People often ask me for a simple overview of what exchanges are. Many people know a bit about trading but are often uncertain of how exchanges really fit into that picture. This article aims to provide a layman's introduction into how exchanges work, what matters to them and why participants in the markets should care.

Almost by definition this will be over-simplified and not entirely accurate but it should nevertheless provide a useful overview for those who are not experts in the space.

Exchanges are Marketplaces for Members

Exchanges provide a market place for members to buy and sell financial instruments, for example shares in a company (stocks). To become a member of an exchange companies must meet certain regulatory criteria. Criteria that essentially guarantees they can meet any financial obligations that they might incur as a result of trading on the exchange.

Typically, such members either trade for themselves, trade on behalf of others or maintain accounts for clients and allow them to trade through their facilities as a broker. In general the number of members of an exchange is small, numbered in the 10s. Traders usually work on behalf of companies who are members, or are themselves clients, of a broker. Anyone who ultimately trades on an exchange through a member is generally referred to as a particpant.

Exchanges Execute Orders

Members of an exchange place orders on the exchange, a statement to buy or sell a certain volume of a particular instrument at a specific price. If orders match (cross at a price with sufficient volume) then they are executed by the exchange, which essentially means a record of the match is created and then sent for settlement and clearing. That's when the actual transaction occurs with instruments and money changing hands as appropriate.

Trades

Publishing this record of the transaction, the trade, is for historical reasons often referred to as “printing to the tape”. The “ticker tape”, that most people are familiar with and which displays the price of a stock, is essentially this record. That is, the “ticker tape” displays the last sale price of the last execution for a given stock.

Orders can be placed on an exchange to indicate a willingness to buy or sell at a certain price. It may not be possible to execute at that price because there is currently no counterparty willing to sell or buy at that same price. However, they may be willing to sell or buy at a better price for them. For example you might wish to buy 100 of XYZ at £10.00 but no one is willing to sell 100 of XYZ at £10.00. Perhaps they only want to sell 100 at £10.05.

Quotes

For any instrument that is liquid, that is, there is both buy and sell interest the quote is the best buy, and best sell, price and volume. This is also referred to as the BBO – the Best Bid and Offer where Bid means Buy and Offer means Sell. For a highly liquid stock (many buy and sell orders with good volume and many price levels) this BBO will change frequently. In the example above the best buy for XYZ might be 100 x £10.00 and the best sell 100 x £10.05. The quote can then be written as: 100 x £10.00 – XYZ – £10.05 x 100.

Exchanges publish Information – Market Data

To understand what is happening in an exchange the exchange must publish two pieces of information:

• what trades are occurring as they happen
• what quotes are available as they update

Companies such as Bloomberg listen to this data and publish it on to their clients and so on. Eventually you can see on the news what the current “price” of a stock is. In many cases the information available to ordinary people is naturally, or deliberately, delayed. Usually by 15 minutes which essentially means ordinary people only know what the price was for a given stock 15 minutes ago.

Typically a trade to quote ratio might be 1:100 which means that there are many more quotes that need to be published compared to trades. In fact in addition to publishing just quotes exchanges will also publish not just the best price and volume, to buy or sell each instrument, but all the price and volumes, for which you could buy or sell an instrument. This is called the order book of the instrument. The quote, or BBO is in fact the top of book since the BBO in fact is the top buy and top sell price and volume. For a highly liquid instrument this might be hundreds of price levels. This data changes very frequently – almost every time an order is placed in fact. As such the volume of data can be huge.

What has happened (in the past)

Consider now that trades that are published are telling you information about what has happened to an instrument. It is concrete evidence of a transaction that occurred at a certain volume and a certain price. On the one hand it tells you something definitive, but on the other hand what it tells you is about something now in the past. It does not necessarily give you much indication about what might happen in the future.

What could happen (in the future)

The quote for each instrument or even more so, the order book, tells you something about what could happen in the future. The more of this information you have the better equipped you are to understand how the market is moving and how best you might trade in it. As such this information is very valuable. If you have sufficient processing power to analyse the data coming from the exchange and combine it with other information such as news feeds you be much more predictive in your trading.

This information about what has happened and what could happen is called Market Data.

How Exchanges earn Revenue

This brings us to the topic of how exchanges make money. Exchanges make money by:

• charging a fee for each execution (trade) on the exchange
• charging for access to market data

Exchanges therefore earn more money the more transactions they process, and their market data is more valuable since it represents a larger portion of the overall market. The rate or volume of orders sent into an exchange is often referred to as order flow. It is possible to discount one of these streams to encourage more transactions (more order flow) overall since the most important aspect of any exchange is having as high a number of orders and transactions as possible.

Anything that incentivises an increase in order flow has a direct positive impact on the potential revenue of an exchange.

Exchanges also earn money from listing instruments on their markets but that is outside the scope of this guide.

Liquidity

We introduced the concept of liquidity in the previous section. In fact liquidity is not easy to define precisely but what we can say is that if we have good liquidity it is easy to find a counterparty to a trade at a reasonable price. A liquid market therefore usually has many buyers and sellers. If we have poor liquidity we will not only have difficulty finding a counterparty to a trade, but we will also be uncertain of what a fair price is for the instrument. Possibly because no-one has bought or sold the instrument recently, or partly because no-one has expressed an interest in buying or selling the instrument. The problem of determining a fair price for an instrument is called price discovery. Price discovery is usually trivial in a liquid market but hard in an illiquid market.

When we talk about liquidity itself we are usually referring to the availability of buyers and sellers to trade in a given instrument. Whether that liquidity is good or bad depends on how much of it there is. This means in some cases when you see the term liquidity you can think of “possibility to trade”.

Lit Liquidity, Dark Liquidity and Dark Pools

In addition to classifying liquidity as good or bad we also talk about lit and dark liquidity. Lit liquidity is the visible market for a given instrument. When members place orders on an exchange, and the effects of those orders are published as market data for others to see, then that liquidity is lit liquidity. The key aspect of lit liquidity is that you can “see” what is happening. This means that should someone place a very large order on one side of the market it will move the price away from it as the market attempts to rebalance itself. This is a basic feature of how markets work. Essentially when a market experiences a large imbalance it will react to address the imbalance and reach a new equilibrium, or price in this case.

As it happens however there are often cases where there are counterparties who both wish to trade large volumes and would be happy to trade at a fair market price. The fair market price in a liquid instrument is typically taken to be the midpoint between the current best buy and the best sell price. In this case those couterparties with large volumes need a way to discover each other without affecting the market at large. To cater for this need the industry has developed the concept of dark pools. A dark pool is a kind of exchange where no market data is published except when a trade occurs and even then the publication of the trade is usually delayed for a period of time depending on the size of the trade.

In such a dark pool matches between buyers and sellers is entirely based on how the order might be split into smaller orders for execution. For example typically participants will specify that they will only execute against other orders that meet a minimum size requirement. Further they may or may not allow their order to be split into smaller orders to execute at different times throughout the day. All trades are executed at the current BBO for the instrument in question. This ensures each party to the trade is given a fair price.

The advantage of dark pools is that it allows the market to execute large trades, often referred to as block trades at a fair price. It does this without causing an unnatural market imbalance, or indeed unnecessary market fluctuations that would result from the market correcting one way to cope with a large buy order and then back again to correct for a large sell order (for example). In this sense the market is less volatile. Market volatility is out of the scope for this primer except to say a less volatile market is usually a good thing.

The disadvantage of dark pools is that because there is only delayed trade data published from the pool it is impossible as an outsider to understand what is actually happening inside the pool. You have to simply trust the dark pool operator and assume they are handling your orders exactly as you asked them to. There have been a number of high profile cases where it turned out this was not the case.

The liquidity (orders to buy and sell) that exists in dark pools is referred to as dark liquidity because, while it exists, we cannot see it until a trade has actually occurred. This brings us to the second problem of dark pools – we are unsure just how much dark liquidity actually exists. Instead we must infer how much liquidity there is based on what was actually executed in the pool. In Europe the Lit to Dark liquidity ratio is approximately 10:1 in terms of value executed.

Pre- and Post-Trade Transparency

We introduced the concept of trade and quote market data, and lit and dark liquidity in the previous sections. When we consider the problem of understanding what is happening in a given market now we are really talking about the transparency of the market.

Market data such as the quote and order book provide us with pre-trade transparency, that is they give us insight into the state of the market since the last trade occurred but before the next trade happens. It is pre-trade information. In order to have good pre-trade transparency we care not just about the what orders exist in the market currently, but also whether those orders represent clear and firm intentions.

For example if we allowed anyone to place orders to buy and sell in a market and published that information for all to see we would only be offering good pre-trade transparency if the orders posted met obligations regarding whether or not they will execute if a suitable counterparty exists. If we allowed everyone to simply choose not to execute then none of the orders could be trusted. In this case we would have lots of market data but poor transparency because we could not “see through” the market data to obtain an accurate picture of the market it represented.

The publication of a trade in response to an execution that has occurred gives us insight into what has just happened in the market. In other words we can see with confidence what the price and volume was of the last trade – we have post-trade transparency. Good post-trade transparency gives us something concrete to reason about with respect to how the market is moving.

The value in post-trade transparency comes from the immediacy of its availability and our ability to correlate to previous trades and changes in order book data (what could happen next). If we delay the publication of trades to some time after they occur, or only partially record the specifics of the trade then it decreases our post-trade transparency.

When we consider lit and dark markets in terms of transparency we can say that lit markets offer both pre- and post-trade transparency. Dark markets usually offer only post-trade transparency.

The general consensus is that good transparency leads to better, fairer markets. However, as discussed there is still a need for dark markets to facilitate fair large block trading.

Market Makers

Members of an exchange can fulfil a number of roles as we discussed earlier. At a simplistic level they might be brokers, companies trading with their own money, or companies trading with someone else's money. In addition some members will be Market Makers. Market makers typically trade using their own money. What differentiates them from other members is that they enter into an agreement with the exchange to help ensure a liquid market in one or more instruments. They do that by maintaining a buy and sell order (a quote) for the instrument throughout the majority of the trading day. This means that even if the market maker would face difficulties maintaining a quote under prevalent market conditions he may still be obligated to do so.

In effect this says that if someone else wants to buy or sell in that instrument there should always be at least some liquidity (available counter-orders) to allow an execution (trade) to occur. The market maker is therefore able to earn the difference between the buy and sell price that they post. For example, in a given instrument the market maker might provide a quote of 100 to buy at £10.00 and 100 to sell at £10.01. If both sides of the quote executed the market maker would earn the difference in price – this difference in price is referred to as the spread and we say that market makers work by “earning the spread”. In this example he would earn the spread if he did indeed buy 100 at £10.00 and then sell 100 at 10.01. Of course the market maker may also be charged a fee for making those two transactions which of course may affect his profit.

Market makers are an essential ingredient to a healthy, liquid market.

In return for ensuring that a liquid market exists for a given instrument the market maker is rewarded with more favourable execution costs, possibly in the form of discounts or perhaps with a favourable pricing model, such as maker-taker. In the maker-taker model the person posting orders on the book actually earns a rebate while the counterparty who “took” the order from the book pays the fee. A problem with pricing models that reward certain trading behaviour directly is that they encourage gaming of the system in attempts to capture the more favourable pricing conditions.

With the electronification of markets and the decimalisation of prices the spread between buy and sell orders has dramatically shrunk. This means that market makers typically earn a smaller spread than they did, say, ten years ago. These days penny spreads (and even sub-penny spreads for low value stocks) are common. Fortunately electronification of markets has also made them faster meaning it is possible to execute many more trades per day than before. For market makers to earn money they need to execute as quickly as possible, that is they need to trade at a high frequency.

High Frequency Trading

It isn't just market makers that like to trade at speed. In general the more trades you can execute successfully the more money you can make. High frequency trading is certainly an important part of a market making strategy but it is not limited to market making. Being able to trade faster than your counterparties offers you the opportunity to react to market data before they can. This means you can “see” where the market is going to move before others and therefore take a better position. Setting aside the specific mechanics behind this, the concept of using information quicker than everybody else to gain a slight advantage and take a better position in the markets is nothing new. This is a cornerstone of how trading works, and has worked, for many years.

What makes so-called high frequency trading more interesting, and harder to reason about, is the speed at which it happens. Literally millions of quotes per second and 1oos of thousands of trades. The only way to operate at the speeds required in modern trading is by the use of automated trading algorithms. This means that the province of high frequency trading is essentially limited to participants with the technical sophistication to create such algorithms and the financial and technical ability to place their systems in close proximity to the actual exchanges on which they will trade. Of course this is not a whole lot different to the advantage that member firms had in the past by being able to place their traders “in the pit”.

As a reminder the purpose of market making is to provide a liquid market for trading. This is achieved primarily by ensuring there is a counterparty to trade with, should you wish to buy or sell a given instrument. However aside from the ability to buy or sell a market maker also increases value for participants by keeping the spread narrow. As we have already noted penny spreads are common today. However it is entirely possible to engage in a market making strategy which is not focused on providing a liquidity or good market value, but instead is focused purely on earning the spread.

In the case where there is no concern for the quality of the market, only in making money from the market, it can mean that strategies are adopted solely based on having a speed advantage over other participants. A problem with this is that trading at such speed has little to do with the fundamental value of the instrument and more to do with transient changes in the market. Such changes can happen over such a short space of time that they do not reflect any real change in the instrument's value. Such high speed strategies can benefit from high volatility in the markets since they can react faster than anyone else. This is a huge over-simplification though as most participants today engage in some form of high frequency trading. These days it is harder to gain advantage from speed alone and when discrepancies in price occur they are quickly normalised precisely because such high speed trading exists.

Despite this so-called high frequency traders (HFTs) have made the headlines in recent years for a variety of reasons. It should at this point be clear that that is a naturally beneficial relationship between HFTs and exchanges. Exchanges in general earn money from transactions, the more the better, and HFTs facilitate trading at higher speeds. One of the problems that occurred in the infancy of HFT was some exchanges sought ways to incentivise HFTs to send order flow to their venues in preference to others. These incentives usually were at the expense of other members on the same exchange.

For example some exchanges adopted pricing models or provided special order types (ways to submit and control how an order is executed) that favoured the strategies that HFTs used. In some cases exchanges worked directly with HFTs to design order types and pricing models that would encourage them to participate more on their exchange. This was a particular issue in the US markets where the regulatory structure made things even worse (but that is outside the scope of this guide).

The single largest event of recent times that pushed high-frequency trading into the spotlight was however the Flash Crash of May 6th 2010.

The Flash Crash and Regulation

On the 6th of May 2010 a two and a half minute flash crash wiped over a trillion dollars off the US markets before a partial recovery later the same day. This was the largest intraday fall in the history of Dow Jones (the DJIA – Dow Jones Industrial Average), a drop of almost 1000 points. This was all in the backdrop of the 2008 Financial Crisis amidst a widespread deep-seated fear for the overall stability of financial markets globally.

What made the Flash Crash of particular note was the speed at which it happened. It was so fast no-one was able to react quickly enough to do anything about it. Naturally this created a fair degree of panic and spurred a quest to find regulations that would help prevent such an event from occurring again. One of the issues identified during the flash crash was the apparent evaporation of liquidity as the market crashed (which made it crash all the faster). The emergency triggers on the many trading algorithms that were in play at the time of the crash were partly to blame. The market was moving too far from their desired positions and too quickly. To prevent further losses they shut down.

Of course what regulators would have liked to have happened was that HFTs engaged in market making would have somehow dampened the crash and prevented the chain reaction of stop order triggering that ensued instead. An extensive investigation was carried out into the Flash Crash and a number of conclusions and recommendations were reached. The upshot of the event was that a number of regulations were enacted in the US and extensive regulations in the form of MiFID II were drafted in Europe. MiFID II was not in direct reaction to the Flash Crash but it has been heavily influenced by it with extensive sections on transparency and high frequency trading.

More recently systematic spoofing — in layman's terms the practice of placing an order that you have no intention of executing, with the objective of causing a move in the market from which you may then benefit — was identified by the CTFC as a primary catalyst of the Flash Crash but regulatory approaches to preventing spoofing are incredibly hard to enforce.