Showing papers on "Algorithmic trading published in 2017"
TL;DR: In this paper, the authors study intraday market intermediation in an electronic market before and during a period of large and temporary selling pressure, and they find that the trading pattern of the most active non-designated high frequency traders did not change when prices fell during the Flash Crash.
Abstract: We study intraday market intermediation in an electronic market before and during a period of large and temporary selling pressure. On May 6, 2010, U.S. financial markets experienced a systemic intraday event—the Flash Crash—where a large automated selling program was rapidly executed in the E-mini S&P 500 stock index futures market. Using audit trail transaction-level data for the E-mini on May 6 and the previous three days, we find that the trading pattern of the most active nondesignated intraday intermediaries (classified as High-Frequency Traders) did not change when prices fell during the Flash Crash.
356 citations
Posted Content•
TL;DR: A financial-model-free Reinforcement Learning framework to provide a deep machine learning solution to the portfolio management problem, able to achieve at least 4-fold returns in 50 days.
Abstract: Financial portfolio management is the process of constant redistribution of a fund into different financial products. This paper presents a financial-model-free Reinforcement Learning framework to provide a deep machine learning solution to the portfolio management problem. The framework consists of the Ensemble of Identical Independent Evaluators (EIIE) topology, a Portfolio-Vector Memory (PVM), an Online Stochastic Batch Learning (OSBL) scheme, and a fully exploiting and explicit reward function. This framework is realized in three instants in this work with a Convolutional Neural Network (CNN), a basic Recurrent Neural Network (RNN), and a Long Short-Term Memory (LSTM). They are, along with a number of recently reviewed or published portfolio-selection strategies, examined in three back-test experiments with a trading period of 30 minutes in a cryptocurrency market. Cryptocurrencies are electronic and decentralized alternatives to government-issued money, with Bitcoin as the best-known example of a cryptocurrency. All three instances of the framework monopolize the top three positions in all experiments, outdistancing other compared trading algorithms. Although with a high commission rate of 0.25% in the backtests, the framework is able to achieve at least 4-fold returns in 50 days.
216 citations
TL;DR: Wang et al. as discussed by the authors examined the performance of the Chinese carbon market and found that the rate of return was negatively associated with expected risk, and the kurtosis in trading volume was excessively high and its fluctuations were highly concentrated.
143 citations
TL;DR: In this article, the authors present a survey of the growing literature on pairs trading frameworks, i.e., relative value arbitrage strategies involving two or more securities, and provide an in-depth assessment of each approach, revealing strengths and weaknesses relevant for further research.
Abstract: This survey reviews the growing literature on pairs trading frameworks, i.e., relative-value arbitrage strategies involving two or more securities. Research is categorized into five groups: The distance approach uses nonparametric distance metrics to identify pairs trading opportunities. The cointegration approach relies on formal cointegration testing to unveil stationary spread time series. The time-series approach focuses on finding optimal trading rules for mean-reverting spreads. The stochastic control approach aims at identifying optimal portfolio holdings in the legs of a pairs trade relative to other available securities. The category “other approaches” contains further relevant pairs trading frameworks with only a limited set of supporting literature. Finally, pairs trading profitability is reviewed in the light of market frictions. Drawing from a large set of research consisting of over 100 references, an in-depth assessment of each approach is performed, ultimately revealing strengths and weaknesses relevant for further research and for implementation.
140 citations
TL;DR: In this article, the authors analyzed the carbon emission market of four representative cities in China and showed that the carbon trading market in China has only achieved weak efficiency, while the semi strong efficiency and the strong efficiency have not been reached.
Abstract: In 2014, China proposed medium and long-term low carbon development goals in China-U.S. Joint Statement on Climate Change that the emission of carbon dioxide would reach its peak and the proportion of non-fossil energy accounted for the primary energy consumption would increase to 20% in 2030. In order to achieve these goals, the unified carbon emission trading system should be put into effect by 2017, the implementation of the unified carbon emission trading system depends on the effectiveness of the current carbon trading market in China. On the basis of the effective market theory and fair game model, the unit root test and the run test are developed to analyze the carbon emission market of four representative cities in China. The results show that (1) the carbon trading market in China has only achieved weak efficiency, while the semi strong efficiency and the strong efficiency have not been reached; (2) with the expansion of the market scale, the increase of trading volume, the carbon trading market would converge from the state of inefficiency to weak form efficiency gradually, and the carbon trading market in China shows signs of restoring market efficiency.
137 citations
TL;DR: This study explores how the performance of the predictive system depends on a combination of a forecast horizon and an input window length for forecasting variable horizons.
101 citations
TL;DR: In this paper, the authors study intraday market intermediation in an electronic market before and during a period of large and temporary selling pressure, and find that the trading pattern of the most active non-designated high frequency traders (classified as High Frequency Traders) did not change when prices fell during the Flash Crash.
Abstract: We study intraday market intermediation in an electronic market before and during a period of large and temporary selling pressure. On May 6, 2010, U.S. financial markets experienced a systemic intraday event, known as the Flash Crash, when a large automated sell program was rapidly executed in the E-mini S&P 500 stock index futures market. Using audit trail transaction-level data for the E-mini on May 6 and the previous three days, we find that the trading pattern of the most active non-designated intraday intermediaries (classified as High Frequency Traders) did not change when prices fell during the Flash Crash.
90 citations
21 Dec 2017
TL;DR: In this article, a C++ implementation on the Intel Xeon Phi co-processor was used for backtesting a simple trading strategy over 43 Commodity and FX future mid-prices at 5-minute intervals.
Abstract: Deep neural networks (DNNs) are powerful types of artificial neural networks (ANNs) that use several hidden layers. They have recently gained considerable attention in the speech transcription and image recognition community (Krizhevsky et al., 2012 ) for their superior predictive properties including robustness to overfitting. However their application to algorithmic trading has not been previously researched, partly because of their computational complexity. This paper describes the application of DNNs to predicting financial market movement directions. In particular we describe the configuration and training approach and then demonstrate their application to backtesting a simple trading strategy over 43 different Commodity and FX future mid-prices at 5-minute intervals. All results in this paper are generated using a C++ implementation on the Intel Xeon Phi co-processor which is 11.4x faster than the serial version and a Python strategy backtesting environment both of which are available as open source code written by the authors.
83 citations
TL;DR: In this article, the authors study performance, concentration, and competition in the high-frequency trading (HFT) industry and find that small differences in HFT firms' latencies are associated with large differences in trading revenues.
Abstract: We study performance, concentration, and competition in the high-frequency trading (HFT) industry. Small differences in HFT firms’ latencies are associated with large differences in trading revenues. The fastest HFT firms capture a higher quantity of profitable trades and perform better in strategies such as liquidity provision and cross-market arbitrage. Consistent with theory suggesting that competition on relative latency leads to a concentrated industry, concentration of HFT revenues and trading volume is high and non-declining over the five-year sample. New entrants are typically slower, earn lower revenues and are more likely to exit, which likely reinforces concentration in the HFT industry.
81 citations
TL;DR: Analytical solutions for the robust optimal strategies are provided, the resulting dynamic programming equations have classical solutions, and a proof of verification is provided about the behavior of the ambiguity averse MM.
Abstract: Because algorithmic traders acknowledge that their models are incorrectly specified we allow for ambiguity in their choices to make their models robust to misspecification. We show how to include misspecification to: (i) the arrival rate of market orders (MOs), (ii) the fill probability of limit orders, and (iii) the dynamics of the midprice of the asset they trade. In the context of market making, we demonstrate that market makers (MMs) adjust their quotes to reduce inventory risk and adverse selection costs. Moreover, robust market making increases the strategy's Sharpe ratio and allows the MM to fine tune the tradeoff between the mean and the standard deviation of profits. Our framework adopts a robust optimal control approach and we provide existence and uniqueness results for the robust optimal strategies as well as a verification theorem. The behavior of the ambiguity averse MM generalizes that of a risk averse MM, and coincide in only one circumstance.
77 citations
19 Nov 2017
TL;DR: This work demonstrates that the Heterogeneous Autoregressive model for Realized Volatility Andersen et al. (2007) applies reasonably well to the BTCUSD dataset and shows that an artificial neural network prediction is capable of approximate capture of the actual log return distribution.
Abstract: Bitcoin has the largest share in the total capitalization of cryptocurrency markets currently reaching above 70 billion USD. In this work we focus on the price of Bitcoin in terms of standard currencies and their volatility over the last five years. The average day-to-day return throughout this period is 0.328%, amounting in exponential growth from 6 USD to over 4,000 USD per 1 BTC at present. Multi-scale analysis is performed from the level of the tick data, through the 5 min, 1 hour and 1 day scales. Distribution of trading volumes (1 sec, 1 min, 1 hour and 1 day) aggregated from the Kraken BTCEUR tick data is provided that shows the artifacts of algorithmic trading (selling transactions with volume peaks distributed at integer multiples of BTC unit). Arbitrage opportunities are studied using the EUR, USD and CNY currencies. Whereas the arbitrage spread for EUR-USD currency pair is found narrow at the order of a percent, at the 1 hour sampling period the arbitrage spread for USD-CNY (and similarly EUR-CNY) is found to be more substantial, reaching as high as above 5 percent on rare occasions. The volatility of BTC exchange rates is modeled using the day-to-day distribution of logarithmic return, and the Realized Volatility, sum of the squared logarithmic returns on 5-minute basis. In this work we demonstrate that the Heterogeneous Autoregressive model for Realized Volatility Andersen et al. (2007) applies reasonably well to the BTCUSD dataset. Finally, a feed-forward neural network with 2 hidden layers using 10-day moving window sampling daily return predictors is applied to estimate the next-day logarithmic return. The results show that such an artificial neural network prediction is capable of approximate capture of the actual log return distribution; more sophisticated methods, such as recurrent neural networks and LSTM (Long Short Term Memory) techniques from deep learning may be necessary for higher prediction accuracy.
TL;DR: This paper studied the pre-and post-publication return predictability of 138 anomalies in 39 stock markets and found that the United States is the only country with a statistically significant and economically meaningful postpublication decline in long/short returns.
Abstract: Motivated by McLean and Pontiff (2016), we study the pre- and post-publication return predictability of 138 anomalies in 39 stock markets. Based on more than a million anomaly country-months, we find that the United States is the only country with a statistically significant and economically meaningful post-publication decline in long/short returns. The surprisingly large differences between the U.S. and international markets cannot be fully explained with general time effects or differences in limits to arbitrage, in-sample anomaly profitability, data availability, or local risk factor exposure. Our results have implications for the recent literature on arbitrage trading, data mining, and market segmentation.
TL;DR: In this paper, the authors examined how investor sentiment and trading behavior affect asset returns in the Korean market and found that high investor sentiment is correlated with high trading behavior, while low trading behavior is associated with low investor sentiment.
Abstract: This article examines how investor sentiment and trading behaviour affect asset returns. By analysing the unique stock trading dataset of the Korean market, we find that high investor sentiment ind...
TL;DR: In this paper, the authors study the dynamics of high-frequency market efficiency measures and provide evidence that these measures comove across stocks and with each other, suggesting the existence of a systematic market efficiency component.
Abstract: We study the dynamics of high-frequency market efficiency measures. We provide evidence that these measures comove across stocks and with each other, suggesting the existence of a systematic market efficiency component. In vector autoregressions, we show that shocks to funding liquidity (the TED spread), hedge fund assets under management, and a proxy for algorithmic trading are significantly associated with systematic market efficiency. Thus, stock market efficiency is prone to systematic fluctuations, and, consistent with recent theories, events and policies that impact funding liquidity can affect the aggregate degree of price efficiency.
TL;DR: The proposal is far superior to the previous flag pattern strategies as regards both profitability and risk and seems to challenge market efficiency in line with other similar studies, in the specific analysis carried out on the DJIA index.
Abstract: We propose an automatic and dynamic trading rule based on flag pattern recognition.The strategy does not depend on the ability of the trader to guess the best configuration of the trading rule.We include several filters for the trades, one of them considering the EMA indicator in short and medium timeframes.The trading rule is applied on a large intraday database for the DJIA index.We can conclude that our proposal is far superior to the previous flag pattern strategies as regards both profitability and risk. In this paper we propose and validate a trading rule based on flag pattern recognition, incorporating important innovations with respect to the previous research. Firstly, we propose a dynamic window scheme that allows the stop loss and take profit to be updated on a quarterly basis. In addition, since the flag pattern is a trend-following pattern, we have added the EMA indicator to filter trades. This technical analysis indicator is calculated both for 15-min and 1-day timeframes, which enables short and medium terms to be considered simultaneously. We also filter the flags according to the price range on which they are developed and have limited the maximum loss of each trade to 100 points. The proposed methodology was applied to 91,309 intraday observations of the DJIA index, considerably improving the results obtained in the previous proposals and those obtained by the buy & hold strategy, both for profitability and risk, and also after taking into account the transaction costs. These results seem to challenge market efficiency in line with other similar studies, in the specific analysis carried out on the DJIA index and is also limited to the setup considered.
TL;DR: In this article, the impact of incomplete risk trading and its impact on investment in the European power sector has been investigated, where the authors apply computable stochastic equilibrium models on a simple market model of the Energy Only type.
TL;DR: In this paper, the authors demonstrate that robust market making increases the strategies' Sharpe ratio and allows the MM to fine tune the trade-off between the mean and the standard deviation of profits.
Abstract: Algorithmic traders acknowledge that their models are incorrectly specified, thus we allow for ambiguity in their choices to make their models robust to misspecification in (i) the arrival rate of market orders, (ii) the fill probability of limit orders, and (iii) the dynamics of the midprice of the asset they deal. In the context of market making, we demonstrate that market makers (MMs) adjust their quotes to reduce inventory risk and adverse selection costs. Moreover, robust market making increases the strategies' Sharpe ratio and allows the MM to fine tune the trade-off between the mean and the standard deviation of profits. We provide analytical solutions for the robust optimal strategies, show that the resulting dynamic programming equations have classical solutions, and provide a proof of verification. The behavior of the ambiguity averse MM is found to generalize those of a risk averse MM and coincide in a limiting case.
TL;DR: The main idea is to use an event-based time scale based on a new way of summarising data, called Directional Changes, combined with a genetic algorithm, to find a trading strategy that maximises profitability in foreign exchange markets.
Abstract: The majority of forecasting methods use a physical time scale for studying price fluctuations of financial markets, making the flow of physical time discontinuous. Therefore, using a physical time scale may expose companies to risks, due to ignorance of some significant activities. In this paper, an alternative and original approach is explored to capture important activities in the market. The main idea is to use an event-based time scale based on a new way of summarising data, called Directional Changes. Combined with a genetic algorithm, the proposed approach aims to find a trading strategy that maximises profitability in foreign exchange markets. In order to evaluate its efficiency and robustness, we run rigorous experiments on 255 datasets from six different currency pairs, consisting of intra-day data from the foreign exchange spot market. The results from these experiments indicate that our proposed approach is able to generate new and profitable trading strategies, significantly outperforming other traditional types of trading strategies, such as technical analysis and buy and hold.
TL;DR: In this article, the authors examined the intraday price discovery and volatility spillover relationship between the CSI 300 equity index and index futures in China and found that index futures plays a dominant role in contributing towards price discovery, with an average yearly information share of about 67%.
Abstract: The introduction of stock index futures in China in 2010 marked an important development in the country's financial markets. It was however not without controversy as regulators blamed the futures market for its role in the stock market crash in 2015. This paper examines the intraday price discovery and volatility spillover relationship between the CSI 300 equity index and index futures in China. Results from the study, covering the period 2010–2015, reveal that index futures plays a dominant role in contributing towards price discovery, with an average yearly information share of about 67%. The price leadership of the futures market, although found to be strong, is diminished in the presence of stringent regulatory trading curbs that were put in place as a response to the crisis. Furthermore, investigation into volatility spillover documents significant return and volatility shocks transmitted from the stock market to the futures market. The evidence, which contradicts regulatory claims, is explained in the context of the unique institutional trading structure in China.
TL;DR: In this article, the authors investigated how lit and dark market fragmentation affects liquidity and found that neither dark trading, nor fragmentation between lit order books, is found to harm liquidity, or at worst does not affect them.
Abstract: Based on data from eight exchanges and a trade reporting facility for a large sample of LSE- and Euronext-listed equities, this article investigates how lit and dark market fragmentation affects liquidity. Neither dark trading, nor fragmentation between lit order books, is found to harm liquidity. Lit fragmentation improves spreads and depth across markets and locally on the primary exchange, or at worst does not affect them. Benefits are greater for large stocks and stocks with less electronic trading. Lit fragmentation however harms the depth of small stocks. Adverse effects on the depth of large stocks result from algorithmic trading and not from fragmentation.
13 Apr 2017
TL;DR: In this article, a neural network-based stock price prediction and trading system using technical analysis indicators is presented, where the model developed first converts the financial time series data into a series of buy-sell-hold trigger signals using the most commonly preferred technical indicators.
Abstract: In this paper, a neural network-based stock price prediction and trading system using technical analysis indicators is presented. The model developed first converts the financial time series data into a series of buy-sell-hold trigger signals using the most commonly preferred technical analysis indicators. Then, a Multilayer Perceptron (MLP) artificial neural network (ANN) model is trained in the learning stage on the daily stock prices between 1997 and 2007 for all of the Dow30 stocks. Apache Spark big data framework is used in the training stage. The trained model is then tested with data from 2007 to 2017. The results indicate that by choosing the most appropriate technical indicators, the neural network model can achieve comparable results against the Buy and Hold strategy in most of the cases. Furthermore, fine tuning the technical indicators and/or optimization strategy can enhance the overall trading performance.
TL;DR: In this paper, the authors investigate how lit and dark market fragmentation affects liquidity and find that neither dark trading nor fragmentation between lit order books is found to harm liquidity, or at worst does not affect them.
TL;DR: In this article, the authors studied the impact of increasing trading frequency in financial markets on allocative efficiency and showed that the trading frequency that maximizes allocating efficiency coincides with the information arrival frequency for scheduled information releases.
Abstract: This article studies the impact of increasing trading frequency in financial markets on allocative efficiency We build and solve a dynamic model of sequential double auctions in which traders trade strategically with demand schedules Trading needs are generated by time-varying private information about the asset value and private values for owning the asset, as well as quadratic inventory costs We characterize a linear equilibrium with stationary strategies and its efficiency properties in closed form Frequent trading (more double auctions per unit of time) allows more immediate asset reallocation after new information arrives, at the cost of a lower volume of beneficial trades in each double auction Under stated conditions, the trading frequency that maximizes allocative efficiency coincides with the information arrival frequency for scheduled information releases, but can far exceed the information arrival frequency if new information arrives stochastically A simple calibration of the model suggests that a moderate market slowdown to the level of seconds or minutes per double auction can improve allocative efficiency for assets with relatively narrow investor participation and relatively infrequent news, such as small- and micro-cap stocks
TL;DR: This work examines the efficacy and the feasibility of developing a stacked generalization system, intelligently combining the predictions of diverse machine learning models, and establishes a novel inferential framework that leads to significantly better trading performance than the considered benchmarks.
Abstract: Multiple FOREX time series forecasting is a hot research topic in the literature of portfolio trading. To this end, a large variety of machine learning algorithms have been examined. However, it is now widely understood that, in real-world trading settings, no single machine learning model can consistently outperform the alternatives. In this work, we examine the efficacy and the feasibility of developing a stacked generalization system, intelligently combining the predictions of diverse machine learning models. Our approach establishes a novel inferential framework that comprises the following levels of data processing: (i) We model the dependence patterns between major currency pairs via a diverse set of commonly used machine learning algorithms, namely support vector machines (SVMs), random forests (RFs), Bayesian autoregressive trees (BART), dense-layer neural networks (NNs), and naive Bayes (NB) classifiers. (ii) We generate implied signals of exchange rate fluctuation, based on the output of these models, as well as appropriate side information obtained by analyzing the correlations across currency pairs in our training datasets. (iii) We finally combine these implied signals into an aggregate predictive waveform, by leveraging majority voting, genetic algorithm optimization, and regression weighting techniques. We thoroughly test our framework in real-world trading scenarios; we show that our system leads to significantly better trading performance than the considered benchmarks. Thus, it represents an attractive solution for financial firms and corporations that perform foreign exchange portfolio management and daily trading. Our system can be used as an integrated part in international commercial trade activities or in a quantitative investing framework for algorithmic trading and carry-trade speculation.
TL;DR: In this article, the authors show that an equity pairs trading strategy generates large and significant abnormal returns, and that this return is not driven purely by the short-term reversal of returns.
Abstract: We show that an equity pairs trading strategy generates large and significant abnormal returns We find that this return is not driven purely by the short-term reversal of returns The evidence related to the cross-sectional variation, the time-series variation, and the persistence of the pairs trading profits, and the determinants of return correlations is consistent with the delay in information diffusion as the driver for the pairs trading strategy Evidence from the liquidity factor and the recent financial crisis suggests that the short-term liquidity provision is not the main cause of the pairs trading strategy
TL;DR: In this paper, the authors introduce novel "doubly mean-reverting" processes based on conditional modelling of model spreads between pairs of stocks, which can capture local market inefficiencies that are elusive for traditional pairs trading strategies.
Abstract: This paper introduces novel ‘doubly mean-reverting’ processes based on conditional modelling of model spreads between pairs of stocks. Intraday trading strategies using high frequency data are proposed based on the model. This model framework and the strategies are designed to capture ‘local’ market inefficiencies that are elusive for traditional pairs trading strategies with daily data. Results from real data back-testing for two periods show remarkable returns, even accounting for transaction costs, with annualized Sharpe ratios of 3.9 and 7.2 over the periods June 2013–April 2015 and 2008, respectively. By choosing the particular sector of oil companies, we also confirm the observation that the commodity price is the main driver of the share prices of commodity-producing companies at times of spikes in the related commodity market.
TL;DR: In this article, the influence of financialization on metal spot prices and in particular on respective volatility has been insufficiently studied, and the potential effects of the lead-lag relationship on futures trading activity of commercial and non-commercial market participants and cash prices and volatility for the major metal commodities: copper, gold, silver, platinum, and palladium.
01 Mar 2017
TL;DR: This research combines Markov decision process and genetic algorithms to propose a new analytical framework and develop a decision support system for devising stock trading strategies and confirms that the model presented in this research can yield higher rewards than other benchmarks.
Abstract: The paper proposed a novel application for incorporating Markov decision process on genetic algorithms to develop stock trading strategies.This predicts the results of applying the Markov decision process with real-time computational power to help investors formulate correct timing (portfolio adjustment) and trading strategies (buy or sell).This study thus uses the excellent genetic algorithm parallel space searching ability to provide investors with the optimal stock selection strategy and capital allocation, and combines them with both constructs to solve the portfolio problem and improve return on investment for investors.This research can solve stock selection, market timing and capital allocation at the same time for investors when investing in the stock market. With the arrival of low interest rates, investors entered the stock market to seek higher returns. However, the stock market proved volatile, and only rarely could investors gain excess returns when trading in real time. Most investors use technical indicators to time the market. However the use of technical indicators is associated with problems, such as indicator selection, use of conflicting versus similar indicators. Investors thus have difficulty relying on technical indicators to make stock market investment decisions.This research combines Markov decision process and genetic algorithms to propose a new analytical framework and develop a decision support system for devising stock trading strategies. This investigation uses the prediction characteristics and real-time analysis capabilities of the Markov decision process to make timing decisions. The stock selection and capital allocation employ string encoding to express different investment strategies for genetic algorithms. The parallel search capabilities of genetic algorithms are applied to identify the best investment strategy. Additionally, when investors lack sufficient money and stock, the architecture of this study can complete the transaction via credit transactions. The experiments confirm that the model presented in this research can yield higher rewards than other benchmarks.
TL;DR: This article found that restrictions placed on the CSI 300 and CSI 500 index futures trading during the recent Chinese stock market crisis deteriorated spot market quality, particularly the September trade restrictions, which can be explained by the sudden risk exposure faced by alpha-strategy traders who stop trading spots after the CSI futures trading restrictions are introduced.
Abstract: Using a difference-in-difference approach, we find that restrictions placed on the CSI 300 and CSI 500 index futures trading during the recent Chinese stock market crisis deteriorated spot market quality, particularly the September trade restrictions. Our results can be explained by the sudden risk exposure faced by alpha-strategy traders who stop trading spots after the index futures trading restrictions are introduced, thus worsening the spot market quality. © 2016 Wiley Periodicals, Inc. Jrl Fut Mark 37:411–428, 2017
TL;DR: In this article, the authors show that specified pools that are eligible to be traded as TBAs have significantly lower trading costs than other SPs and that dealers hedge SP inventory with TBA trades, and are more likely to prearrange trades in SPs that are difficult to hedge.
Abstract: Agency mortgage-backed securities (MBS) trade simultaneously in a market for specified pools (SPs) and in the to-be-announced (TBA) forward market. TBA trading creates liquidity by allowing thousands of different MBS to be traded in a handful of TBA contracts. SPs that are eligible to be traded as TBAs have significantly lower trading costs than other SPs. We present evidence that TBA eligibility, in addition to characteristics of TBA-eligible SPs, lowers trading costs. We show that dealers hedge SP inventory with TBA trades, and they are more likely to prearrange trades in SPs that are difficult to hedge.