statistical arbitrage pairs trading strategies

• plos.org
• create account
• sign in

• Loading metrics

Artless Access

Peer-reviewed

Research Clause

Optimal statistical arbitrage trading of Berkshire Hathaway stock and its replicating portfolio

• Che-Ming Yang

Optimal applied math arbitrage trading of Berkshire Hathaway stock and its replicating portfolio

• An-Sing Chen,
• Che-Ming Yang

x

• Published: January 15, 2022
• https://doi.org/10.1371/diary.pone.0244541

Figures

Pinch

In this paper, we work use of the replicating plus for applied mathematics arbitrage trading, where the replicating plus is constructed aside a portfolio that mimics the returns from a divisor model. Using the replicating asset in the linguistic context of statistical arbitrage has never been through with before in the literature. A refreshing optimal statistical arbitrage trading manakin is practical, and we derive the average transaction length and return for the Berkshire A stock and its replicating asset. The results show that the statistical arbitrage method proposed away Bertram (2010) is profitable by victimisation the replicating asset. We also compute the average returns under different transaction costs. For the statistical arbitrage using the replicating asset of the factor exemplar, mediocre time period returns were at least 33%. Hardiness is examined with the Sdanamp;P500. Our results can provide circumvent fund managers with a new technique for conducting statistical arbitrage.

Citation: Chen A-S, Yang C-M (2021) Optimum statistical arbitrage trading of Berkshire Hathaway stock and its replicating portfolio. PLoS ONE 16(1): e0244541. https://doi.org/10.1371/journal.pone.0244541

Editor: Alejandro Raul Hernandez Montoya, Universidad Veracruzana, MEXICO

Received: Jan 28, 2022; Accepted: Dec 13, 2022; Publicized: Jan 15, 2022

Copyright: © 2022 Chen, Yang. This is an open get at clause distributed low the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The DOI/admittance act to access code the data for this study is: DOI 10.17605/OSF.Io/ZM27G.

Funding: A.S. Chen standard funding for this work from Taiwan's Ministry of Science and Technology, Grant number To the highest degree 105-2410-H-194-044-MY2. The funders had no role in study intention, data collection and analysis, determination to publish, or preparation of the holograph.

Competing interests: The authors have got declared that No competing interests exist.

1. Introduction

Pairs trading is a statistical arbitrage concept, and information technology has two types: one is the applied math arbitrage, and the other is jeopardy arbitrage. Information technology usually involves using two related securities in the same industry or with synonymous characteristics and is entered into when their prices abnormal from the equipoise state. The application of pairs trading can be roughly divided into terzetto main steps. First, look for two securities whose prices have the comparable trend in a given period. Indorse, observe the change in the spread between the two securities during the subsequent trading period. Third, if the 2 securities feature an equilibrium family relationship, a long (short) neutral portfolio can be constructed. When the scatter of the two securities reverts to its historical miserly, the put back is reversed. Existing inquiry finds that pairs trading is correlative to concepts, much as cointegration, correlation of stock prices, mean statistical regression, overreaction, contrarian selection, and price trends [1–7]. It is undeniable that the effectiveness of pairs trading mainly depends on the modeling and prediction of the spread time series of the two assets (securities, index, Beaver State commodities).

If investors want to get the maximum profits, they should look for the spread of pairs with high variance and strong mean reversion. The usual method acting is to construct a stationary, mean-returning synthetic asset as a linear combination of securities [8]. Notwithstandin, some researchers argue that this method has little help for forecasting [9, 10]. In light of this, the applied mathematics arbitrage method is developed. For example, pairs trading and its generalizations depend on the expression of mean-reversion spreads, but the mean-lapse spreads essential have a certain stage of predictability. So, researchers have developed loss protection methods for pairs trading [11] or used algorithms to estimate trade duration and find optimal preset boundaries [12]. Pairs trading is a simple concept. If the two blood-related stocks' circularize expands and diverges from chemical equilibrium, investors buns short the high priced stock and buy the low price stock; the investor wish profit when the spread converges back towards equilibrium. The concept of pairs trading potty be practical to any equilibrium relationship of financial markets or to a portfolio of securities some held poor, and the others held long. It uses statistical methods to identify two related stocks and then exploits the potential short condition congenator mispricing 'tween them and finds all possible portfolios.

In applied mathematics arbitrage trading, investors calculate the historical distance between the standardized daily recall paths and choose the pair with the smallest trading distance. If the future price is similar to the past, the paste terms may meet again, resulting in a positive return in the nought-cost portfolio. Investors can maximize profits by shorting the overpriced and purchasing the underpriced. However, this method acting still has some problems, such equally when to deal out to maximize the profit of opposite trading. Bertram [13] uses the statistical arbitrage trading supported (the Ornstein Uhlenbeck process) to drive the timing of pairs trading entranceway and exits. Cummins and Bucca [14] followed Bertram's method acting and achieved good results. Do and Faff [9, 10] try the impact of trading costs on pairs trading profitability. For statistical arbitrage, issues such as when, how, and the impact of dealings costs are important.

The relationship between chance and return has always been a unreassuring matter in academe and application. Fama and French's three-constituent model is designed to capture the carnal knowledg between median counte and size and the price ratios [15]. The three-broker model significantly improved CAPM because IT adjusted for the outperformance tendency of strategies supported on the additional factors. Although the three-broker model canful explicate most of the sprout returns, there are still researchers WHO think that information technology is non right-down [16–18]. Carhart [16] added the impulse factor and proposed a four-factor simulate. Titman et al. [17] argue that increasing capital investments subsequently leads to negative benchmark-adjusted returns. Novy-Marx [18] measures profitableness using the gross profits-to-assets and shows that it provides more or less the assonant power equally reserve-to-grocery in foretelling the crosswise of average returns. The five-factor model added the factors of gainfulness and investment; the evidence showed that the trio-factor model was lean for explaining expect return because IT ignored a lot of the variation in modal returns related to profitability and investing [19]. The trial-and-error tests of the fivesome-factor model consumption the average returns on portfolios guitar-shaped connected size, B/M (book to grocery store ratio), profitability, and investment. Recently, newer gene models, such American Samoa BAB and QMJ, are proposed [20, 21]. Frazzini et aluminum. [22] studied Warren Buffett's company, Berkshire Hathaway's profitability, and carrying out; they propose several quantifiable factors to analyze its performance. They refer to the papers by Carhart [16], Frazzini and Pedersen [20], and Asness et al. [21] to propose an alternative factor model. In this study, we will refer to that As the "Buffet-factor model." We present analytic formulae and solutions for calculating optimal applied mathematics arbitrage strategies with transaction costs using the Buff-factor model to form the replicating portfolio. We bear that the synthetic plus formed aside the Berkshire Hathaway old-hat and its replicating portfolio can be described by the Ornstein Uhlenbeck process. Our main results show that the replicating portfolio can be effectively matched with the innovative asset in a pairs trading applied mathematics arbitrage framework and verify that this method is rewarded.

In past tense literature, most enquiry explores the arbitrage of two similar nature assets. There does not exist research along victimization the factor model to create replicating assets for arbitrage. The important contribution of this paper is using the component model to form a replicating asset and then constructing the synthetic plus with other assets for statistical arbitrage. Methodologically, we form the replicating asset (portfolio) by using the Buffett- and basketball team-factor simulate following the method described in Asness et al. [21]. We verify that this method can indeed create replicating assets that exhibit similar properties to Berkshire A stock and that the replicating asset can be paired with the original Berkshire A stock for statistical arbitrage, profitably. To implement the statistical arbitrage, we refer to the findings of Bertram [13] and apply them to our experiments. In particular, we utilise the Counter-, five-factor mock up, and the Ornstein Uhlenbeck process to perform statistical arbitrage for the Berkshire A stock and the Sdanamp;P 500 portfolio. The sequent analysis provides the mathematical framework which can buoy be wont to explore the relationships between the replicating portfolio and Berkshire's stock and go insight into the kinetics of trading strategies. We use the Ornstein Uhlenbeck process to build a continuous trading strategy for the original asset and its replicating portfolio and cipher the sell distance and the return of the strategy based on the transit time of the process. The results of this paper show that Berkshire A paired with its replicating portfolio provides returns of at least 33% under applied mathematics arbitrage and Sdanadenosine monophosphate;P500 at least 4.8%.

The remainder of the theme is methodical as follows. Department 2 discusses the literature related to statistical arbitrage and factor models. Section 3 presents the data and statistical arbitrage trading pattern of the Ornstein Uhlenbeck process. Section 4 presents the results of the verifiable depth psychology and examines robustness to variable transaction costs. Section 5 concludes and summarizes the main results of the paper.

2. Related literature

Literature on statistical arbitrage

The most plebeian form of pairs trading involves forming a portfolio of 2 related stocks whose relative pricing departs from its 'labyrinthine sense.' The equilibrium whitethorn involve the concepts of cointegration, mean regression, overreaction, and reversal strategies. The success of statistical arbitrage depends on determination two suitable securities, then moulding and forecasting of spread time series. Ace popular method is the distance approach, which explores different dimensions and implications of pairs trading strategies, such As accounting information, news, liquidity, sensitivity, dealing costs, etc. [8–10, 23–25].

Jegadeesh and Titman [5] find the strategies that buy up stocks that have outperformed in the past, and curt stocks that have underperformed in the past produce significantly positive returns. They argue that the profitability of these strategies is not owing to their systematic hazard or delayed stock price reactions to common factors. Gatev et al. [8] freshman proposed the distance approach. They use all U.S. stocks from the CRSP daily dataset from 1962 to 2002. They design a easy algorithmic program for choosing pairs and test the gainfulness of several straightforward, self-financing trading rules. They cypher the sum of Euclidean squared distance (SSD) for n stocks and choose the smallest SSD to concept a portfolio. They then choose to enter the transaction when the price of the paired plus is greater than two standard deviations, and when the price competitory the average price, they sell the paired plus. They get average annualized superfluous returns of most 11% for the pinnacle pairs portfolios and that the profits come non look to be caused by simple mean lapse. However, Do and Faff [9] apply the Gatev et AL. [8] methodology with more modern data and find the profit show a declining trend when the naive trading dominion is used. Do and Faff [10] change the 2010 paper on the U.S. CRSP stemm but run the sample period from 1963 to 2009. In the newer study, they considered transaction costs and allowed securities to be matched across 48 Fama-European nation industries. This restriction created more meaningful pairs, because all sample companies are now matched within the same sectors. Their results record that the pairs trading strategy clay profitable, albeit at much more coy levels. Cohen and Frazzini [26] find substantial customer-provider links in the U.S. stock commercialise that allow return predictability in the context of this strategy.

Another statistical arbitrage concept is cointegration. The method acting of cointegration describes as follows: first, choose two cointegrated neckcloth price series; second, construct a long (short) position when stocks deviate from their long-run equilibrium; at last, close the position after convergence or at the end of the trading period [27–30]. Hong and Susmel [31] implement the cointegration approach to common stocks. They analyze pairs-trading strategies for 64 Asian shares recorded in their local markets and recorded in the US as ADRs from 1991 to 2000. They assume these pairs to be cointegrated but provide no test results in their paper. Vidyamurthy [29] uses the Arbitrage Pricing Theory (APT) to identify stocks with akin democratic come back components. They developed a framing for forecasting using the cointegration method and analysis the mean reversion of the residuals. Under the variable cointegration approach, Dunis and Ho [32] use cointegration relationships to construct exponent-tracking portfolios for the EuroStoxx 50 index. They and so take contrary subsets of the forefinger constituents and calculate the joint cointegration transmitter for these constituents and the EuroStoxx 50 index. They find that the tracking baskets have an outperformance versus the bench mark in terms of complete returns and Sharpe ratio. Caldeira and Moura [33] utilize the univariate cointegration approach to the 50 to the highest degree melted stocks of the Brazilian stock certificate index Bovespa. They show that this strategy generates excess returns of 16.38% per year. Moreover, it also gives a Sharpe Ratio of 1.34 and a low correlation with the market.

Mean reversion is key to pairs trading. The stochastic spread method acting models the average-lapsing march of pairs trading as an Ornstein Uhlenbeck process. Elliott et al. [34] advance a ignoble-reverting Gaussian Andrei Markov chain sit for the spreadhead, which is observed in Gaussian noise. They comparison the model with later observations of the pass aroun to find proper investment decisions. They conceive this approach path stern be practical to any financial market to attain wealth, flat though information technology is at times out of equilibrium. Most financial info is non-Stationary and not-Gaussian. Hence, it is necessary to deal with the stability of the time series information. The Ornstein Uhlenbeck litigate is a stationary Gauss Andrei Markov process, and is homogeneous one of these days. This process has a long-run-term mean and exhibit mean reversal. The process can be viewed a modification of random walk in continuous clock or Wiener process. The Ornstein Uhlenbeck process can be considered as the endless-time analog of the AR (1) process. Because the Ornstein Uhlenbeck process is static, the return is settled.

Bertram [13] derives the entry/passing time and analytical formula for the trading thresholds for synthetic assets nutmeg-shaped aside pairs, whose price assumptions follow the Ornstein Uhlenbeck process. He showed that the optimal thresholds were symmetric around the mean both for maximizing the return per unit time and the Sharpe ratio. His result also provides the optimal entry and exit points for arbitrage trading at a conferred dealing cost. Cummins and Bucca [14] believe a mental investor would aim for a high-profit opportunity. So, the demythologised pairs trade in should have a compounding of the lowest drift in spread mean and highest spread variance features. They practice Bertram's method to the spread trading for crude oil and refined products markets. Their result showed evidence of aggregate upward and downward mean reversion, and profitable strategies with Sharpe ratios of greater than two. Their research shows that Bertram's method acting has lucrativeness voltage for non-Gaussian processes. Avellaneda and Lee [35] and D'Aspremont [36] mature a method acting of synthetic mean-relapsing portfolios, which uses the concept of Elliott et alia. [34]. They use orthodox correlation analytic thinking to reconstruct mean-reversion portfolios with a limited number of assets. Their result shows that the systematic component of store returns explains between 40 and 60% of the variableness. Huck and Afawubo [37] research the performance of a pairs trading system supported various pairs-survival of the fittest methods. They exercise the components of the Sdanamp;P 500 index finger arsenic an observation target. They argue that when the stock price deviates from equilibrium, the investor can enter the merchandise (recollective or short) after dominant for risk and transaction costs. They also show that the distance method acting generates insignificant excess returns, but the cointegration method provides a high, stable, and iron payof.

Literature on the Phoebe-ingredien and Buffett-factor pose

Below the capital asset pricing mould (CAPM), the expected return of a breed is related to its Beta coefficient. All the same, this model has difficulties in explaining actual stock commercialize returns. First, its assumptions are nasty to maintain in the real capital marketplace. Second, the factors touching expected returns may be more than one. The Arbitrage Pricing Theory (APT) is an extension of CAPM. The pricing manikin given by both the APT and the CAPM are models low equilibrium. The difference is that the APT theoretical account is based on the concept of arbitrage. If the market does not reach chemical equilibrium, there will be run a risk-at large arbitrage opportunities in the market. The APT model uses several factors to explicate the expected return on risky assets, and according to the atomic number 102-arbitrage principle, there is an approximately linear family relationship between the equilibrium return of risk assets and the risk factors.

Fama and European country [15] feel that companies with smaller market esteem and higher book value/ market value ratio are more possible to achieve an fair plac of return above market levels. So, they join the sizing factor (SMB) and value factor (HML) with the original CAPM. The resulting sit is the familiar triad-factor modeling. This model can explain about 90% of stock returns. Carhart [16] uses a four-factor model, which includes the market divisor, size gene, value factor, and impulse broker, to mastery the encroachment of systemic risk connected stocks. His results read that the quaternary-factor model is an advance concluded the Fama-Gallic's three-factor model and was able to explain the short‐term persistence in equity bilateral investment firm returns. Titman et al.. [17] find the firms that well increment capital investments after achieve negative benchmark-adjusted returns. The negative deviant capital investment/return relation is shown to be stronger for firms that have more considerable investment discernment, i.e., firms with higher cash flows and lower debt ratios. Novy-Zeppo [18] identifies a proxy for expected profitability that is strongly related to the middling return. His results show that gross net-to-assets has the unvaried power as book-to-market in predicting the cross-section of average returns. Also, profitable firms produce significantly higher returns than unprofitable firms, despite having importantly higher valuation ratios. Fama and daniel Chester French [19] added the factors of profitability and investment to the three-factors model and showed that the v-factor model explains between 71% and 94% of the cross-section variance of expected returns on the portfolios.

Asness et al. [21] define 'quality' security as one that has characteristics that an investor is willing to pay a higher price for, namely, stocks that are safe, profitable, growing, and well managed. They present a upper-class-minus-junk (QMJ) factor to describe these characteristics, and they find that stocks sorted on this cistron have fundamental lay on the line-adjusted returns in the U.S., and globally crossways 24 countries. The QMJ factor is constructed by a portfolio that longs high-quality stocks and drawers low-quality stocks. Frazzini and Pedersen [20] hash out a grocery store-neutralized betting-against-beta (BAB) factor, which is constructed by a portfolio that longs leveraged low-Beta assets and shorts high-beta assets. They find two key results: (1) high beta is associated with low alpha (2) stocks sorted on the BAB factor produces significant positive adventure-adjusted returns.

3. Data and methodology

Therein paper, we examine the statistical arbitrage trading between the Berkshire Hathaway stock and its replicating portfolio. For clearness of exposition, in the remainder of this report the terminus "replicating portfolio" will be replaced with the term "replicating asset" where appropriate for increased clarity. The empirical tests of this study consist of 2 parts. First, we utilize the five-factor model and Buffett-factor example to construct the replicating plus. Second, we use the Ornstein Uhlenbeck process to test the applied mathematics arbitrage trading of the synthetic asset constructed from the original Berkshire Hathaway stock and its replicating asset.

For this study, we utilize the factors data of the five-factor and the Buffett-factor out model and the price of Berkshire Hathaway stock. The various factors of the models are obtained from the Kenneth R. French data program library and the AQR data library. Berkshire Hathaway's neckcloth return and price data are from the CRSP database. Berkshire Hathaway's stock has two classes: class A and class B. Berkshire Hathaway introduced parcel class B in April 1996. For this study, concenter on Class A stocks supported the longer length of the data. We use time unit information from 1980/3/17 to 2022/09/28. We use the five-factor pose, and the Buffett-factor model to construct the replicating asset for Berkshire Hathaway class A stock. We denote information technology with BerkA*. We then use the replicating asset and the original Berkshire Hathaway stock to conduct the statistical arbitrage experiments. The information period used in this study spans from 1980/03/17 to 2022/09/28, which gives a total of 9720 daily observations.

Fama and French [15] design the three-factor model to conquer the relation 'tween average return and Size (market capitalization, computed atomic number 3 the price times shares superior) and the relation between average return and monetary value ratios like B/M. Still, Titman et al.. [17] and Novy-Harpo [18] Federal Reserve note that the tierce-agent modelling misses untold of the variation in average returns associated with profitability and investment. The three-agent model is incomplete. Fama and French [19] add profitability and investment factors to the three-cistron manikin. The five-factor model is presented as follows: (1)

A cinque-factor framework that captures the size of it, value, profitability, and investment patterns in average standard returns can perform better than the three-factor model. In the five-factor model, R _it is the give back on security or portfolio i for period t, R _ft is the risk-release return, R _{element 109} is the return on the value-weight (VW) food market portfolio. The size gene SMB _t is the return from a diversified portfolio of small stocks little the fall along a diversified portfolio of colossal stocks. A positive loading on SMB reflects a tendency to buy small stocks [15], but Berkshire's negative loading reflects a tendency to buy large stocks [22]. The value factor HML _t is the difference between the returns on diversified portfolios of high gear and low B/M stocks. Berkshire farm animal shows a positive tilt, which reflects a tendency of buying stocks that possess a high book value comparative to their food market value [22]. The profitableness factor RMW _t is the deviation between the returns on varied portfolios of stocks with robust and weak profitability. The investment factor CMA _t is the difference between the returns on diversified portfolios of the stocks of low and squeaking investment firms, and e _it is a zero-mean residual.

Buffett's factor model was proposed by Frazzini et al. [22]. They express that Berkshire Hathaway's performance hindquarters comprise explained mainly by exposures to appreciate, ground-hugging-risk, and quality factors. It has been certificated that value stocks have high average returns than growth stocks [38, 39], and high-quality stocks outperform than junk stocks on the average [21]. Among the traded U.S. stocks, Berkshire Hathaway has one of the highest Sharpe ratios for more than 30 years. Frazzini et al. [22] identify single features of Buffett's portfolio: "safe" (humble volatility and David Low genus Beta), "cheap" (i.e., stocks with low price-to-book ratios (value stocks), and piercing-quality (stocks that are profitable, lasting, development, and have soaring payout ratios). Notwithstandin, the stocks exhibit these characteristics usually perform fountainhead, not only the buy in that Buffett bought. These cured-renowned factors seem not in full capture the performance of Berkshire stock.

Carhart [16] constructed a four-factor model by combining the deuce-ac-factor manikin of Fama and French [15] and the momentum personal effects of Jegadeesh and Titman [5]. The impulse factor UMD is constructed aside buying the stocks that surpass the market and shorting the stocks that underperform. UMD is superficial to Berkshire stock. Frazzini et al. [22] expanded the four-divisor model and added the equity constituent (Betting Against Beta (BAB)) of Frazzini and Pedersen [20] likewise as the character factor (Quality Negative Trash (QMJ)) of Asness et al.. [21]. The BAB factor is constructed by a portfolio that longs low-beta assets, leveraged to a beta of one: and shorts high-beta assets, de-leveraged to a Beta of unitary. The QMJ cistron is constructed past a portfolio that longs high-quality stocks and shorts under-tone stocks. The Buffett-factor model of Frazzini et al. [22] as follows: (2) We modify the equation of Carhart's [16] and add the factors of BAB and QMJ and let its features be seamless with the five-factor model. The equation is specific American Samoa follows: (3)

This section provides the mathematical exposition of the Ornstein Uhlenbeck process statistical arbitrage trading manikin of Bertram [13]. Under this model, a continuous trading strategy is formed by a series of separate proceedings executed happening a consecutive-fourth dimension stochastic action. Thus, umteen trading strategies are a role of the frequency which these transactions occur. The trading frequency is delimited by how many times the scheme trades per unit time. So, we can mould the price of the listed security P _t as, (4) X _t denotes the synthetic plus. For our experiments we define X _t = log(Berkshire stock price)−log(replicating asset price). Since the replicating plus is constructed using the theoretically accurate asset pricing models of Eqs (2) and (3); X _t , the spread 'tween two asset log price series should follow a zilch-mean Ornstein Uhlenbeck process. An Ornstein Uhlenbeck process is cruciform. Mathematically, X _t lavatory be written equally the tailing stochastic differential equality, (5) where α and η dangt; 0, and W _t is a Wiener process. Suppose T is the complete trading period. Define the entry levels of the trading scheme by X _t = a and exiting the barter at X _t = m. We assume that a danlt; m, this implies that a danlt; 0 and m = −a. A complete trading oscillation time is presented as follows: (6) Let r(a,m,c) denote the replication per trade As a function of a,m, and transaction cost. Given that the X _t represents the log price, the continuously compound rate of return for a single trade after accounting for transaction cost can be written atomic number 3 r(a,m,c) = (m−a−c). In a profitable scheme, the return must pass the dealings costs from entry moving to exit. Bertram [13] shows that subordinate these conditions, the expected value per unit clock and the variance of the riposte per unit of measurement time for the strategy arse be written arsenic, (7) (8) where E(T) and V(T) are the mean and variableness of T. Even though the process X _t is stationary, the go back of all swop is stochastic; the sentence frame for return realization is random. A trade may take a age ahead reaching the die level and have a significant deviation off from the exit level during the time frame. We can use the first pass time of the Ornstein Uhlenbeck process to figure out the expression of the strategy return and variance. Bertram [13] assumes adanlt;m, adanlt;0, and −a = m. T _a→m represents the prison term to transition from a to m, and T _m→a is the time to transition from m to a, and the independency of the two times follows from the Markovian property of the Ornstein Uhlenbeck process. The maximum expected return is then given A follows: (9) and the optimal value of a will fill the followers equation: (10)

4. Results

This section presents the results of optimal applied mathematics arbitrage trading of Berkshire Anne Hathaway blood line with its replicating asset. First, we construct a replicating asset, which bequeath have similar peril and return characteristics with the actual Berkshire A stock price by using the pentad-factor model (Eq (1)) and the Buffett-factor mannequin (Combining weight (3)). The factor loadings of the replicating asset are estimated away regressing the unnecessary return of Berkshire A on right-hand-side factors of Eq (1) and Combining weight (3). Specifically the surplusage return of Berkshire A is the as the dependent variable and the factors on the right-hand-side of Eq (1) and Equivalent (3) are the independent variables. The estimated coefficients are then used as the portfolio weights for the construction of the replicating plus. The returns of the replicating portfolio will, in the long run for, match the returns of the Berkshire A stock, since the replicating portfolio is constructed from theoretically correct asset pricing model specifications. In the remainder of our paper, we will denote the replicating portfolios as simply Buffett- OR five-divisor exemplar. Second, we use the replicating portfolios as input to the pair trading pretense tests. The five- operating room Buffet-factor models capture just about of the factors affecting stock price return. Assuming the investor is coherent, the investor will review past performance and adjust the trading strategy in each fixed flow. Therefore, in our experiments, we modeling this behavior by presumptuous that investors will close their open positions (hold losses or capitalize the gains) the last trading Clarence Day of each year and reset their positions the first trading twenty-four hour period of the incoming year exploitation the information adequate that day. This reset will prevent the prices of the two paired assets (Berkshire A stock and the replicating asset) from drifting too FAR apart. Specifically, at the origin of apiece year, investors leave recalculate the theoretical value of the replicating plus and equivalence it with the market value of Berkshire A stock and take the appropriate positions for the next cycle (short the overpriced asset and long the underpriced plus).

Fig 1A and 1B exhibit the stock price trends and neckcloth toll returns of Berkshire Anne Hathaway A from 1980/03/17-2018/09/28. In Fig 1A and 1B, we can find the stock price dropped importantly, and the store price return also fluctuated importantly more than otherwise times during the fiscal crisis.

Common fig tree 1.

a: The trend of Berkshire A stock price, b: Berkshire A stock terms returns. Fig 1A shows Berkshire A stock price drift from 1980/3/17-2018/09/28. The figure shows that Berkshire A standard price rises very promptly. Information technology as wel shows that the turn back excitableness for Berkshire A is also relatively eminent. Libyan Fighting Group 1A shows that the stock price fell significantly during the period of the recent financial crisis, but the stock price besides rises faster after the business enterprise crisis. Fig 1B shows the return of Berkshire A stock terms from 1980/3/17-2018/09/28. Fig 1B shows that the volatility of stock price return was significantly high in 1987 and 2008 than in other periods.

https://Interior.org/10.1371/diary.cornpone.0244541.g001

In our experiments, we employ two different factor models to construct the replicating asset. Fig 2 shows the returns of the replicating assets constructed using the Buffett- and five-ingredien models. Libyan Fighting Group 2A and 2B shows that the gross fall pattern of the replicating assets is similar to the return pattern of the underivative asset in Fig 1B. In the remainder of this theme, we use BerkA* to denote the theoretic stock price of the replicating asset constructed by using the Buffett- or five-factor models.

Figure 2.

a: Buffett-factor posture returns, b: pentad-factor model returns, c: Buffett-factor model returns and five-factor manakin returns for the sub-period (2008–2009). Fig 2A shows the returns of the replicating plus constructed by using the Buffett-factor model. Ficus carica 2B shows the returns of the replicating asset constructed by victimisation the five-element mannikin. We can happen that the overall behavior patterns of the two replicating assets are same to those of the original Berkshire A stock. Fig 2C overlays the information presented in Fig 2A and 2B in one plot of ground using a shorter time frame, 2008–2009. This form shows that the information presented in Fig 2A and 2B, which depict the returns of the replicating assets obtained from using the Buffett- and five-factor models are not identical to each past. In the figure, the maximum return of the Buffett-element simulation is 7.2%; the minimum return is -0.068%. The maximum return of the cinque-gene theoretical account is 6.26%; the minimum rejoinder is -0.0733.

https://doi.org/10.1371/journal.pone.0244541.g002

It should Be noted that Fig 2A and 2B, which depict the returns of the replicating assets obtained from using the Buffett- and Little Phoeb-factor models are not identical to each otherwise. They solitary look similar to each other because the time-historic period for the two figures is quite lasting, spanning from 1980/3/17-2018/09/28, which causes the two figures to looking at very similar to each early. Al-Jama'a al-Islamiyyah al-Muqatilah bi-Libya 2C plots the information in Figure 2A and 2B using a shorter period, from 2008–2009 only, and overlayer the results of the deuce models on upper side of all other victimization only i graph. Al-Jama'a al-Islamiyyah al-Muqatilah bi-Libya 2C shows much clearly that the data presented in FIG 2A and 2B are not isotropous to apiece other.

Fig 3A shows the price behavior of the replicating asset constructed using the Buffet-factor model. Fig 3B shows the cost deportment of the replicating asset constructed using the cardinal-factor model. Figs 3A, 3B and 1A register nearly identical patterns and trends. The figures show that the replicating assets constructed away the Buffet- and cinque-factor simulate does an excellent job of replicating the Berkshire stock and thereby tin can serve atomic number 3 a good opposing asset in pairs trading applied mathematics arbitrage. Figure 3C plots the info in Fig 3A and 3B using a shorter period, from 2008–2009 only, and overlay the results of the two models on top of each other using only single chart. Fig 3C shows that the information presented in Fig 3A and 3B are not identical to each other.

Fig 3.

a: Replicating asset price of Buffett-factor posture, b: Replicating asset price of five-factor role model, c: Replicating asset price of Buffett-factor model and replicating asset price of five-factor model for the sub-period (2008–2009). Fig 3A shows the replicating plus price when the replicating asset is constructed using the Buff mannikin. Fig 3B shows the replicating plus price when the replicating asset is constructed using the five-constituent model. Overall price deportment closely matches the price conduct of the original Berkshire A monetary value. Fig 3C overlays the information presented in Fig 3A and 3B in one plot using a shorter time frame, 2008–2009. This form shows that the information conferred in FIG 3A and 3B, which depict replicating asset Leontyne Price when the replicating asset is constructed using the Buffet- and five-component models, are non identical to each other. The maximum apprais of replicating assets price of Buffett- factor model is $141,109; the minimum value is $73,088. The maximum value of the five-factor model is $140,284; the minimum value is $73,375.

https://Interior.org/10.1371/journal.pone.0244541.g003

We following apply the Ornstein Uhlenbeck work on arbitrage strategies discussed in Bertram [13] and Cummins and Bucca [14] to find the optimum entry and exit points for the initiation and surcease of pairs trading. We set out α = 0.006439, η = 0.05401 for Buffett-factor exemplary, and α = 0.01702, η = 0.05748 for five-ingredien model. Figs 4 and 5 stage the results of using the Buffett- and five-divisor simulate as the replicating plus. The figures present the plots of the expected return and entry-level with different transaction costs. They show that the trading frequency and dealing cost has a strong shape on the profitableness of the trading strategy. Trading strategies are influenced by "a" and transaction cost to form the expected return bands. In the trading bands, for a precondition transaction cost, the larger "a" makes the expected pass smaller. On the other hand, for a given "a" value, the larger the dealings cost makes the expected return small. In sum, the "a" and transaction costs are discriminatory to the awaited render of the trading strategy. The equations for the mean and disagreement of the return give up us to determine the trading bands that optimize the trading strategy. We use Eq (8) to get a, when adanlt;0. The expected tax return is equal to 0 when adangt;0.

Fig 4.

The relationship of expected revert, a, and c when the replicating plus is constructed by the Snack counter-factor manakin. This calculate presents the relationship between the expected return, a, and transaction be when the replicating plus is constructed by the Buffett-factor model. This example uses parameters α = 0.006439, η = 0.05401, c = 0.001~0.007.

https://doi.org/10.1371/journal.cornpone.0244541.g004

Common fig 5.

The relationship of expected return, a, and c when the replicating asset is constructed by the five-factor out model. This figure presents the relationship between the expected return, a, and transaction cost when the replicating asset is constructed by the fivesome-factor sit. This example uses parameters α = 0.01702, η = 0.05748, c = 0.001~0.007.

https://Department of the Interior.org/10.1371/diary.pone.0244541.g005

Table 1 shows the expected returns of using the optimal statistical arbitrage strategies for different transaction costs and "a." The optimum solution for "a" is from Combining weight (8). In the Table, "c" represents the dealings cost, and "a" represents the submission-level. Table 1 shows that "a" and expected payof will become smaller as the transaction costs increase, regardless of whether the replicating asset is constructed using the Buffett- Beaver State five-factor model.

Figs 6A and 7A plot the family relationship between the dealings costs and "a." Figs 6B and 7B secret plan the relationship between the transaction costs and the expected return of the optimal trading scheme. Fig 6A and 6B correspond to the results of exploitation the Buffett-factor framework to construct the replicating asset. Fig 7A and 7B are the results of victimization the five-broker model. These figures demo that higher transaction costs wish cut down some the "a" and the expected return of the optimal trading strategy, regardless of which modelling is used to construct the replicating asset. Figs 6B and 7B, however, testify that the expected return of the optimal trading strategy is not that sensitive to transaction costs.

Common fig 6.

a: c vs a for Buffet-factor model, b: c vs expected return for Snack bar-cistron example. Fig 6A shows the relationship between transaction cost and Fig 6B shows the relationship between transaction cost and the due render from the optimal trading strategy. The replicating asset for Figure 6 is constructed using the Buffett-factor model.

https://doi.org/10.1371/journal.pone.0244541.g006

Fig 7.

a: c vs a for quintet-factor model, b: c vs expected generate for five-factor pose. Libyan Islamic Fighting Group 7A shows the human relationship between transaction cost and a. FIG 7B shows the kinship 'tween dealings cost and the expected return from the optimal trading scheme. The replicating asset for Fig 7 is constructed using the quint-factor model.

https://doi.org/10.1371/daybook.pone.0244541.g007

Fig 8A and 8B show the relationship between the synthetic plus toll and the incoming level "a" and exit level "m." Fig 8A provides the results from victimisation the Buffett-divisor model to retrace the replicating asset. Fig 8B provides the result of using the five-factor model to construct the replicating plus. For Fig 8A, α = 0.006439, η = 0.05401, and the "a" and "m" are obtained assuming c = 0.001. For Fig 8B, α = 0.01702, η = 0.05748, and the "a" and "m" are obtained assumptive c = 0.001. "c" represents the transaction cost; "a" represents the entry level; "m" is the die down level.

Libyan Islamic Grou 8.

a: The behavior of the synthetic plus formed from Berkshire A stock price and the replicating asset BerkA* when the replicating asset is constructed using the Buffett-divisor exemplary, b: The behavior of the synthetic plus fan-shaped from Berkshire A stock price and the replicating asset BerkA* when the replicating asset is constructed using the five-factor model. Libyan Islamic Fighting Group 8A presents the relationship between the synthetic asset price and the entry level "a" and exit level "m." In Fig 8A, the results are computed using the Buffett-factor model to construct the replicating asset. In computation Fig 8A, α = 0.006439, η = 0.05401, and the "a" and "m" are obtained forward c = 0.001. "c" represents the dealing cost; "a" represents the entry level; "m" is the exit raze. Libyan Islamic Grou 8B presents the relationship between the synthetic plus price and the entry level "a" and exit level "m." In Fig 8B, the results are computed using the five-factor model to construct the replicating asset. In computer science Fig 8B, α = 0.01702, η = 0.05748, and the "a" and "m" are obtained assuming c = 0.001. "c" represents the transaction cost; "a" represents the entry level; "m" is the go level.

https://doi.org/10.1371/journal.pone.0244541.g008

The results presented show the feasibility of victimisation synthetic assets formed aside using the actual stock and a replicating asset constructed from ingredien models, without the need to find real assets with similar properties for statistical arbitrage. In the experiments, ii different factor models were used to replicate the original asset, Berkshire A. Applied math arbitrage applied to the polysynthetic asset oven-shaped from the primary Berkshire A and the replicating asset results in arbitrage profits of leastways 33% in potential return in Table 1. Our experiments show that the factor model can be used with success to produce replicating assets, which in turn can be combined with the unconventional asset to form a counterfeit asset that throne be used in pairs trading for optimal applied mathematics arbitrage. We express that the synthetic asset pliable from the replicating asset and the germinal Berkshire A blood can give utile ledger entry and exit points for statistical arbitrage trading at divers transaction costs.

Hardiness tests

To test for the robustness of our method acting, we redo our experiments using the Sdanamp;P500 index as the fair game asset to be replicated instead of Berkshire A inventory. We suss out to see if the proposed optimum arbitrage scheme is workable low-level the indistinguishable method. We again use the Buffett-and five-agent poser to mold the replicating asset. Estimations of parameters were done by using the day-after-day data of the Sdanamp;P 500 index. The synthetic asset is formed from the replicating asset and the actual SdanA;P 500. We calculate the difference in the logarithm price of the two assets to create the synthetic asset and name IT as SP500*. We and then calculate the parameters for the Ornstein Uhlenbeck work on and puzzle out α = 0.008735, η = 0.007170 for Buffett-factor model, and α = 0.009195, η = 0.007369 for five-factor model. Figs 9 and 10 show the trading bands' tests of the Sdanamp;P500 trading scheme. The deuce figures usher the indistinguishable overall pattern every bit those of Figs 4 and 5. The "a" and transaction toll also have an adverse effect on the expected return bands. We use Eq (8) to calculate the best "a" value for a acknowledged dealings cost, and calculate the expected transaction at the fast transaction cost and "a". As ahead, the tests use two different models to conception the replicating asset, the Snack counter- and five-gene model. Table 2 shows the optimal strategies at different transaction costs. In Table 2, we find that the expected regress of statistical arbitrage of are least 4.8% for the synthetic asset constructed by the ii different factor models.

Fig 9.

The relationship of expected devolve, a, and c for the optimal statistical arbitrage of the Sdanamp;P500 using the replicating asset constructed by the Buffet-factor model. This figure presents the relationship of expected return, a, and transaction monetary value for the best statistical arbitrage of the Sdanamp;P500 using the replicating asset constructed past the Buffett-factor model. This example uses parameters α = 0.008735, η = 0.007170, c = 0.001~0.007.

https://Department of the Interior.org/10.1371/journal.pone.0244541.g009

Fig 10.

The relationship of supposed return, a, and c for the optimal statistical arbitrage of the Sdanamp;P500 using the replicating asset constructed aside the five-constituent posture. This figure presents the kinship of expected return, a, and transaction cost for the optimal applied mathematics arbitrage of the SdanAMP;P500 using the replicating asset constructed away the five-factor model. This example uses parameters α = 0.009195, η = 0.007369, c = 0.001~0.007.

https://Interior Department.org/10.1371/diary.pone.0244541.g010

Tables 1 and 2 show that the obtained returns under statistical arbitrage victimization the replicating portfolio can depend dramatically on the pick of the asset. There are roughly possible reasons for this: when conducting optimal statistical arbitrage under our framework, where the synthetic plus is constructed so that information technology follows an Ornstein Uhlenbeck process, the synthetic asset with a larger disagreement will likely allow many considerable profits since the synthetic asset away construction should drift back and forth around zero in the long-run, meaning that the synthetic asset with the larger disagreement will have entry-story "a" and exit level "m" further away from zero, which prospective leads to higher winnings. In our study, the variance of the synthetic asset bottle-shaped using Berkshire and its replicating portfolio is larger than the variance of the synthetic asset saddle-shaped using the SdanA;P 500 and its replicating portfolio; regardless of whether we use the Buffett-factor modeling to construct the replicating portfolio or the five-component exemplar to construct the replicating portfolio. Thus, the high returns for the statistical arbitrage using Berkshire than that of using the Sdanamp;P 500 is in line with our framework. A possible implication that canful equal drawn from Table 2 is the possibleness that investors can obtain higher returns under the applied math arbitrage frame described in this study if they rear end find assets when paired with its replicating portfolio, which results in a synthetic asset with high disagreement.

Next, we execute analyses in price of Sharpe ratios. Table 3 shows the expected Sharpe ratios from optimal statistical arbitrage of the synthetic plus formed by Berkshire A and its replicating asset BerkA*. Table 4 shows the corresponding Sharpe ratios when the logical plus is keel-shaped by the Sdanamp;P500 and its replicating asset Sdanamp;P500*. As in the previous analyses, we use two assorted models to construct the replicating asset, the Buffet- and five-factor sit. The overall patterns of the Sharpe ratios conform tight to those for the expected returns computed in Tables 1 and 2; and does not affect the main results of this paper, the feasibility of constructing a trading strategy based on statistical arbitrage by constructed a replicating plus.

5. Conclusion

This study provides a fresh method of using replicating assets formed past factor models for optimal statistical arbitrage. The synthetic substance asset formed past the original target asset and the replicating asset follows the Ornstein Uhlenbeck work on closely. Our experiments illustrate the viability of forming the synthetic asset from the original target asset and the replicating asset constructed from factor in models. The contributions of this field of study are as follows: first, we show that the component model can be used to construct a replicating asset that can embody paired with the freehand target asset to perform applied mathematics arbitrage. This method is shown to be executable low-level several dealing costs assumptions. Since the synthetic plus in that study is football-shaped from the original target plus and the replicating asset constructed from factor models, the method works even when it is not possible to find oneself a fit tradable asset to partner off with the original target asset. Bertram [13] solved the optimal entry even out and exit level for statistical arbitrage low-level the generalized Ornstein Uhlenbeck operation. The synthetic asset formed by the original prey asset and the replicating asset was shown to satisfy the Bertram [13] best statistical arbitrage conditions. Second, we apply the Bertram [13] optimal statistical arbitrage formula to compute the trade length and the arbitrage returns for our replicating asset and aim plus dyad. These experiments allowed for the analysis of versatile trading strategies, including the affect of transaction costs. As illustrated in Fig 8, using the factor models to make the replicating asset for applied math arbitrage allows straightforward determination of the entry levels and choke levels of the arbitrage transactions. We use Berkshire A and Sdanadenylic acid;P 500 to form a synthetic asset with the replicating plus constructed from the factor models. The results of this paper indicate that using the divisor models to cast the replicating asset and using the optimal root of the Ornstein Uhlenbeck process to execute statistical arbitrage allows for the determination of first appearance and exit points under the Bertram [13] optimal statistical arbitrage conditions and is profitable. We believe this come near of using the factor models to construct the replicating plus for statistical arbitrage can be applied to some financial market to gain ground wealth.

References

1. 1. Bossaerts Simon Peter. Common nonstationary components of plus prices. Journal of Economic Dynamics and Ascendance. 1988; 12(2–3): 347–364. https://doi.org/10.1016/0165-1889(88)90045-0
   • View Article
   • Google Learner
2. 2. Bossaerts Peter, and Green Richard C.. A General Sense of balance Model of Changing Risk Premia: Theory and Tests. The Review of Financial Studies. 1989; 2(4): 467–493. https://doi.org/10.1093/rfs/2.4.467
   • View Article
   • Google Bookman
3. 3. Bondt De, Werner M., and Richard Thaler. Does the Securities market Overreact? The Journal of Finance. 1985; 40(3): 793–805. https://doi.org/10.1111/j.1540-6261.1985.tb05004.x
   • View Article
   • Google Scholar
4. 4. Lo Andrew W., and A. Craig MacKinlay. When are contrarian profits due to stock market overreaction? The refresh of financial studies. 1990; 3(2): 175–205. https://doi.org/10.1093/rfs/3.2.175
   • View Article
   • Google Scholar
5. 5. Jegadeesh N., and Titman S.. Returns to Buying Winners and Selling Losers: Implications for Origin Market Efficiency. The Journal of Finance. 1993; 48(1): 65–91. https://doi.org/10.1111/j.1540-6261.1993.tb04702.x
   • View Article
   • Google Assimilator
6. 6. Poterba James M., and Summers Lawrence H.. Poor reversion in stock prices: evidence and implications. Diary of Financial Economics. 1988; 22(1): 27–59. https://doi.org/10.1016/0304-405X(88)90021-9
   • View Clause
   • Google Scholar
7. 7. Lehmann Robert I N. Fads, martingales, and market efficiency. The Quarterly Journal of Economics. 1990; 105(1): 1–28. https://doi.org/10.2307/2937816
   • Regar Article
   • Google Scholar
8. 8. Gatev E., Goetzmann , Goetzmann William N.and K. Geert Rouwenhorst. Pairs trading: Performance of a congener-prise arbitrage rule. The Critical review of Financial Studies. 2006; 19(3): 797–827. https://doi.org/10.1093/rfs/hhj020
   • Horizon Article
   • Google Scholar
9. 9. Suffice Binh, and Faff Robert. Does simple pairs trading still work? Financial Analysts Journal. 2010; 66(4): 83–95. https://doi.org/10.2469/faj.v66.n4.1
   • View Clause
   • Google Scholar
10. 10. Bash Binh, and Faff Robert. Are pairs trading profits robust to trading costs? The Journal of Fiscal Research. 2012; 35(2): 261–287. https://doi.org/10.1111/j.1475-6803.2012.01317.x
    • View Clause
    • Google Scholar
11. 11. Lin Yan-Xia, Michael McCrae, and Chandra Gulati. Personnel casualty protection in pairs trading direct minimum profit bounds: A cointegration overture. Journal of Applied Mathematics and Decision Sciences. 2006; 2006(4): 1–14. https://doi.org/10.1155/JAMDS/2006/73803
    • Scene Article
    • Google Scholar
12. 12. Puspaningrum Heni, Lin Yan-Xia, and Gulati Chandra M.. Determination the optimal pre-set boundaries for pairs trading scheme supported cointegration technique. Daybook of Statistical Theory and Practice. 2010; 4(3): 391–419. https://doi.org/10.1080/15598608.2010.10411994
    • View Clause
    • Google Scholar
13. 13. Bertram , William K. Analytic solutions for optimal statistical arbitrage trading. Physica A. 2010; 389(11): 2234–2243. https://doi.org/10.1016/j.physa.2010.01.045
    • View Article
    • Google Scholar
14. 14. Cummins Check off, and Bucca Andrea. Quantitative gap trading on crude oil color and cultured products markets. Quantitative Finance. 2012; 12(12):1857–1875. https://doi.org/10.1080/14697688.2012.715749
    • View Article
    • Google Bookman
15. 15. Fama Eugene F., and French Kenneth R.. Common risk factors in the returns on stocks and bonds. Daybook of Financial Economic science. 1993; 33(1): 3–56.
    • View Clause
    • Google Scholar
16. 16. Carhart , Mark M. Happening persistence in mutual fund performance. The Journal of Finance. 1997; 52(1): 57–82. https://Department of the Interior.org/10.1111/j.1540-6261.1997.tb03808.x
    • View Article
    • Google Scholar
17. 17. Titman Sheridan, Saint John the Apostle Wei K. C., and Feixue Xie. Capital investments and tired returns. Journal of Financial and Vicenary Analysis. 2004; 39(4): 677–700. https://doi.org/10.1017/S0022109000003173
    • Horizon Article
    • Google Scholar
18. 18. Novy-Marx Robert. The other side of measure: The gross profitability premium. Journal of Financial Economics. 2022; 108(1): 1–28. https://DoI.org/10.1016/j.jfineco.2013.01.003
    • View Article
    • Google Scholar
19. 19. Fama Eugene F., and French Kenneth R.. A five-factor asset pricing model. Journal of Financial Economics. 2022; 116(1): 1–22. https://doi.org/10.1016/j.jfineco.2014.10.010
    • View Article
    • Google Scholar
20. 20. Frazzini Andrea, and Lasse Heje Pedersen. Betting Against Beta. Journal of Financial Political economy. 2022; 111(1): 1–25. https://Interior.org/10.1016/j.jfineco.2013.10.005
    • View Article
    • Google Scholar
21. 21. Asness Clifford S., Frazzini Andrea and Lasse Heje Pedersen. Quality Minus Junk. Review of Accounting Studies. 2022; 24(1): 34–112. https://doi.org/10.1007/s11142-018-9470-2
    • Persuasion Article
    • Google Scholar
22. 22. Frazzini Andrea, Kabiller David and Lasse Heje Pedersen. Buffett's Alpha. Financial Analysts Journal. 2022; 74(4): 35–55. https://doi.org/10.2469/faj.v74.n4.3
    • View Clause
    • Google Scholar
23. 23. Mori Masaki, and Ziobrowski Alan J.. Performance of pairs trading strategy in the U.S. REIT Market. Real Estate Economics. 2011; 39(3): 409–428. https://doi.org/10.1111/j.1540-6229.2010.00302.x
    • View Article
    • Google Scholar
24. 24. Broussard John Paul, and Vaihekoski Mika. Profitability of pairs trading strategy in an illiquid market with five-fold share classes. Journal of World Financial Markets, Institutions and Money. 2012; 22(5): 1188–1201. https://doi.org/10.1016/j.intfin.2012.06.002
    • View Article
    • Google Scholar
25. 25. Huckaback Nicolas. The high sensitivity of pairs trading returns. Applied Economics Letters. 2022; 20(14): 1301–1304. https://doi.org/10.1080/13504851.2013.802121
    • Regar Clause
    • Google Student
26. 26. Cohen Lauren, and Frazzini Andrea. System links and foreseeable returns. The Journal of Finance. 2008; 63(4):1977–2011. https://doi.org/10.1111/j.1540-6261.2008.01379.x
    • Persuasion Article
    • Google Scholar
27. 27. Engle Robert F., and Granger C. W. J.. Co-Desegregation and error correction: Representation, idea, and testing. The Econometric Society. 1987; 55(2): 251–276. https://www.jstor.org/stable/1913236
    • View Clause
    • Google Scholar
28. 28. Johansen Søren. Statistical depth psychology of cointegration vectors. Journal of Economic Dynamics and Control. 1988; 12(2–3): 231–54. https://doi.org/10.1016/0165-1889(88)90041-3
    • View Clause
    • Google Scholar
29. 29. Vidyamurthy Ganapathy. Pairs trading: Quantitative methods and analysis. John Wiley danamp; Sons, Hoboken, N.J. 2004.
30. 30. Galenko A., Popova E., and Popova I. Trading in the presence of cointegration. Journal of Alternative Investments. 2012; 15(1): 85–97. https://doi.org/10.3905/jai.2012.15.1.085
    • Though Article
    • Google Scholar
31. 31. Hong, G., and Susmel, R. Pairs-trading in the Asian ADR market, Working paper, University of Houston. 2003. https://www.bauer.uh.edu/rsusmel/Academic/ptadr.pdf
32. 32. Dunis Christian L., and Ho Richard. Cointegration portfolios of European equities for index tracking and market neutral strategies. Daybook of Asset Management. 2005; 6(1): 33–52. https://doi.org/10.1057/Palgrave.jam.2240164
    • View Article
    • Google Scholar
33. 33. Caldeira João, and Moura Guilherme V.. Excerpt of a portfolio of pairs settled on cointegration: A applied math arbitrage scheme. Brazilian Follow-up of Finance. 2022; 11(1): 49–80. https://dx.DoI.org/10.2139/ssrn.2196391
    • View Clause
    • Google Scholar
34. 34. Elliott Robert J., John Van Der Hoekand Malcolm William P.. Pairs trading. Quantitative Finance. 2005; 5(3): 271–276. https://Interior.org/10.1080/14697680500149370
    • Scene Article
    • Google Scholar
35. 35. Avellaneda Marco, and Gypsy Rose Lee Jeong-Hyun. Statistical arbitrage in the US equities market. Decimal Finance. 2010; 10(7): 761–782. https://doi.org/10.1080/14697680903124632
    • View Article
    • Google Assimilator
36. 36. D'Aspremont Alexandre. Identifying small mean-reverting portfolios. Quantitative Finance. 2011; 11(3): 351–364. https://DoI.org/10.1080/14697688.2010.481634
    • View Article
    • Google Scholar
37. 37. Huck Nicolas, and Afawubo Komivi. Pairs trading and excerption methods: is cointegration superior? Practical Economic science. 2022; 47(6): 599–613. https://doi.org/10.1080/00036846.2014.975417
    • View Article
    • Google Scholar
38. 38. Rosenberg Barr, Reid Kenneth, and Lanstein Ronald. Persuasive certify of market inefficiency. Journal of Portfolio Management. 1985; 11(3): 9–16. https://doi.org/10.3905/jpm.1985.409007
    • View Article
    • Google Scholarly person
39. 39. Fama Eugene F., and French Kenneth R.. The Crosswise of Expected Stock Returns. The Diary of Finance.1992; 47(2): 427–465. https://doi.org/10.1111/j.1540-6261.1992.tb04398.x
    • View Article
    • Google Scholar

statistical arbitrage pairs trading strategies

Source: https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0244541

Posted by: webberaninme.blogspot.com

Share this post