trading day

Quantitative and Derivative Strategies (“QDS”) of Morgan Stanley’s Institutional Equity Division Sales Desk have prepared this piece. This is not a product of Morgan Stanley’s Research Department and you should not regard it as a research report.

Q U A N T I T A T I V E A N D D E R I V A T I V E S T R A T E G I E S

Quant Times FOR INSTITUTIONAL USE ONLY F E B R U A R Y 3 , 2 0 1 1

Stages of the US Trading Day – Optimize Your Execution Equity markets continue to evolve rapidly. The widespread use of algorithms, both on the liquidity demanding and the providing side has impacted the intraday price discovery process. Greater concentration in liquidity demand has led to a ‘synchronization’ in trading behavior. Beta-driven trading now accounts for a greater percentage of trading volume. In this paper, we review the impact of these developments on market stability and efficiency. We introduce a new way of thinking about liquidity provisioning, and show how this can be used to optimally interact with trade flows.

Three Stages of Trading Day in US

• Price Discovery (9:30-10:00): Overnight news on a macro and micro level gets priced into risky assets.

• Institutional Portfolio Rebalancing (10:00-13:00): Stock-specific portfolio rebalancing, driven by longer-term alpha views.

• Beta-Driven Trading (13:00-16:00): Market is dominated by beta trading, such as mutual fund flows or index fund rebalances.

Liquidity Demand and Supply – Footprint of Algorith ms

• Automated execution strategies and the way they are used can lead to repetitive patterns.

• We differentiate between front-loaded and back-loaded liquidity provisioning – profitability depends on liquidity demanding strategies in use.

• Complexity in execution strategies is critical in limiting price impact.

Implications for Execution Strategies

• Market participants need to be aware of the impact of the synchronization of behavior.

• Opportunistically adjusting execution strategies improves market impact cost.

• We propose alternative solutions and discuss how buy side trading desks should approach tailoring their execution strategies to lower price impact.

QUANTITATIVE AND DERIVATIVE

STRATEGIES

United States

Simon Emrich +1 212 761 8254 [email protected]

Amir Khandani +1 212 761 8679 [email protected]

Relevant QDS Articles “Stages of the Trading Day – Think Globally, Act Regionally”, February 2011.

“Recent Liquidity Imbalances and Price Evolution”, May 2010.

“US Market: Status Quo Ante, But With A Twist”, May 2010.

“Using Smarter Algorithms vs. Smarter Use of Algorithms”, April 2010.

“Liquidity, “Endogenous Risk’ and Quant Returns”, April 2009.

“Quant 2.0?”, December 2007.

FOR INSTITUTIONAL USE ONLY

1 This material is not a solicitation of any offer to buy or sell any security or other financial instrument or to participate in any trading strategy. This material was not prepared by the

Morgan Stanley research department. Please refer to important information and qualifications at the end of this material. The information contained herein does not constitute advice. Morgan Stanley is not acting as your advisor (municipal, financial otherwise) and is not acting in a fiduciary capacity.

When to Trade What

The microstructure of equity markets, and of the behavior of its participants, shows marked patterns, particularly on an intraday basis. In this note, we introduce some of the key observed patterns. We will argue that these patterns can be linked both to the structure of equity markets and to the decisions made by market participants who trade to express their views while keeping their ultimate intended positioning private. Understanding the microstructure of such interactions and the objectives of market participants is critical in selecting optimal trading strategies.

At a high level of abstraction, trading in risky assets can be viewed as a liquidity provisioning strategy – we ‘buy low and sell high’, hoping to earn a return equal to or in excess of the appropriate risk premium. Portfolio construction, from asset allocation to stock selection, has this guiding objective function.

When it comes to execution strategies, however, a number of different reasons can drive trading decisions. Not all of these reasons are compatible with the objective of being a liquidity provider to the market place. In many cases, constraints lead to execution strategies demanding liquidity and hence a market impact cost. For example, managing cash flows into or out of an institutional portfolio such as a mutual fund might lead to liquidity demanding trading practices. Similarly, systematic rebalancing strategies, particularly for index-replicating portfolios, can lead to liquidity demanding executions. Lastly, the empirical concentration in assets under management leads to large order sizes relative to the instantaneous liquidity. In an

environment of increasingly fragmented equity markets, this might tilt execution strategies away from liquidity provisioning towards liquidity demanding.

These different reasons for trading, as well as the current practice of execution using systematic, algorithmic trading strategies, have led to a stratification of the trading day into three distinct phases. Exhibit 1 shows the average intraday evolution of a number of market statistics, as well as some derived statistics, for S&P 500 constituents in 2010. Based on these statistics, as well as on our observations on market participants’ trading decisions, we decompose the trading day into three distinct stages:

• Price Discovery (9:30-10:00): Overnight news on a macro and micro level gets priced into risky assets.

• Institutional Portfolio Rebalancing (10:00-13:00): Stock-specific portfolio rebalancing, driven by longer-term alpha views

• Beta-Driven Trading (13:00-16:00): Market is dominated by beta trading, such as mutual fund flows or index fund rebalances as well as derivatives delta and gamma hedging.

Because of these distinct stages, market impact of a given trading strategy will vary depending on the time of day at which it is executed. In many cases, stock-specific trading during a beta-dominated period can be attractive from an impact perspective, for example. We recommend tailoring execution strategies based on such regularities in the marketplace.




Exhibit 1: The Three Stages of the Trading Day – 20 10

Source: Morgan Stanley Quantitative and Derivative Strategies

Stages of the Trading Day

The left side of Exhibit 1 shows the intraday evolution of market statistics such as the median Bid/Ask Spread, the median available size at the NBBO (National Best Bid/Offer), the cross-sectional standard deviation of stock returns and the median turnover (normalized volume). The sidebar on the next page provides details on the calculation of these metrics. We have introduced most of them in an earlier paper1, where we discussed the evolution of their intraday patterns over time. As an example, we found that Turnover has become increasingly back-loaded in the day, with more than 30% of total trading volume now occurring during the last hour of the day. Meanwhile, most of the relative price moves of stocks, as measured by the Return Dispersion, happen early in the day, which also leads to higher bid/ask spreads early in the morning (our Spread Profile).

The right side of Exhibit 1 shows a number of derived measures. Our Liquidity Measure is a function of the average percentage change in prices for a given unit of turnover. The higher the liquidity measure, the lower the market impact for a single-stock trade. The Index Turnover Attribution Measure shows the proportion of single-stock dollar volume that can be explained by the market

1 See “Using Smarter Algorithms vs. Smarter Use of Algorithms”, Institutional Investor Journals’ A Guide to Global Liquidity II, Spring 2010.

capitalization of the stock. The higher this measure, the more likely it is that index slice trading, such as that coming from index fund inflows/outflows, drives single-stock trading. Lastly, we show the Median Pairwise Correlation of index constituent returns, using a rolling 30-minute window of 1-minute returns.. This correlation metric is based on the pairwise correlation across all stocks in the S&P 500 on an intraday basis, rather than on the more commonly used close-to-close correlation.

Exhibit 1 shows the average shape of our statistics for 2010. We also investigate whether there are changes in these variables over time. Exhibit 2 presents a heatmap view for some of the key statistics from January 2008 to November 2010. To show the intraday shape more clearly, the normalized charts on the right hand side take out each day’s average value of the statistic. The normalized statistics show that the averages in Exhibit 1 reflect market conditions on most trading days – turnover increases after 15:00, pairwise correlation after 13:00. Index turnover attribution declines in the morning and then increases after 14:00 and particularly after 15:30, while the available size at the NBBO increases throughout the day.

All four measures increase significantly after 13:00, in particular during the last hour of trading. We attribute this to a greater proportion of beta-driven trading.




Clearly, there are also periods where the averages change, particularly during stress situations in the market – during April and May 2010, or during Q4 2008. We find that average pairwise correlation, turnover and index turnover attribution were all elevated during these periods, indicating a greater macro focus during this period leading to more index trading.

Conversely, liquidity demand often becomes more persistently one-sided during these periods. This leads to a decline in market depth, as measured by available size at the NBBO. Market-making liquidity provisioning strategies, which are based on relatively rapid order decay, become less profitable when liquidity demand is persistently one-sided. They reduce their activity as a result. Since market-making strategies are typically the marginal setters of the NBBO spread, this leads to a widening of spreads. Price change accelerates to attract liquidity suppliers to clear the market. However, if there is significant rebalancing of similarly constructed portfolios (such as during the ‘quant crisis’ of August 2007), liquidity imbalances can persist for multiple days2. They need not have a significant impact on the market index level if – as in August 2007 – the imbalances are concentrated in a long/short, market neutral portfolio of stocks.

We also observe a number of artifacts in the data around full hours and half hours of the trading day – such as spikes in turnover, particularly at 10:00, and increases in available size at the NBBO every half hour. We believe that these are driven by biases both in trader behavior and in the implementation of systematic trading strategies, which lead to updating waves occurring at these focal times. These artifacts are prevalent and persistent throughout our datasets and across most of the statistics we consider.3

2 See ‘Quant 2.0?’, Morgan Stanley Quantitative and Derivative Strategies, December 2007, and Amir Khandani and Andrew Lo, “What Happened To The Quants in August 2007?’, Journal of Portfolio Management, Fall 2007

3 Macro news releases at 10:00 have an impact. However, they cannot fully explain the patterns. First, we observe the patterns daily, while macro news releases are sparser. Second, these 10:00 patterns are a relatively recent phenomenon – similar analysis for 2005 shows much smaller spikes at 10:00.

INTRADAY MARKET METRICS – CALCULATION

DETAILS

Throughout the paper, we define and compute the following metrics:

Spread: Median of bid-ask spread relative to the volume weighted average interval price for all index constituents.

Available size: For each security, calculate the average of last ask and bid size at the NBBO (National Best Bid/Offer) for every interval. We report the median across all index constituents.

1-Minute Return Dispersion: Cross-sectional standard deviation of 1-minute returns. To arrive at a more robust metrics, we remove the top and bottom 5% of securities based on 1-minute return. This trimming procedure has little impact on the intraday evolution of dispersion.

Cumulative Return Dispersion: Cross-sectional standard deviation of cumulative returns since the open price. We use the same trimming procedure as for the 1-minute return dispersion. Excluding overnight returns changes the starting point of the cumulative return profile but not the shape.

Turnover : Calculated as the interval volume divided by the number of shares outstanding for each index constituent.

Liquidity Measure: the negative of the cross-sectional regression coefficient β of

iii Tr εβ +=

for all stocks i, where r i is the 1-minute return (calculated using vwap price during current and previous intervals) of stock i, and Ti its turnover (defined as the volume divided by the number of shares outstanding) during the same minute. One such regression is run for each minute of each day. We then average the metric across all trading days during 2010.

Index Turnover Attribution Measure: the R2 of the cross-sectional regression

iii εβ += mcapvol$

for all stocks i, where $voli is the 1-minute dollar volume for stock i, and mcapi is the market capitalization of the stock. One such regression is run for each minute of each day. We then average the metric across all trading days during 2010.

Median Pairwise Correlation: correlation among all pairs of stocks (~125,000 pairs for the S&P 500) for each minute using the prior 30 minutes of returns.

Each of these metrics is calculated for each minute of the trading day. We then average the results across all trading days during 2010. We have excluded May 6, 2010 from the analysis. In the exhibits, we normalize each metric to vary between 0 and 1 to allow a comparison across the different metrics.




Exhibit 2: S&P 500 – Evolution of Intraday Metrics over Time (Jan 2008-Nov 2010)

Log Turnover Profile Normalized Log Turnover Profile

Median Pairwise Correlation Profile Normalized Median Pairwise Correlation Profile

Log Available Size at NBBO Normalized Log Available Size





Exhibit 2 (continued): S&P 500 – Evolution of Intra day Metrics over Time (Jan 2008-Nov 2010)

Index Turnover Attribution Profile Normalized Index Turnover Attribution Profile


Trading Objectives and Investment Horizons

Market participants differ in their objective function and their trading patterns. While most participants’ strategies are based on liquidity provisioning (‘Buy low, sell high’), the period over which that liquidity is provided varies. At one extreme, we have intraday holding periods for higher-frequency, market-making type strategies. At the other extreme, we have the multi-year holding periods of some valuation based long-term stock selection strategies.

Exhibit 3 provides a schematic view of the market as a set of interacting liquidity providers. As we move up the pyramid, the liquidity provisioning/holding period timescale increases. At the top, we place ‘Buy-to-Hold’ market participants as the terminal investors. With sufficiently long liability horizons, they base their investment decisions on the income or cash flow potential of assets, rather than on expectations of a capital gain.

Market participants in the layers below the ‘Buy-to-Hold’ investors are then liquidity providers to the layers above them. They ‘Buy-to-Sell’, in this admittedly simplistic view. This liquidity provision takes two forms, which we call ‘Front-Ended’ and ‘Back-Ended’. Front-Ended liquidity provisioning is akin to market making activity, while Back-Ended liquidity provisioning can be seen as a momentum strategy. As the investment horizon shortens, we see greater Front-Ended liquidity provisioning strategies, and greater turnover / share of overall trading volume.

Exhibit 3: Investment Strategy Pyramid


For example, ‘Fundamental’ market participants have skill and a comparative advantage in evaluating the fundamentals of an asset, particularly as it relates to the rate and growth of income generated by the asset. This skill enables this class of investors to separate liquidity needs of other market participants from changes in the fundamental drivers of income generated by the asset. They therefore become natural liquidity providers in those situations where other market participants need liquidity, even though the fundamentals of the underlying asset have not changed. This liquidity provision will command a premium, which is extracted through Front-Ended strategies.

At the same time, the Fundamental market participants spend resources identifying assets that will be of interest to the ‘Buy-to-Hold’ investors in the future, based on their

Increasing Holding Period

‘Buy - To-Hold ’

HFTs / Market Making

Statistical Arbitrage

Quants

Fundamental




views of the fundamentals of the asset. Purchasing these securities is a Back-Ended form of liquidity provisioning, since they pre-position themselves in expectation of future liquidity demand – ‘buying low to sell high later’.

A similar argument can be made for other market participant types in Exhibit 3. Their investment horizon is shorter, and the means by which they identify assets that will be of interest to market participants with an investment horizon longer than their own is different, but the same principle applies.

In our view, this distinction between front-end versus back-end liquidity provision, as well as between the different market participant types, is a useful framework for understanding the interactions between different market participants, and their impact on the price discovery process. In particular, we believe that this approach can potentially be more illuminating than the common academic approach of ‘informed traders’, ‘noise traders’ and ‘market makers’.

The varying holding periods lead to different ‘alpha decays’ and hence urgency of trading. Schematically, Exhibit 4 shows the relationship between reasons for rebalancing portfolios and trading urgency. News-driven trading will have a higher urgency than valuation-based strategies where the liquidity premium is expected to accrue over a longer time period. This should be reflected in the execution strategy as well – valuation-based strategies should lead to patient trading strategies.

The greatest variability in urgency is in trades that are driven by cash flow, or by the need to fund other trades. These trades are generally driven by institutional requirements, such as constraints on the amount of cash held in a fund vehicle, or on beta exposure. They may be part of a portfolio rebalancing trade – for example, the portfolio manager decides to build a position in a new asset, and funds this position by trading a slice of the remainder of the portfolio. In such cases, the urgency of the funding trade will depend not on the expected alpha decay of these positions, but on the urgency of the trade in the new portfolio constituents4.

4 In practice, such transitions can be smoothed using other instruments. Futures contracts or ETFs are frequently used for beta and cash management. To the extent that the price impact for these instruments is lower than for the stocks to be traded, dynamically managing such a beta overlay while slowly trading the stocks to be transitioned can be advantageous for reducing price impact. Depending on the size

Exhibit 4: Alpha Decay for Different Types of Trade s


Trading Patterns and Market Dynamics

Even though the hierarchy of market participants suggests that most operate as liquidity providers over time, the day-to-day practice of actual trading follows a simpler pattern. Urgency demands in trading and concentration in order size mean that most of the time, the market makers at the bottom of the pyramid act as the counterparties in trades for all other market participants.

The interaction between the ‘instantaneous’ liquidity providers and other market participants behaving as liquidity demanders is dominated by uncertainty about future order flow. ‘Front-Ended’ market makers are willing to provide liquidity if they expect the intensity of the liquidity demand they ar interacting with to decay relatively quickly. This is the case either if the liquidity demand occurs for idiosyncratic reasons (akin to the ‘noise traders’ in the academic literature) or if the total parent order size is not too large. In those cases, the risk premium available from instantaneous liquidity provision may be attractive. In other cases, a ‘Back-Ended’ momentum strategy may be more attractive – for example, if we expect the liquidity demand from a parent order to persist for a certain length of time (or in the case of an ‘informed trader’).

of the transition, dedicated transition managers can assist in the optimal structuring of such trades.

Urg

ency

/ A

lpha

Dec

ay

“News”-driven Cash Flow / Funding

Valuation - based

Reason for Portfolio Rebalance

Hig

h Lo

w




Uncertainty about the persistence of liquidity demand and the gradual resolution of this uncertainty can account for most of the intraday patterns we showed in the first section of this paper.

During the first part of the trading day (the ‘Price Discovery’ phase), overnight macro and company specific news can lead to one-sided liquidity demand as prices adjust to the news. The likelihood that order flow during this period is driven by idiosyncratic requirements of market participants is low. Standing in front of such order flow is typically not a profitable proposition for a market maker. As a result, bid/ask spreads are wider and the available size at the best price is relatively small. Each executed order in the early part of the day provides greater information about the future direction of order flow. Prices adjust more quickly per unit of volume, which we capture in our ‘Liquidity Measure’.

As the macro and company specific news becomes incorporated into prices, the likelihood of idiosyncratic liquidity demand through institutional buy and sell orders increases. During this ‘Institutional Portfolio Rebalancing’ phase, uncertainty is centered on the private information that institutions have about the total size and direction of their stock specific liquidity demand. Given the typical large size of this liquidity demand relative to the instantaneously available size, institutions typically stretch their trading over the trading day. With each executed child order, uncertainty around the private information about stock specific liquidity demand is reduced.

Depending on the predictability of the trading pattern over time, the speed at which this information is absorbed by the market will vary. Typically, we find that the cross-sectional return dispersion – a measure of differences in stock specific liquidity demand – decreases very rapidly during the morning hours, indicating that uncertainty around the private information reduces quickly. As a result, our Liquidity Measure is much higher during this period than during the first 30 minutes of trading.

During the afternoon, yet another source of uncertainty enters the market, centered on the beta demand of investors. Market participants adjusting their beta exposure – through mutual fund and ETF flows, dynamic option hedging strategies, or other systematic beta trading strategies – drive more of the trading activity. This is reflected in our ‘Index Turnover Measure’, which shows that a greater portion of

the observed trading volume can be explained by index trading, as well as by the rising average pairwise correlation of stock returns during the afternoon. As a result, our ‘Liquidity Measure’ increases significantly during this period, indicating lower price impact per unit of volume.

One interpretation of this effect is that ‘the market’ expects less stock-specific trading during this period, because of the dominance of beta-driven trading. Any stock-specific liquidity demand is subsumed in the diversification of beta-driven trading. Selectively exploiting this market expectation by shifting idiosyncratic stock trading into the afternoon may lead to smaller impact on prices than an equivalent trade in the morning. This is predicated on the trade not being ‘too large’, which would lead other market participants to adjust their expected probabilities of trades being driven by idiosyncratic stock orders.

Macro and Micro Drivers of Intraday Trends

The relative importance of macro and micro drivers of stock level liquidity demand changes through the day. Macro drivers are significant during early morning and during the afternoon, while micro drivers are significant in morning trading but fade during the afternoon. We can observe the impact of macro drivers more directly by focusing on differences in trading patterns between an index ETF (we use the SPY as an example) and a ‘typical’ large cap stock chosen from the S&P 500 constituents.

Exhibit 5 contrasts a number of intraday metrics for the SPY ETF and a ‘typical’ large cap stock for 2010. We observe distinct differences in these metrics:

• Turnover Pattern: Turnover is skewed towards the close both for the ETF and for the representative stock. However, there is greater volume in the ETF during the early part of the day. We have argued earlier that beta-driven trading affects the underlying stocks predominantly during the afternoon. The turnover pattern for the ETF shows that during the morning, beta-driven trading still happens, but is concentrated in index products such as ETFs. The trading impact on the underlying stocks during the afternoon may be driven by the residual beta demand necessary to clear markets.




• Spread Profile: Bid/Ask spreads are widest in the morning. However, while single stock spread declines throughout the day, ETF spreads plateau after 13:00, reflecting potential unanticipated beta trades during the afternoon. The ETF shows a significant spike in spreads at 10:00, driven by macroeconomic news that get released at that time on some days.

• 1-Minute and Cumulative Dispersion Profile: Both ETFs and single stocks show a greater dispersion early in the day (and a spike at 10:00). However, while the cumulative ETF return dispersion accelerates during the last 2 hours of trading, we do not observe this on a single-stock level.

Exhibit 6 directly compares the intraday evolution of the intraday bid/ask spread and available size for the SPY and our representative single stock. This comparison shows the

liquidity providers’ uncertainty about macro-driven vs micro-driven liquidity demand. While the available size for the SPY and the single stock shows similar patterns – in particular the increase after 14:30 – the bid/ask spreads behave very differently.

For our single stock, uncertainty about stock-specific liquidity demand decreases quite rapidly after markets have opened. This is reflected in the declining spread in the first hour. Spread then drifts slowly lower until 14:30 when it starts to decline rapidly again as beta-driven trading takes over.

For the SPY, in contrast, we see a spike in spread at 10:00, driven by macroeconomic news releases on certain days. Following a decline and plateau in the spread, it then increases again after 14:00 before fading into the close. This reflects the additional uncertainty around beta-driven trading and about who will hold the overnight beta risk, which dominates the market during the afternoon session.

Exhibit 5: Intraday Trading Profile for SPY vs Sing le Stock – 2010

Intraday Trading Profile for SPY ETF Intraday Trading Profile for ‘Typical’ Large Cap St ock





Exhibit 6: Intraday Evolution of Spread and Availab le Size for SPY and Single Stock – 2010

Intraday Spread of SPY vs Stock Intraday Available Size of SPY vs Stock


In Exhibit 7 we link the behavior of the SPY and our representative stock, to highlight the changing importance of macro vs. micro risks on the stock level price discovery process. Both SPY and stock return dispersion are highest in the morning. They then decline before increasing again after 14:00. Stock return dispersion is relatively higher early in the day, as indicated by the higher residual stock return dispersion5. Notably, the increase in stock return dispersion after 14:00 is driven exclusively by increasing SPY dispersion – the residual dispersion remains flat until just before the close.

This is also shown by the increasing correlation of returns between the 1-minute stock returns and the SPY returns, as well as by the increasing percentage of stock return dispersion explained by the SPY dispersion.

We thus find that stock level behavior and price discovery is driven by macro and micro risks to varying degrees throughout the trading day. Most importantly, we find that during the afternoon, macro or beta considerations become

5 We perform the univariate pooled regression for our representative stock across all trading days

εβα ++= SPYstock retret

for every minute of the trading day. The residual dispersion is then given by

( )ρσσ ε −= 122stock

and the proportion of stock return dispersion explained by the SPY is equal to

2ρ=RSQ

dominant in determining stock level risk and volume patterns.

Exhibit 7: Intraday Trading - Macro vs. Micro Uncer tainty – 2010


Historical Changes: 2010 vs. 2005

How have the intraday dynamics of the price discovery process changed over time? In particular, is the dominance of beta trading in the afternoon a recent phenomenon? Analyzing the dynamics over time allows us to determine how adaptive trading strategies need to be.

In Exhibit 8 we compare some of the basic metrics on spread, available size, dispersion and turnover for 2010 with those in 2005. The charts on the left compare the actual values for our metrics. On the right, we show normalized values (by subtracting the daily average), to allow us to focus on changes in the intraday shape of the metrics.




The two most significant changes between 2005 and 2010 are the much lower bid/ask spreads throughout the day, as well as the much higher turnover. Looking at the normalized data, we see that the intraday pattern of these two metrics has also changed – Turnover has become significantly more back-loaded, with a greater proportion of volume now occurring in the last hour of the trading day. Meanwhile, the bid/ask spread declines much faster now in the first hour of trading, and then continues to decline throughout the trading day, while in 2005, we see a spike in spreads just before the close.

Another notable change in the intraday metrics is that trading behavior appears to have become more synchronized. In particular, we observe spikes in Spread, Dispersion, Turnover and other metrics on every full half hour during 2010. The 10:00 spikes are particularly pronounced. These patterns were much weaker during 2005.

We attribute these results to two key changes in liquidity demand. First, we see a rise in the use of systematic execution strategies. To the extent that these are anchored to fixed time windows (for example, having target trading volume ‘waves’ for each half hour of the trading day), we would expect to see synchronization effects.

Exhibit 8: Stages of the Trading Day - S&P 500 in 2005 vs 2010

Comparison of Raw Metrics Comparison of Normalized Metrics


Exhibit 9: Intraday Liquidity, Index Turnover and C orrelation Profile – 2005 vs 2010





Furthermore, we see a greater proportion of beta-driven trading during the afternoon, which leads to lower uncertainty stemming from single-stock liquidity demand during that period. This is reflected in the normalized Available Size profile, which shows a much larger increase during the last hour of trading in 2010 than it did in 2005.

Exhibit 9 shows this through our derived measures. The Median Pairwise Correlation has been higher in 2010 than in 2005 throughout the trading day, reflecting the greater macro risk contributions. More importantly, this metric continues to increase into the end of the trading day during 2010, while in 2005 we saw a decrease during the last 30 minutes. This decrease, which is still observed in most markets outside the US6, reflects unanticipated stock level trading volume into the close.

In the US, the market has become dominated by beta trading during this period, meaning that correlation no longer declines into the close. This is shown by our Index Turnover Measure. In 2010, this measure is not only higher throughout the day, but it also increases more dramatically into the close than it did in 2005.

Exhibit 10 returns to our sample large cap stock, and shows the importance of macro vs micro drivers on the price discovery process, as measured by the R-squared of our pooled regression. The importance of macro drivers has increased since 2005 – while the R-squared of the regression during 2005 was around 20-30%, this has now risen to reach almost 70% just before the close. In addition to this increase in the level of R-squared, we also note that the intraday shape of the curves has become more extreme. Since 2007, the period after 13:00 shows an almost monotonic increase in the R-squared.

6 See our companion piece covering global markets, ‘Stages of the Trading Day –

Think Globally, Act Regionally’, Morgan Stanley Quantitative and Derivative Strategies, February 2011.

Exhibit 10: Macro vs Micro Drivers of Stock Prices


Execution Strategies

Based on our analysis of the intraday market behavior, we believe that the market day can be segmented into three distinct periods – the pricing in of overnight idiosyncratic and macro news at the open and shortly thereafter, the period of institutional portfolio rebalancing that follows, and the period of beta-driven trading during the afternoon. Market participants’ objective functions can account for a significant portion of this observed segmentation, in our view.

This has implications for optimal execution strategies. Given the continual change in intraday market characteristics, we believe that execution strategies have to be reviewed and adjusted on an ongoing basis. In the current market environment, concentration in liquidity demand means that parent orders are typically far larger than the instantaneously available liquidity. This requires orders to be executed over some period, either intraday or across multiple days. Algorithmic execution strategies are commonly used to manage at least part of the parent order, in addition to more opportunistic strategies.

When to use which algorithmic trading strategies is thus of particular importance when determining the optimal execution strategy. Exhibit 11 summarizes some of the Morgan Stanley equity trading algorithms. We categorize them along two dimensions. ‘Aggressiveness’ involves a tradeoff between speed of execution and market impact – greater aggressiveness generally will mean greater market impact in a shorter time. ‘Execution Strategies’ can either follow a pre-determined schedule, or they can dynamically react to market conditions.




Exhibit 11: Morgan Stanley Algorithmic Trading Stra tegies

TWAP NIGHT VISION VWAP ARRIVAL PX

LOW NIGHTOWL

LOW ARRIVAL PX

MED PRICE REACT

NIGHT OWL MED

ARRIVAL PX HIGH

NIGHT OWL HIGH SORT

LOW

Aggressiveness HIGH

Source: Morgan Stanley Electronic Trading

The ‘alpha decay’ of trades in Exhibit 4 – the speed with which private information embedded in a trade is expected to be priced in – will play a role in determining the optimal execution strategy. Information that is expected to be priced in quickly should lead to greater urgency in trading. This accounts for the willingness of market participants to trade early in the day, despite the large bid/ask spreads and high cross-sectional dispersion of returns.

For trades driven by slower alpha decay – for example valuation-based portfolio rebalancing – we would expect to see less aggressiveness in trading and potentially greater use of schedule-based execution strategies. This can account for the market behavior during the ‘institutional rebalancing’ phase. Lastly, for trading that is driven by beta adjustments – for example due to fund flows – the trading strategy should reflect the fact that these trades do not contain any stock-specific information. This leads to the relatively low cross-sectional dispersion and tight bid/ask spreads during the afternoon, in our mind.

Market participants have synchronized on this segmentation of the trading day. In our mind, this has been driven by a number of factors. First, we have seen an increased concentration of liquidity demand. This has made execution more challenging, and has led to greater adoption of algorithmic trading strategies. Second, when market participants decide on an execution strategy that appears optimal in a given market climate, they often do not consider the impact that other market participants adopting the same execution strategy will have. This can lead to a

synchronization in trading strategies. Third, the combination of greater concentration in liquidity demand and more synchronization in trading strategies has led to more opportunities for short-term liquidity provisioning. Instantaneously available liquidity has increased as a result, as has the fragmentation in trading venues.

What does this mean for execution strategies? We believe that the alpha decay threshold between opportunistic and passive trading strategies has shifted, such that even relatively longer alpha decay trades now have to be approached from an opportunistic, liquidity seeking perspective. This view is driven by the concentration in order flow in the market, which leads to a greater divergence between market impact of a liquidity demanding strategy and that of a liquidity providing strategy.

In practice, this means that shifting the start time of a trade to later in the day can often be beneficial, in our view. To the extent that intraday liquidity demand patterns get priced in relatively early in the trading day, taking these patterns into account and opportunistically trading at a later point may lead to lower market impact for a given trade size.

The second implication for execution strategies involves the optimal release schedule of private information. In a market where liquidity demand is concentrated amongst relatively few market participants, delaying the release of private information is critical. This requires a trading strategy that is ‘non-invertible’, in the sense that the market can not easily infer overall trading intentions through individual child

BENCHMARK EXECUTION STRATEGIES

TWAP Aims to execute trades evenly over a specified time period.

VWAP Seeks to match the volume-weighted average price over a given time period.

Arrival Price Benchmarks an order against the mid-point price when it enters the market, balancing the tradeoff between market impact and execution risk.

Price React Adds a level of price sensitivity to the Arrival Price strategy by increasing an order’s participation in the market as the stock’s price moves in a favorable direction.

LIQUIDITY-SEEKING STRATEGIES

Night Vision A dark liquidity aggregator, seeking liquidity across multiple dark venues including MS POOLSM.

Night Owl Maximizes liquidity access across dark pools, ECN/MTF non-displayed limit order books and displayed markets simultaneously.

SORT A direct market access (DMA) destination that provides high-speed seamless access to all major liquidity sources, for trading exchange-listed names, ETFs and options.




orders going through the market. In many cases, schedule-based algorithms are used in such a way that private information is released sooner than necessary. We believe that a more complex, opportunistic approach to using such execution strategies can be beneficial to market impact incurred. Randomization in the placing of child orders can certainly help. However, we believe that complexity has to be introduced in the usage of algorithms themselves as well.

Trading Day Status Quo

The current market price discovery process is the result of a confluence of factors, including greater automation in execution strategies, greater concentration in orders and with it greater order sizes, and the emergence of new forms of liquidity provisioning strategies. Compared to the period prior to 2007, the market shows greater instantaneous efficiency, in the sense of equilibrating supply and demand on an intraday basis.

The question is whether this instantaneous efficiency also leads to efficiency over longer periods. In our mind, this is not always the case – trading constraints and habits can lead to greater market impact than necessary, given the size and urgency of the trade. To correct this, we believe that market participants have to increase the complexity of their execution strategies, in order to prevent the market from fully pricing in their private information too soon.

Morgan Stanley Electronic Trading and QDS have developed a number of practical recommendations for our clients on optimal execution strategies, given a set of constraints on urgency and alpha decay. Please contact your sales representative for further details.


© 2011 Morgan Stanley. All rights reserved.

Important Information and Qualifications The information provided herein was prepared by sales, trading, or other non-research personnel of one of the following: Morgan Stanley & Co. Incorporated, Morgan Stanley & Co. International PLC, Morgan Stanley MUFG Securities Co., Ltd, Morgan Stanley Capital Group Inc. and/or Morgan Stanley Asia Limited (together with their affiliates, hereinafter “Morgan Stanley”), but is not a product of Morgan Stanley's Equity Research or Fixed Income Research Departments. This communication is a marketing communication and is not a research report, though it may refer to a Morgan Stanley research report or the views of a Morgan Stanley research analyst. We are not commenting on the fundamentals of any companies mentioned. Unless indicated, all views expressed herein are the views of the author’s and may differ from or conflict with those of the Morgan Stanley Equity Research or Fixed Income Research Departments or others in the Firm. For additional information, research reports and important disclosures, see https://secure.ms.com.

Morgan Stanley is not acting as a municipal advisor and the opinions or views contained herein are not intended to be, and do not constitute, advice, including within the meaning of Section 975 of the Dodd-Frank Wall Street Reform and Consumer Protection Act.

The information provided herein has been prepared solely for informational purposes and is not an offer to buy or sell or a solicitation of an offer to buy or sell the securities or instruments mentioned or to participate in any particular trading strategy. These materials have been based upon information generally available to the public from sources believed to be reliable. No representation is given with respect to their accuracy or completeness, and they may change without notice. Morgan Stanley on its own behalf and on behalf of its affiliates disclaims any and all liability relating to these materials, including, without limitation, any express or implied representations or warranties for statements or errors contained in, or omissions from, these materials. Morgan Stanley and others associated with it may make markets or specialize in, have or may in the future enter into principal or proprietary positions (long or short) in and effect transactions in securities of companies or trading strategies mentioned or described herein and may also perform or seek to perform investment banking, brokerage or other services for those companies and may enter into transactions with them. We may at any time modify or liquidate all or a portion of such positions and we are under no obligation to contact you to disclose any such intention to modify or liquidate or any such modification or liquidation. Morgan Stanley acts as “prime broker” and lender for a number of hedge funds. As a result, Morgan Stanley may indirectly benefit from increases in investments in hedge funds.

Unless stated otherwise, the material contained herein has not been based on a consideration of any individual client circumstances and as such should not be considered to be a personal recommendation. We remind investors that these investments are subject to market risk and will fluctuate in value. The investments discussed or recommended in this communication may be unsuitable for investors depending upon their specific investment objectives and financial position. Where an investment is denominated in a currency other than the investor’s currency, changes in rates of exchange may have an adverse effect on the value, price of, or income derived from the investment. The performance data quoted represents past performance. Past performance is not indicative of future returns. No representation or warranty is made that any returns indicated will be achieved. Certain assumptions may have been made in this analysis which have resulted in any returns detailed herein. Transaction costs (such as commissions) are not included in the calculation of returns. Changes to the assumptions may have a material impact on any returns detailed. Potential investors should be aware that certain legal, accounting and tax restrictions, margin requirements, commissions and other transaction costs and changes to the assumptions set forth herein may significantly affect the economic consequences of the transactions discussed herein. The information and analyses contained herein are not intended as tax, legal or investment advice and may not be suitable for your specific circumstances. By submitting this communication to you, Morgan Stanley is not advising you to take any particular action based on the information, opinions or views contained herein, and acceptance of such document will be deemed by you acceptance of these conclusions. You should consult with your own municipal, financial, accounting and legal advisors regarding the information, opinions or views contained in this communication.

These materials may not be distributed in any jurisdiction where it is unlawful to do so. The products described in this communication may not be marketed or sold or be available for offer or sale in a number of jurisdictions where it is unlawful to do so. This publication is disseminated in Japan by Morgan Stanley MUFG Securities Co., Ltd only to those investors who are “Professional Investors” (tokutei toushika) as defined in the Financial Instrument Exchange Law of Japan and may not be redistributed to other types of investors without the prior written consent of Morgan Stanley; in Hong Kong by Morgan Stanley Asia Limited; in Singapore by Morgan Stanley Asia (Singapore) Pte. and/or Morgan Stanley Asia (Singapore) Securities Pte. Ltd., each regulated by the Monetary Authority of Singapore, which accept responsibility for its contents; in Australia by Morgan Stanley Australia Limited A.B.N. 67 003 734 576, A.F.S.L. No. 233742 and/or Morgan Stanley Australia Securities Limited A.B.N. 55 078 652 276, A.F.S.L. No. 233741, which accept responsibility for its contents; in Canada by Morgan Stanley Canada Limited, which has approved of, and has agreed to take responsibility for, the contents of this publication in Canada; in Spain by Morgan Stanley, S.V., S.A., a Morgan Stanley group company, which is supervised by the Spanish Securities Markets Commission (CNMV) and states that this document has been written and distributed in accordance with the rules of conduct applicable to financial research as established under Spanish regulations; in the United States it is directed only to institutional clients and is distributed by Morgan Stanley & Co. Incorporated, which accepts responsibility for its contents; and in the United Kingdom it is directed only to those persons who are eligible counterparties or professional clients and must not be acted on or relied upon by retail clients (each as defined in the UK Financial Services Authority's rules) and is distributed in the European Union by Morgan Stanley & Co. International plc, except as provided above. Private U.K. investors should obtain the advice of their Morgan Stanley & Co. International plc representative about the investments concerned. In Australia, this publication, and any access to it, is intended only for “wholesale clients” within the meaning of the Australian Corporations Act. The trademarks and service marks contained herein are the property of their respective owners. Third-party data providers make no warranties or representations of any kind relating to the accuracy, completeness, or timeliness of the data they provide and shall not have liability for any damages of any kind relating to such data.

HYPOTHETICAL PERFORMANCE RESULTS HAVE MANY INHERENT LIMITATIONS, SOME OF WHICH ARE DESCRIBED BELOW. NO REPRESENTATION IS BEING MADE THAT ANY ACCOUNT WILL OR IS LIKELY TO ACHIEVE PROFITS OR LOSSES SIMILAR TO THOSE SHOWN. IN FACT, THERE ARE FREQUENTLY SHARP DIFFERENCES BETWEEN HYPOTHETICAL PERFORMANCE RESULTS AND THE ACTUAL RESULTS AND THE ACTUAL RESULTS SUBSEQUENTLY ACHIEVED BY ANY PARTICULAR TRADING PROGRAM.

ONE OF THE LIMITATIONS OF HYPOTHETICAL PERFORMANCE RESULTS IS THAT THEY ARE GENERALLY PREPARED WITH THE BENEFIT OF HINDSIGHT. IN ADDITION, HYPOTHETICAL TRADING DOES NOT INVOLVE FINANCIAL RISK, AND NO HYPOTHETICAL TRADING RECORD CAN COMPLETELY ACCOUNT FOR THE IMPACT OF FINANCIAL RISK IN ACTUAL TRADING. FOR EXAMPLE, THE ABILITY TO WITHSTAND LOSSES OR TO ADHERE TO A PARTICULAR TRADING STRATEGY IN SPITE OF TRADING LOSSES ARE MATERIAL POINTS WHICH CAN ALSO ADVERSELY AFFECT ACTUAL TRADING RESULTS. THERE ARE NUMEROUS OTHER FACTORS RELATED TO THE MARKETS IN GENERAL OR TO THE IMPLEMENTATION OF ANY SPECIFIC TRADING PROGRAM WHICH CANNOT BE FULLY ACCOUNTED FOR IN THE PREPARATION OF HYPOTHETICAL PERFORMANCE RESULTS AND ALL OF WHICH CAN ADVERSELY AFFECT ACTUAL TRADING RESULTS.

Any estimates, projections or predictions (including in tabular form) given in this communication are intended to be forward-looking statements. Although Morgan Stanley believes that the expectations in such forward-looking statement are reasonable, it can give no assurance that any forward-looking statements will prove to be correct. Such estimates are subject to actual known and unknown risks, uncertainties and other factors that could cause actual results to differ materially from those projected. These forward-looking statements speak only as of the date of this communication. Morgan Stanley expressly disclaims any obligation or undertaking to update or revise any forward-looking statement contained herein to reflect any change in its expectations or any change in circumstances upon which such statement is based. Prices indicated are Morgan Stanley offer prices at the close of the date indicated. Actual transactions at these prices may not have been effected.

The trademarks and service marks contained herein are the property of their respective owners. Additional information on recommended securities discussed herein is available on request. This communication or any portion hereof, may not be reprinted, resold or redistributed without the prior written consent of Morgan Stanley.

The trading of futures or options on futures contains inherent risks. We remind investors that these investments are subject to market risk and will fluctuate in value. The investments discussed or recommended in this communication may be unsuitable for investors depending upon their specific investment objectives and financial position. Where an investment is denominated in a currency other than the investor’s currency, changes in rates of exchange may have an adverse effect on the value, price of, or income derived from the investment.

Quantitative and Derivative Strategies (“QDS”) of Morgan Stanley’s Institutional Equity Division Sales Desk have prepared this piece. This is not a product of Morgan Stanley’s Research Department and you should not regard it as a research report.

trading day

Documents