Microstructure and the Market Making Game.

Microstructure 1Microstructure 1Microstructure refers to the process or mechanism by which trades take place.

We study microstructure for several reasons:–Clarifies concept of market efficiency (How prices come to reflect information).–Dealer is basic intermediary (Understanding how they earn profits, set spreads, etc., helps us understand all intermediaries.)–Regulatory Issues.–Trading Norms (e.g., Should we trade at the open, in the third or fourth market, etc.?)–Introduction to Liquidity.

•MeasurementMeasurement

Additionally, microstructure affects the way we measure stock prices, and ignoring implications of the trading institution may lead to erroneous inference.

2 Examples:– When Issued Anomaly

After a company announces a stock split, it is often 6 weeks before the split becomes effective (i.e., the ex-day). During this period, the NYSE suggests that the specialist begin trading in the post-split priced stock (on a when-issued basis). So, for example GE is currently selling for 120. It declares a 3-for-1 split on Sept. 10, that will go into effect Oct. 21.

Measurement 2.Measurement 2.

The NYSE starts trading GE on a when-issued basis - if the stock is at 120, when-issued will be at 40. An anomaly was discovered: the when-issued stock were more expensive than the regular stock. So, here when the stock is at 120, then when-issued may be at 40 1/8 or 40 1/4.

One might look at this “anomaly” and start to conjecture on its cause. However, it is the result of the institution (or microstructure). In particular, the regular shares are being bought and sold actively, so that the average price will be in the middle of the bid-ask spread. The when-issued shares, on the other hand, are very illiquid. The only trades here are the public buying from the specialist.


So the when-issued price is always an ask quote (the top of the bid-ask spread), while the actual stock is in the middle of the spread. This is the cause of the “When-Issued Anomaly.”

• January / Small Firm Effect (?)Another anomaly (much more important and well-

known) is the January Effect. It has been noted that the smallest stocks (by market capitalization) have very high positive returns in the month of January (and essentially negative returns in the other 11 months), while the opposite is true of large firms. Furthermore, the return for small firms appears relatively too big to be explained by CAPM.


For example, the smallest 10% of NASDAQ stocks may have an average return of 40% in Januarys. However, the bid-ask spread on these small stocks may be on average 45%. This large spread may explain why the phenomenon may persist. (Although it can’t explain why we see the phenomenon.)

The point: If you ignore the institution that generates the data, you may think there’s something fishy - when there really is not.

Market Making GameMarket Making Game

Our primary tool to develop an understanding of microstructure is the market making game. In the basic market making game, there are 3 types of players:

• Informed Trader• Liquidity Trader (aka noise trader)• Dealer

There are variations on the basic game, but it is important to fully understand the situations that characterize each player type.

Informed TraderInformed Trader

We introduce the security in the game by understanding the informed trader. There is a stock whose value at the end of the period is drawn from a probability distribution. All players know the nature of this probability distribution. The informed trader learns the actual draw before the period starts. During the period the players may trade the stock with one another, and at the end of the period, all positions in the stock are marked-to-market.

Informed Trader 2Informed Trader 2

The informed trader has no trading constraints or position requirements. Within the game, her objective is simply to make as much money as possible.

In the game, there are no short-selling constraints, so from any player’s point-of-view, buying and selling are symmetric.

Liquidity TraderLiquidity Trader

In the confines of the game, the liquidity trader must trade during the period to meet a liquidity constraint, or else incur a penalty. As is the case for the other 2 types of players, the positions are marked-to-market at the end of the period.

There are no constraints on the trades that a liquidity trader can make -- the constraint applies only to the position at the end of the period.

Liquidity Trader 2Liquidity Trader 2

The end-of-period constraint (or liquidity shock) for each liquidity trader is drawn from a probability distribution, and is seen only by that trader. The distribution from which these draws are taken is the same for all liquidity traders, and the draws are taken independently.

DealerDealer

Dealers do not learn the end-of-period asset value, nor do they receive a liquidity constraint. They must enter a bid quote and an ask quote in the stock before the market can open.

The bid quote is the price at which the dealer is willing to buy the stock (in response to a market order to sell), and the ask quote is the price at which the dealer is willing to sell the stock (in response to a market order to buy).

Dealer 2Dealer 2

All trades in the game are for 1 share of the stock - so all quotes are for 1 share.

When there is more than 1 dealer in a stock, a market order is automatically routed to the inside spread - i.e., the highest bid and the lowest ask.

As with the other 2 types of players, dealers’ positions are marked-to-market at the end of the period.

The StockThe Stock

In the base case, the end-of-period value is drawn from a Normal Distribution, with a mean of $100, standard deviation of $10, min of $65, and max of $135.

As described above, all players know this information about the stock’s probability distribution.

ExampleExample

Consider a game with 1 stock, 3 dealers, 1 informed trader, and 4 liquidity traders. All 8 players are informed of the distributions:

– Stock value drawn from N(100,10) [65 , 135]– Liquidity Shocks drawn from

U [ -5 , -4, -3, -2, -1, +1, +2, +3, +4, +5]

At this point, you should characterize the distribution of aggregate liquidity shocks.

Example 2Example 2

The insider learns the end-of-period asset value is $93.

(All players understand that the insider knows the value, and that only she knows this value.)

Liquidity Trader 1 gets a draw of -2.

Liquidity Trader 2 gets a draw of +1.

Liquidity Trader 3 gets a draw of +4.

Liquidity Trader 4 gets a draw of -2.

(All players understand that each liquidity trader has such a draw, and no one else knows what it is.)

Example 3Example 3

Before the market period can start, all 3 dealers are prompted to enter a bid and ask quote (which includes the unconditional expectation of $100).

For example, Dealer 1: Bid $76, Ask: $128

Dealer 2: Bid $97.5, Ask: $101.5

Dealer 3: Bid $99, Ask: $111.

Now the clock starts ticking. All traders see all 6 quotes, and the fact that the inside spread is: $99 bid, $101.5 ask.

Dealers are free to revise quotes at any time.

Concept ChecksConcept Checks

You should be able to answer the following questions:

– What would the dealers do in the absence of liquidity traders (i.e., what would the spread be?)

– What is the distribution of aggregate liquidity shocks - and why is this important?

– How will the degree of risk aversion of the dealers affect their spread-setting and expected profits?

– What is the expected profit of insiders? (or at least the sign)

– What is the expected profit of liquidity traders? (or at least the sign)

RealityReality

The purpose of the game is to enable you to build intuition in a relatively simple setting. We will tweak the basic assumptions in this regard, as well.

In reality, there may exist an insider with knowledge of an event before the rest of the market. Different countries have different regulatory attitudes about the insider’s ability to profit personally from this information. In the US, our WSJ articles suggest that the adverse selection problem is more severe in options markets than in the stock market. (Why?)

Insider TradingInsider Trading

From The House of Morgan:[In the Roaring 20’s Morgan partners] were in an

excellent position to take advantage of insider trading, which was a common vice of the 1920s, and not illegal. Not only did Jazz Age Wall Street echo with rumors, but being in a position to plant false reports was considered a mark of financial maturity. Lax Stock Exchange rules and meager corporate reports made inside information more valuable, and investors milked their Wall Street friends for news. Inside tips didn’t guarantee success - many investors perished on Black Thursday clutching them - yet they were profitable enough to be considered a major perk of Wall Street employment. (pp. 305-6)

Reality 2Reality 2

The market described above is a dealer market - all trades involve a dealer. This is similar to NASDAQ, pre-1998, where, in aggregate, dealers have a monopoly franchise on market making (at least for small trades).

Liquidity traders may seem the most disconnected from reality, but it is simply that their motives for trading lie outside of the game’s framework.

Efficient MarketsEfficient Markets

It should be clear that whether or not markets are efficient depends on your point of view.

Does the insider ever have the opportunity to make infinite profits?

Information TheoryInformation Theory

What does the following have to do with the market making game?

In the Second World War, the Allies had broken the Japanese and German secret codes. Many American and British lives were sacrificed to prevent the Axis from learning that the Allies had broken their codes.

A direct use of intercepted and de-coded messages was the assassination of Japanese Admiral Yamamoto, in 1943. Interestingly, it was hubris that prevented the Japanese from realizing that the Allies had broken their code.

Where have the Insiders Gone?Where have the Insiders Gone?

This short article from the WSJ suggests that NYSE specialists are not worried about insiders. This is probably because the NYSE is essentially an auction market - as the article notes, only 10% of NYSE trades involve the specialist trading on his own account.

Options markets are another matter. The article notes that almost all options trades involve a dealer trading on her own account. (Of course there are competing dealers on the options markets.)

What does an option dealer do once he’s taken a position? She can (and generally does) Delta Hedge - i.e., trade in the stock to construct a portfolio whose value will not change when the stock price changes (a small amount).

Actual InstitutionsActual InstitutionsThe market making game is an example of a

pure dealer market. All trades are routed to a dealer, who is required to maintain a posted bid and ask at all times.

A close analog of this is NASDAQ prior to 1999 – for retail investors.

Odd Eighth AvoidanceOdd Eighth Avoidance

An interesting study by Bill Christie and Paul Schultz showed that in the early 1990’s, NASDAQ market makers did not use odd eighths in setting their bid and ask quotes. So, for example IBM might be trading on the NYSE with a posted spread of 80 – 1/8 (meaning 80 bid, 80.125 asked). On NASDAQ, MSFT, which is very similar to IBM would trade at 80 – ¼.

8

1

8

1

Odd Eighth Avoidance – How?Odd Eighth Avoidance – How?

In light of this, we should ask why MSFT’s management was not pressured (for example by large institutional investors) to list on the NYSE.

The reason is that such investors trade in the “third market.” An example of the third market is Instinet - an ECN (Electronic Clearing Market). In the third market, traders avoid the market makers.

So this system hurt only retail traders.

Regulatory ReformRegulatory Reform

The Justice Department, SEC, and private lawsuits forced NASDAQ to allow an open limit order book. So now, if I want to buy MSFT, when the standing quotes are 80 – 1/4 , I can submit a limit order to buy 100 shares at 80.125 (or now 80.07). My “bid” will be displayed on the NASDAQ computers.

Knight-TrimarkKnight-Trimark

Our WSJ article on this large market-making company ties-in with the BB article on trading desks. The main theme here is that monopoly access to order flow information is valuable.

NASDAQ has been transformed from a market where people make money off spreads to one where people make money off information. . . . When you have 30% on the order flow you can make some darn good guesses.

An example: ORCL is currently at 34. Knight has 13,000 orders to buy at prices between 33 and 34 (these are “hidden limit orders”) and 30,000 orders to sell between 34 and 35.

Knight-Trimark 2.Knight-Trimark 2.

In this setting, Knight may infer that there is more selling pressure than buying pressure, and the stock price is apt to fall from 34.

Problems:– Limit Orders may be withdrawn at any time.– This is information about liquidity, and not

fundamentals. A corporate announcement would change everything.

– By taking advantage of this information, Knight is putting its customers at a disadvantage (on average).

NYSE: An Auction MarketNYSE: An Auction Market

The NYSE designates a specialist – or monopoly market maker for each stock. But most volume is direct trades between traders. In a sense, the specialist must compete to make a market with all NYSE traders.

Specialist Gone BadSpecialist Gone Bad

Consider the following example from the NYSE of the special advantages that a dealer has. A particular stock opened on October 19, 1987 at 60. Throughout the day most of the trading involved the public selling - and this particular dealer doing a lot of buying. At the end of the day, the stock was at 30, and the dealer under water. Over night, many market orders to buy at the open arrived. (Of course, these buyers anticipated buying the stock around 30.) The specialist opened the stock at 50 - selling out of his inventory. This surprised the buyers, but allowed the specialist to break even on the losing trades from the previous day. By the end of the day the stock was at 30.

Bad Specialist 2.Bad Specialist 2.

The NYSE has strict price continuity rules that constrain a specialist’s ability to move the price between trades. These rules may be circumvented by halting trade in the stock, or at the open. Generally, when there is an order imbalance at open, the specialist will not open the stock until advising the crowd of the imbalance.

The specialist in this case was severely disciplined by the NYSE.

Why?