration accuracy is expressed by the root mean square error. Toward the end, this paper also includes a few opinions on issues that are believed to hinder a correct correspondence between diverse images.

Weighted total least squares formulated by standard least squares theory

Weighted total least squares formulated by standard least squares theoryThis contribution presents a simple, attractive, and flexible formulation for the weighted total least squares (WTLS) problem. It is simple because it is based on the well-known standard least squares theory; it is attractive because it allows one to directly use the existing body of knowledge of the least squares theory; and it is flexible because it can be used to a broad field of applications in the error-invariable (EIV) models. Two empirical examples using real and simulated data are presented. The first example, a linear regression model, takes the covariance matrix of the coefficient matrix asQA=Qn⊗Qm, while the second example, a 2-D affine transformation, takes a general structure of the covariance matrixQA.The estimates for the unknown parameters along with their standard deviations of the estimates are obtained for the two examples. The results are shown to be identical to those obtained based on thenonlinearGauss-Helmert model (GHM). We aim to have an impartial evaluation of WTLS and GHM. We further explore the high potential capability of the presented formulation. One can simply obtain the covariance matrix of the WTLS estimates. In addition, one can generalize the orthogonal projectors of the standard least squares from which estimates for the residuals and observations (along with their covariance matrix), and the variance of the unit weight can directly be derived. Also, the constrained WTLS, variance component estimation for an EIV model, and the theory of reliability and data snooping can easily be established, which are in progress for future publications.

Automatic Georeferencing of Aerial Images Using Stereo High-Resolution Satellite Images

For airborne surveys, the GPS/INS system has become the primary source for aerial image georeferencing. However, alternative automated georeferencing is required to serve as a backup for georeferencing when GPS/INS-based georeferencing is not feasible. High-resolution satellite images (HRSI) have been globally available with better spatial resolution and increasing positional accuracy. Therefore, HRSI has high potential as a ground control source for aerial image georeferencing which usually requires accurate 3D ground control points. Unfortunately, single imagery, which is often used as a reference, contains relief displacement due to objects on the ground introducing positional errors, unless it is not true ortho-rectified which is costly and time consuming. Therefore, in this study, a stereo HRSI-based automated georeferencing approach is proposed. The use of stereo images can avoid the impact of relief displacement and requires no external height information. The proposed method is based on a multi-scale image matching approach utilizing a combination of SIFT (Scale-Invariant Feature Transform) and RANSAC (RANdom SAmple Consensus). In the georeferencing step, the bundle adjustment with outlier removal of Baarda's data snooping was utilized. Experimental results for a strip of aerial images with stereo Ikonos images showed its potential as a backup system for automated georeferencing.

Contrarian Flows, Consumption and Expected Stock Returns

We investigate the relation between contrarian flows, consumption growth and market risk premium. We construct a contrarian flows measure by summing up the capital flows to stocks that go against the total flow of the aggregate market. We show that the contrarian flows are negatively influenced by the same quarter consumption growth. Because during bad times majority of investors who are affected by the negative shock reduce their equity exposure and these extra supply of risky assets are absorbed by those contrarian investors who are least affected by the consumption shock. Using quarterly stock market data, we find that the contrarian flows forecast market returns at short to intermediate horizons. Almost all of the predictability is explained by the component of contrarian flows that is determined by consumption growth. Moreover, the predictability is stronger for growth stocks than for value stocks and hence negatively predicts the value premium. This is because the contrarian flows measure the market risk premium and growth stocks bear more discount rate risk than value stocks. Out-of-sample tests show that the main results are robust to data-snooping bias.