Modeling Stock Order Flows and Learning Market-Making from Data

Stock markets employ specialized traders, market-makers, designed to provide liquidity and volume to the market by constantly supplying both supply and demand. In this paper, we demonstrate a novel method for modeling the market as a dynamic system and a reinforcement learning algorithm that learns profitable market-making strategies when run on this model. The sequence of buys and sells for a particular stock, the order flow, we model as an Input-Output Hidden Markov Model fit to historical data. When combined with the dynamics of the order book, this creates a highly non-linear and difficult dynamic system. Our reinforcement learning algorithm, based on likelihood ratios, is run on this partially-observable environment. We demonstrate learning results for two separate real stocks.

Predicting stock index volatility: can market volume help?

This paper explores a number of statistical models for predicting the daily stock return volatility of an aggregate of all stocks traded on the NYSE. An application of linear and non-linear Granger causality tests highlights evidence of bidirectional causality, although the relationship is stronger from volatility to volume than the other way around. The out-of-sample forecasting performance of various linear, GARCH, EGARCH, GJR and neural network models of volatility are evaluated and compared. The models are also augmented by the addition of a measure of lagged volume to form more general ex-ante forecasting models. The results indicate that augmenting models of volatility with measures of lagged volume leads only to very modest improvements, if any, in forecasting performance.

What Does the Market Learn from Stock Offering Revisions?

We examine the disclosure of size revisions of seasoned stock offerings to see what information revisions impart to investros. Revisions could deliver firm-originated infoirmation, which discloses something managers know about the firm. Alternatively, they could disseminate market-originated information, which is information market participants have but which is not conveyed until trading takes place. Our results reject the notion that revisions reveal firm-originated news. Instead, the results are consistent with the market-originated news hypothesis and suggest a mechanism that investros and underwriters use to learn about the demand for an offering.

Types and market classes of live stock,

"First edition, July 1915."

Modeling equity market integration using smooth transition analysis:a study of Eastern European stock markets

This paper assesses the extent to which the equity markets of Hungary, Poland the Czech Republic and Russia have become less segmented. Using a variety of tests it is shown there has been a consistent increase in the co-movement of some Eastern European markets and developed markets. Using the variance decompositions from a vector autoregressive representation of returns it is shown that for Poland and Hungary global factors are having an increasing influence on equity returns, suggestive of increased equity market integration. In this paper we model a system of bivariate equity market correlations as a smooth transition logistic trend model in order to establish how rapidly the countries of Eastern Europe are moving away from market segmentation. We find that Hungary is the country which is becoming integrated the most quickly. © 2005 ELsevier Ltd. All rights reserved.

The relative importance of domestic and global factors in explaining Australian stock returns

In this study, we explore the relative importance of the several documented factors in explaining the behaviour of stock returns for a sample of 157 Australian companies over the period 1993–9. In line with prior evidence, we contend that the influence of global (market, industry and currency) factors is related to the extent of a firm's international activity. We find that Australian firms are in large part impacted by domestic factors with the level of sensitivity declining as the level of international activity increases. In contrast to prior literature, we also show that Australian firm returns are related to regional market, global industry and currency factors and the firm's sensitivity to these factors is an increasing function of its level of international activities.

Stock splits: Implications for investor trading costs

Stock splits are known to have a negative effect on market quality—while stock prices adjust consistently with the split's scale, the bid/ask spread and market depth do not. Two possible explanations for the relative increase in spread are that (i) splits cause an increase in market maker costs that are passed along to investors or (ii) splits provide a mechanism for market makers to increase excess profits. Using a robust econometric methodology, we find evidence of the latter, which raises questions about the motivation of the splitting practice. We also document that while NASDAQ spreads appear to adjust more fully than those of NYSE/AMEX stocks, NASDAQ spreads are higher in general.

O efeito da distribuição sobre o market share em diferentes canais

Apesar dos avanços nos estudos sobre os efeitos das estratégias de canais de distribuição, a relação entre distribuição e participação de mercado é ainda um tópico pouco explorado, especialmente em mercados emergentes. Por essa razão, a proposta central deste artigo é investigar a relação entre a variáveis de distribuição e market share no mercado brasileiro, especificamente no setor de bebidas, comparando dois importantes tipos de canais do mercado de consumo: grandes e pequenos supermercados. Para tanto, foi conduzida uma pesquisa empírica com base em um banco de dados contando com mais de 180 Stock Keeping Unit (SKUs) comercializados por três fabricantes na região sudeste do Brasil. Por meio de análises estatísticas, os resultados indicam existir uma relação convexa e crescente entre as variáveis estudadas, ou seja, existe um ponto de inflexão no qual o crescimento do market share é mais acentuado em razão da distribuição. Além disso, verificou-se que as curvas que descrevem essas variáveis apresentam efeitos diferentes quando comparados grandes e pequenos supermercados.

Open and closed positions and stock index futures volatility

In this paper we analyze the relationship between volatility in index futures markets and the number of open and closed positions. We observe that, although in general both positions are positively correlated with contemporaneous volatility, in the case of S&P 500, only the number of open positions has influence over the volatility. Additionally, we observe a stronger positive relationship on days characterized by extreme movements of these contracting movements dominating the market. Finally, our findings suggest that day-traders are not associated to an increment of volatility, whereas uninformed traders, both opening and closing their positions, have to do with it.

Industry concentration and market volatility

In this paper our aim is to gain a better understanding of the relationship between market volatility and industrial structure. As conflicting results have been documented regarding the relationship between market industry concentration and market volatility, this study investigates this relationship in the time series. We have found that this relationship is only significant and positive for Spain. Our results suggest that we cannot generalize across different countries that market industrial structure (concentration) is a significant factor in explaining market volatility.

Market impact of international sporting and cultural events

We study market reaction to the announcements of the selected country hosting the Summer and Winter Olympic Games, the World Football Cup, the European Football Cup and World and Specialized Exhibitions. We generalize previous results analyzing a large number and different types of mega-events, evaluate the effects for winning and losing countries, investigate the determinants of the observed market reaction and control for the ex ante probability of a country being a successful bidder. Average abnormal returns measured at the announcement date and around the event are not significantly different from zero. Further, we find no evidence supporting that industries, that a priori were more likely to extract direct benefits from the event, observe positive significant effects. Yet, when we control for anticipation, the stock price reactions around the announcements are significant.