Joel Hasbrouck (NYU) has written a concise teaching note that serves as a primer to securities markets trading protocols. The document is ~100 pages and gives a 10,000 ft. view of rules for U.S securities markets. Here are some points from the teaching note:

  • Informed trading works like a tax on uninformed traders

  • A sequential trade model to hypothesize the behavior of informed, uninformed and passive traders

  • Increased demand in any market, from candy bars to cars, is also generally associated with a price increase. So why is order impact in security markets remarkable? A puzzle arises in connection with the relative size of the order and the price impact. For example, a firm might have 10 Million shares of stock outstanding. At a price of, say, $20 per share, the firm’s equity market capitalization is $200 Million. A 1,000 share marketable buy order might move the market share price upwards by $0.01. The total value of the order is about $20,000. Yet the market capitalization of the firm has increased by 10 Million x $0.01 = $100,000, about five times the total value of the order. It is as if an over-eager car buyer, accepting the dealer’s high initial “list price” offer, caused the price of that model to increase for everyone else in the world.

  • Why bids and offers are not generally refreshed at the same level, once acted upon? Toy model to explain the behavior

  • Base-rate fallacy : psychological tendency to “underweight the prior”

  • How does an Intermarket sweep order ( ISO) get executed? What sort of protection is offered by RegNMS? Are invisible order at the top of order book given protection? These questions are answered via detailed examples

  • Amongst all the venues, trading volume is highest for “NASD ADF”. “ADF” stands for “Alternative Display Facility”. This system provides a trade reporting channel for venues that aren’t exchanges. By most accounts, the largest contributors to ADF-reported volume are “dark pools”.

  • Financial markets are sometimes compared to casinos. There are good arguments against this view. Financial markets are not, in the long run, zero-sum games. (In a zero-sum game, there’s a loser for every winner.) Financial markets perform important functions in allocating capital across time and different uses, and in risk management. These functions can create wealth, not simply redistribute it.

  • Some of this consolidation has encountered resistance from regulators concerned about competitiveness. In 2011, a contemplated merger of the NASDAQ OMX and NYSE Euronext was effectively blocked by the US Department of Justice. A planned combination of the NYSE Euronext and DeutscheBourse (the principal German exchange) was similarly derailed.

  • Workings of a double auction market

  • Bid-shielding : When bidding opens on the sterling silver tea set, a bidder enters a price of $10, and then (via an accomplice or a separate online identity) bids $10,000. As the $10,000 establishes the high bid, no lower bids will be accepted. Immediately before the deadline, the $10,000 bid is cancelled, leaving the $10 bid the winner

  • The Globex limit order market for US futures is an electronic platform, on which trades are generally reported promptly and automatically. It recently came to widespread attention, however, that the public trade report slightly lagged the confirmation messages that were transmitted to the parties involved in the trade. A limit order seller, for example, would know that his ask had been hit slightly before the rest of the market. This was quickly recognized as a potential problem and corrected.

  • Markets that employ randomized clearing times include: the London Stock Exchange , the Tel Aviv Stock Exchange, Tel Aviv Stock Exchange, Euronext Amsterdam ,Euronext Lisbon , and Xetra. In the U.S., however, neither the NYSE nor NASDAQ use randomized clearing times. Instead, they attempt to stabilize the price determination process by restricting the information available to auction participants and also by restricting order entry and modification immediately prior to the clearing Randomizing the clearing time in an auction. This link describes how it happens on NSE.

  • Some observers have argued for wider use of single-price auctions. There has been a steady but modest flow of auction IPOs, mostly for smaller firms. But even after the Google auction established the viability of the mechanism, traditional book building continues to dominate.

  • Some markets have adopted a two-stage auction procedure that clears the market for CDSs and the bonds simultaneously

  • The process of simultaneously representing customer orders and trading on one’s own behalf is called dual trading. In all markets it is highly regulated

  • If a customer and market maker are both bidding 100, the customer’s order must be filled in full before the market maker can buy. This is true even if the market maker’s bid has time priority. At the New York and American Stock Exchanges, this was historically a long-standing rule. Formerly, in some US equity markets, a dealer had no obligation to display a customer bid, even if it bettered the best displayed bid in the market. The dealer could even trade through the customer’s bid. This was changed by the SEC’s Order Handling Rules in the mid-1990s.

  • Who are the defacto market makers ? The competition between the (human) market makers and off-floor customers reached a tipping point with the introduction of Reg NMS (2005). Reg NMS mandates trade-through protection of bids and offers. This is subject to certain restrictions. It only applies to the top of the book (a market center’s best bid and offer). An additional restriction, though, concerns access. For a bid or offer to be protected, it must be immediately accessible for execution. The bids and offers of non-automated markets or market participants are not covered. This elevated the relative status of customers who invested in the fastest technology. Their bids and offers have large supplanted those of the traditional market makers. These customers have become the new “de facto” market makers.

  • A designated market maker has a formal affiliation with the sponsoring exchange. The relationship is regulated and monitored, with the intent of holding the market maker to a higher standard. De facto market makers have no formal affiliation or obligations. They generally supply liquidity, but this provision is opportunistic. There is no penalty for withdrawal in a volatile market. Nor are there prohibitions against destabilizing trades.

  • Generally, trades in quote driven markets are not publicly reported. For investors who are accustomed to the comprehensive last-sale reporting available in most equity markets, this may come as a surprise. There is no consolidated feed, for example, that reports trades in foreign exchange or US government bonds

  • “Dark trade” is not an unreported trade. The darkness reflects the fact that neither the buyer nor the seller is posting a v isible bid or ask. Post-trade, the execution is reported like any other.

  • Dark executions typically arise from one of following mechanisms:

  • A hidden (undisplayed) limit order in a limit order market (like NASDAQ’s ISLAND) that also handles visible orders

  • A NASDAQ market maker trades against a customer order at the NBBO at a time when the MM’s own quotes are behind the NBBO.

  • The trade occurs in a crossing network or dark pool that posts no quotes of its own, but matches buyers and sellers at prices determined by the NBBO

  • Trades occurring in crossing networks are sometimes described as “zero impact”. When the trade is reported, it can’t be determined whether the aggressor was a buyer or seller. Nothing is shown prior to execution. If there is no execution, nothing is shown.

  • After extensive debate Canadian regulators adopted restrictive measures on dark trading. Generally, large dark executions (above $100,000) are permitted at or within the NBBO. Smaller orders can be executed in the dark only if there is price improvement relative to the visible bid or offer.

  • Pipeline darkpool scandal

  • Entire books have been written about stock price forecasting, and it is probably a safe bet that the diversity of techniques and paucity of reliable results are unmatched by any other field of scientific endeavor.

  • BATS markets offers stop orders, pegged orders, discretionary orders, iceberg orders and many more exotic order types

  • Temporary and Permanent impact models based on intraday and daily data – can be considered as toy models, just to give on some starting point

  • The tick size is the price of time priority.

  • Order splitting strategies – VWAP, TWAP, participation based, order splitting with pre-trade benchmarks

  • Bertsimas and Lo’s paper relating to optimal control explained using simple math

  • Maker/taker pricing on NASDAQ

  • Some exchanges offer inverted maker/taker pricing. This might seem to go against competitive logic, but actually isn’t.

  • Some retail investors believe that the brokers send their orders to exchanges where they interact in some central fashion with all the other buyers and sellers, large and small. Certainly at one point in living memory, this was for many stocks the reality. The SEC Concept Release states, “A review of the order routing disclosures required by Rule 606 of Regulation NMS of eight broker-dealers with significant retail customer accounts reveals that nearly 100% of their customer market orders are routed to OTC market makers,”

  • Payment for order flow has been a hot topic of debate amongst regulators, exchanges, market participants

  • A consistent thread running through all SEC market regulation (including Reg NMS) has been balancing order and market competition. This is also known as the fragmentation vs. consolidation debate.

  • There are four main parts of RegNMS

  • Order protection rule: The automatic execution condition for protections was far-reaching. It made a distinction between “fast” and “slow” markets. It forced floor markets (notably the NYSE) to quickly become electronic.

  • Access rule: For a market to have its quotes protected under the order protection rule, anyone must be able to execute against those quotes quickly and inexpensively. The market must give everyone equal access.

  • Subpenny pricing rule: Congress’s Common Cents Pricing Act of 1997 established the penny as the price increment in US stock market. The rule, however, only banned sub-penny quoting - and not sub-penny trading - and as such the problem of sub-pennying still exists.

  • Market data rules: BBO and trade reports should be provided at low cost, addressed “tape shredding”

  • When Reg NMS was being debated, the modern era of high-frequency trading was just beginning. It passed notice that an exchange might want to discriminate among its market data customers on the basis of speed.