Order Submission Behavior Research Articles

Latent Segmentation of Stock Trading Strategies Using Multi-Modal Imitation Learning

While exchanges and regulators are able to observe and analyze the individual behavior of financial market participants through access to labeled data, this information is not accessible by other market participants nor by the general public. A key question, then, is whether it is possible to model individual market participants’ behaviors through observation of publicly available unlabeled market data alone. Several methods have been suggested in the literature using classification methods based on summary trading statistics, as well as using inverse reinforcement learning methods to infer the reward function underlying trader behavior. Our primary contribution is to propose an alternative neural network based multi-modal imitation learning model which performs latent segmentation of stock trading strategies. As a result that the segmentation in the latent space is optimized according to individual reward functions underlying the order submission behaviors across each segment, our results provide interpretable classifications and accurate predictions that outperform other methods in major classification indicators as verified on historical orderbook data from January 2018 to August 2019 obtained from the Tokyo Stock Exchange. By further analyzing the behavior of various trader segments, we confirmed that our proposed segments behaves in line with real-market investor sentiments.

The impact of weather on order submissions and trading performance

Using an account-level dataset consisting of the complete order submissions of various types of traders on the Taiwan Futures Exchange, we perform intraday analyses to investigate whether the weather influences investors' order submission behaviors and subsequent trading performance. The results of our empirical study show that the weather conditions exclusively affect the order aggressiveness of domestic individual traders, not of institutional investors or individual day traders. We also find that the weather negatively affects the trading performance of individual traders. The average daily returns of the most weather-sensitive individual traders are approximately 0.06% (22% per year) lower than those of the least weather-sensitive individual traders. Our study is the first to verify the impact of weather on trader order submission decisions; in addition, our findings provide an alternative explanation for the generally inferior trading performance of individual traders—they are just too weather-sensitive when submitting orders.

Who knows more and makes more? A perspective of order submission decisions across investor types

This research examines the impacts of order submission decision across investor types on overall stock returns and trading performances by using a unique dataset and new measures that present several new findings. In terms of order aggressiveness and trade size, different types of investors display distinct order submission behaviors· Professional investors outperform other investors in the near term following their current-day trading, while aggressive and large sells derive better trading performance for all institutional investors. Only foreign investors and other institutions have better performance from more aggressive buys· Finally, individual traders contribute to total trades more than institutions do, but they tend to use the least aggressive price and smallest trade size per order. Compared to professional investors, individuals perform poorly.

Whose trades move stock prices? Evidence from the Taiwan Stock Exchange

Analyzing firms listed on the Taiwan Stock Exchange, this research examines whose trades move stock prices. To emphasize the roles of different types of investors, we construct the weighted price contribution (WPC), explore sensitivities of the price contribution to order submission behavior, and then test whether their order aggressiveness and trade size, along with current price contribution, affect future stock returns. Our empirical results reveal several new findings. First, individual traders account for more than 80% of WPC, while institutional investors make up less than 20% of WPC. Second, although mutual funds make up the smallest proportion of WPC, their WPC per order (or per lot) and their order aggressiveness are both the highest among all types of traders. Finally, price contributions generated by large orders of professional institutions are closely related to future stock performance, although they account only for a small proportion of total WPC.

Price limit changes, order decisions, and stock price movements: an empirical analysis of the Taiwan Stock Exchange

This study examines the relationships among price limit changes, order submission decisions, and stock returns in the Taiwan Stock Exchange. Specifically, we first investigate whether price limit changes affect overall investors’ order aggressiveness and trade size. We then analyze the relationships across trader types. Finally, we analyze the effects on future stock returns. Our analysis yields the following findings. (1) Stock traders’ order submission decisions are quite different in the post- versus pre-event periods, as they place orders at more aggressive prices but with smaller sizes after the price limit changes. (2) Buy and sell orders demonstrate asymmetric effects with sell orders responding to a policy change in a more diversified way than do buy orders. (3) The sensitivities of order submission decisions to a policy change are significantly different across trader classes. Foreign investors are conservative while investment trusts drastically change their post-event order submission behaviors. (4) Professional institutions’ trade size is a better predictor of future stock returns than order aggressiveness.

Market transparency and closing price behavior on month-end days: Evidence from Taiwan

On February 20, 2012, the Taiwan Stock Exchange launched an information disclosure mechanism that changed its closing auction system from a black box to a more transparent system, with indicative best bid and ask prices disseminated every 20 s throughout the five-minute closing call period. This paper examines the impact of this change on closing price behavior on month-end days. The results show that, following the introduction of the new mechanism, day-end returns and effective spreads decreased significantly on month-end days, suggesting an improvement in market quality and a decline in possible closing price manipulation on month-end days. In addition, the trading volume at market close increased. Further examination of order submission behavior and order imbalance in the closing interval shows that the increase in trading volume at closing is a result of increased market liquidity due to improved transparency. Overall, the introduction of the new closing disclosure rule has effectively enhanced market quality on month-end days.

Analysis of the Spot Market’s T+1 Trading System Effects on the Stock Index Futures Market

T+0 trading system, or day trading system, which allows the investors to buy and sell shares in one day, is a universal trading system in international markets. By contrast, T+1 trading system, which is implemented in China’s stock market, allows investors to sell shares which are bought today only on the next day. T+1 trading system in spot market is a Chinese problem left over by history, and a characteristic in Chinese market. Stock index futures, the financial derivatives based on the stock, are the products of the development of the capital market at certain stage. For the stock index futures are generated based on the stock, they must be closely related. On that basis, what’s the impact of the Chinese special T+1 trading system in spot market on the pricing efficiency, market liquidity and market volatility of the stock index futures market? The paper adopted the method based on the agent-based computational finance which’s different from the traditional method, and built the agent-based computational cross-market platform which includes both several stocks and stock index futures based on MASON. The paper simulated the experiment on the platform to look out the effect on the market quality of the stock index futures market when T+1 trading system in spot market turned into T+0 trading system, consisting mainly of the efficiency of price discovery, market liquidity, market volatility and arbitrage investors’ order submission behavior. The results showed that, compared with the T+1 trading system in spot market, T+0 trading system improved the efficiency of price discovery and market liquidity of stock index futures market, did not raise market volatility, enhance the enthusiasm of arbitrage investors’ order submission behavior. From the perspective of the stock index futures market, we believe that we should restore the T+0 trading system in stock market timely, so as to active the market and improve the efficiency of market information diffusion, to promote the steady and sound environment of the stock index futures market.

The reactions to on-air stock reports: Prices, volume, and order submission behavior

The purpose of this paper is to analyze the real-time responses of stock prices, volume, and order submission behavior across investor groups to 2894 on-air stock reports from 9:16a.m. to 1:15p.m. during regular trading hours from 10/11/2010 to 12/31/2010 in Taiwan. First, positive (negative) reports move stock prices upwards (downwards) in real time, accompanied by increasing trading volume. However, the observed price movements are short-lived and vanish 14days afterwards. Second, responding to the reports, individual investors trade more actively and aggressively than institutional investors do. The overreaction of individual investors is responsible for the observed price movements.

Investor sentiment, order submission, and investment performance on the Taiwan Stock Exchange

We examine investor sentiment, order submission decisions, and investment performance on the Taiwan Stock Exchange. We first specifically investigate whether investor sentiment affects investors' order aggressiveness and/or order size. Second, we empirically analyze the relation between investor sentiment and order submission strategies across various investor classes. Third and finally, we examine the relation between investor sentiment, order submission decisions, and the short- and long-term investment performance. Our analysis yields the following findings: 1) People's order submission behavior is quite different in optimistic versus pessimistic periods. 2) Buy and sell orders demonstrate different patterns and asymmetric effects. 3) The sensitivity of order submission decisions on investor sentiment is significantly different across different trader classes. 4) Mutual funds, compared with other investors, perform better on the Taiwan Stock Exchange.

Position limit for the CSI 300 stock index futures market

The aim of this study was to find the optimal position limit for the Chinese stock index (CSI) 300 futures market. A low position limit helps to prevent price manipulations in the spot market, and thus keeps the magnitude of instantaneous price changes within the tolerance range of policymakers. However, setting a position limit that is too low may also have negative effects on market quality. We propose an artificial limit order market with heterogeneous interacting agents to examine the impact of different levels of position limits on market quality, measured as liquidity, return volatility, efficiency of information dissemination, and trading welfare. The simulation model is based on realistic trading mechanisms, investor structure, and order submission behavior observed in the CSI 300 futures market.Our results show that on the basis of the liquidity status in September 2010, raising the position limit from 100 to 300 could significantly improve market quality and at the same time keep the maximum absolute price change per 5s below the 2% tolerance level. However, the improvement becomes only marginal if the position limit is further increased beyond 300. Therefore, we believe that raising the position limit to a moderate level can enhance the functionality of the CSI 300 futures market, which should benefit the development of the Chinese financial system.

Position-Limit Design for the CSI 300 Futures Markets

The aim of this paper is to find the optimal level of position limit for the Chinese Stock Index (CSI) 300 futures market. A small position limit helps to prevent price manipulations in the spot market, thus able to keep the magnitude of instantaneous price changes within policy makers' tolerance range. However, setting the position limit too small may also have negative effects on market quality. We propose an artificial limit order market with heterogeneous and interacting agents to examine the impact of different levels of position limit on market quality, which is measured by liquidity, return volatility, efficiency of information dissemination and trading welfare. The simulation model is based on realistic trading mechanism, investor structure and order submission behavior observed in the CSI 300 futures market. Our results show that based on the liquidity condition in September 2010, raising the position limit from 100 to 300 can significantly improve market quality and at the same time keep maximum absolute price change per 5 seconds under the 2% tolerance level. However, the improvement becomes only marginal when further increasing the position limit beyond 300. Therefore, we believe that raising the position limit a moderate level can enhance the functionality of the CSI 300 futures market, which benefits the development of the Chinese financial system.

Technical analyses and order submission behaviors: Evidence from an emerging market

The usefulness of technical analyses has never reached a consensus. Unlike most literature studying stock price behaviors surrounding the presence of technical trading signals, this paper examines the heterogeneity in order submission behaviors of investors in the Taiwan Stock Exchange. Our results show that professional institutional investors, particularly foreign investors, behave in a manner consistent with the suggested strategy by the KD trading rule. Namely, after the presence of buy (sell) signals, they intend to buy (sell) the stocks. Conversely, individual investors, acting like liquidity suppliers, tend to net sell (buy) those stocks.

Order submission behaviors and opening price behaviors: evidence from an emerging market

Recent studies document that mutual funds demonstrate an overnight information advantage over foreign investors in Taiwan, a fast emerging market. However, this inequality contradicts the documented foreign investors’ growing market dominance and information advantage. Motivated by the inconsistency, we apply intraday data to examine the linkage between investors’ order submission behaviors and the opening price behaviors of the top net-trade stocks of professional institutional investors. We find that mutual funds exhibit a more persistent and aggressive trading pattern than foreign investors. The order submission behavior of aggressive individuals, following closely mutual funds’ trades, mainly drives the observed informational differences.

The Effects of Removing Price Limits and Term Structure upon Order-Submission Behaviors of IPOs

This paper analyzes the effects of removing price limits and term structure upon order-submission behaviors in IPOs. Many Emerging markets impose limits on trading prices within a trading day to prevent the market from overreaction, especially in IPOs market. With a sample of Taiwan IPOs, the deregulation of price limits in first five trading days offers us an opportunity to insight into how the price limits affect investors' order behaviors. Besides, two anomalies in the pricing of IPOs are well documented: the short-run underpricing and the long-run overpricing phenomenon. Thus, we propose the term structure effect on order-submission behaviors similar to pricing behaviors in different periods and then examines above effect how to affect order-submission behaviors . Our findings are: (1) the previous information indicators of limit order book significantly affect order aggressiveness of market participants in IPOs market; (2) the term factor slightly affect order submission behaviors; (3) removing price limits decreases the degree of relationship between information contents and order aggressiveness.

The Order Submission Behaviors Surrounding Open-Market Repurchase Announcements: The Examination of a Missing Link Embedded in the Signaling Hypothesis

This paper highlights a concern for a link possibly missing in the traditional justification of the signaling hypothesis of open-market repurchases (OMRs). To recover the missing link, we employ the order-level data for the Taiwan stock market to contrast the order submission behaviors among different groups of investors surrounding OMR announcements - who trade the repurchased stocks and how they trade them. As a result, only from the standpoint of price behavior, OMR announcements are reliable signals to which the price reactions of those stocks are favorable. However, the observed order submission behaviors among most investors contradict the essence of the signaling hypothesis.

Market vs. Limit Order Submission Behavior at a Nasdaq Market Maker

This paper considers the order type decision (market vs. limit order) of retail investors. The key question addressed is the extent to which stock-specific characteristics influence the order decision. The focus is primarily empirical, being based on a sample of proprietary order data obtained from a Nasdaq market maker. A simple theoretical model of order submission is put forward to guide the empirics. The stock-specific characteristics considered are the spread, price, volatility and trading activity of the stock. These characteristics serve as explanatory variables in a multinomial logit model which predicts the probability of four discrete order types, based on the aggressiveness of the order. By comparing the results from a 1996 sample with those from a 1998 sample, the impact of the landmark Nasdaq changes of 1997 is also assessed. It was found that limit orders are more prevalent when spreads are wide, as investors attempt to mitigate against high transaction costs. It was also found that limit orders are more prevalent for stocks with high price volatility, suggesting the public investors may have a comparative advantage over professional market makers in the provision of liquidity for volatile stocks. The market reforms of 1997 tended to increase the incidence of limit orders.

Screen Information, Trader Activity, and Bid-Ask Spreads in a Limit Order Market

A key focus of empirical work on limit order markets is the relative importance of individual pieces of information in characterizing order submission and trade execution. We enlarge this focus to include an examination of pricing behavior, using data on index futures trading in a pure electronic limit order book market. A theoretical link between order, trade, and cancellation arrival rates, and the distribution of bid-ask spreads is empirically implemented. Evaluation of models across different information sets is based on relative ability to predict market activity and pricing out-of-sample. A main finding of the paper is the importance and superiority of information embodied in continuous individual traders' actions in characterizing order submission behavior and the structure of pricing. The book information on chararcteristics of resting orders alone cannot explain subsequent order submission, trade, or pricing behavior, and has little impact on the shape of the spread distribution.