Basic statistics

Forex brokerage firms servicing retail clients began to emerge in force in 1999. Retail brokers channel the trading interest of an estimated 5-6 million Forex retail traders worldwide.

Worldwide, Forex Datasource estimates that there are approximately 100 Forex broker dealers that have more than 1000 clients each. To get a sense of how much trading volume passes through the hands of Forex brokers, it is useful to look at statistics from the largest broker in the market, Forex Capital Markets (FXCM). As of Jan 2009, FXCM claims to have more than 125,000 accounts trading its platforms. Also according to official sources, the FXCM monthly trading volume is $0.5 trillion - as a reference, the foreign exchange volume that normally trades per day in the interbank market is US$ 3.2 trillion.

What a broker needs to start doing business

Before a Forex brokers opens the door for business, they have invested time and money to prepare the backoffice that will allow them to offer electronic trading of currencies to retail traders. The elements that they must have are: a trading platform, a sophisticated backoffice system, and a bridge interface to interact with participants in the interbank currency market.

The trading platform is what the retail trader experiences when they see charts, news, prices, and quotes. All of this information comes from the broker backoffice systems: price engine, trade servers, account servers, web servers, news servers, etc. The trading platform integrates all of this information in a format that, hopefully, is user-friendly and intuitive. Both the platform and the backoffice are part of the same system.

If a broker offers more than one platform, they have added complexity because they had to have one more element (not pictured in the diagram above): a front-end bridge interface between the new platform and the existing backoffice systems. On a side note, the more platforms a broker offers, the more complex will be to manage currency risk, keep the systems synchronized, and maintain systems uptime.

Getting FX price quotes.

Before prices appear in a trader platform, a Forex broker will use the API (application protocol interface) instructions that the major bank will provide it to setup a link that allows brokers to get price quotes in various currencies. These banks are major participants in the interbank market and are to the broker what we call liquidity providers.

Using the API, the Forex broker needs to build a connection between its backoffice and the backoffice of the liquidity providers. A typical broker will get prices from 1-4 banks - banks like Deutsche Bank, JP Morgan Chase, Citibank, HSBC, UBS, etc. A small Forex broker will typically have only one liquidity provider and will be at the mercy of that bank or bigger broker for whatever prices the sole liquidity provider gives it. It is useful to keep in mind that no matter what broker you choose, you will only be trading with a small portion of the Forex interbank market, because numerous other liquidity providers will not be connected to the broker of your choice.

width=628px

The Forex broker will thus use this connection to banks, or back-end bridge interface, to aggregate or bring together prices from various sources. The prices from the different banks will never be exactly the same, but price quotes among the largest banks are very close.

Not knowing if a client will buy or sell, the broker needs to be able to deliver both, the bid and the ask prices. But before the broker can show a bid/ask quote to clients, it must be able to secure one or multiple bid/ask prices that will allow it to earn a profit from the spread. Deciding what price to show requires complex algorithms inside a price engine that determines what price and spread to show for each pair given conditions set by the liquidity providers and limitations of their backoffice systems.

Example of a simple order: how trader and broker can make a profit

For example, a trader sees an opportunity to buy EURUSD. Let's say the broker will show its clients a spread of 3 pips on EURUSD, for example 1.2700-1.2703. Client submits an order request to buy at the 1.2703 price through the platform. The order is relayed by the broker to its back-end bridge. The broker gets these simultaneous tradeable prices from three liquidity providers:

Bank 1: EURUSD 1.2701-02, Bank 2: EURUSD 1.2700-01, Bank 3: EURUSD 1.2702-03.

Broker now confirms the trade at the quoted price and charges the trader $30 per $100,000. Broker then offsets the trade risk by buying at 1.2701 from Bank 2 (the lowest asking price).The reason for offsetting the trade with the interbank market is so that the broker remains indifferent to whether the client makes or loses money.

Let's further assume that when the trade will be closed, the broker sees the following prices from its liquidity providers:

Bank 1: EURUSD 1.2801-03, Bank 2: EURUSD 1.2799-1.2801, Bank 3: 1.2801-02.

The broker decides to show its client 1.2799-1.2802 at the time when the client closes the trade. The client trade was closed (sold) at 1.2799 (the bid price). Meanwhile, the broker closes its trade with the interbank market at 1.2802 (the highest bid price) from Bank 3.

Now, we do some basic math to calculate how much the trader made and how much the broker made from this trade:

Client opened the trade at 1.2703 and closed it at 1.2799, a 96 pip gain minus 3 pips in spread (for EURUSD, each pip is worth $10): 960 pips x $10 = $960, minus $30 spread = net gain of $930

Broker opened its trade at 1.2701 and closed it at 1.2802, a 101 pip gain minus $2 in fees to the banks (brokers pay banks a per million fee that in our example is $10 per million or $1 per $100,000:

101 pips x $10 = $1010, minus $960 client gain, minus $2 bank fee, plus $30 spread gain = $78

In addition to bank fees per amount transacted, brokers will usually have to pay a fee to their prime broker and another fee to other technology providers that provide the ultra-fast connection that allows the trade to be booked quickly. So the net broker gain in our example may turn out to be more like $74, still a very handsome profit and no matter if the trader wins or loses.

3 Types of Brokers

There are three types of brokers: no-dealing desk brokers, no-dealing desk ECNs, and dealing desk brokers. The example above shows the profit profile of a no dealing desk broker with multiple liquidity providers. A no-dealing desk ECN will have very similar results, except that the broker profit will be 50% smaller and instead of a $30 spread cost to clients, the all in cost (commission +smaller spread) might be something like $20 - a savings of $10 per $100,000 traded.

In the case of a dealing desk broker, the calculation of profit/loss for a broker is different and more complex. Typically, a dealing desk broker will manage the risk of a pool of many trades, not just the profit or loss of one trade or one account. The dealing desk will try to make money managing the net position of all longs (buys) and shorts (sells) in any currency pair. A dealing desk is much more profitable for a broker, but it also exposes the firm to wild swings in profit/loss depending on whether the dealing desk bet its net position wisely. In some ways, the dealing desk is similar to one very large Forex trading account, and a team of internal traders actively manage the outcome of this large account.

In a dealing desk model, the broker can still offset a trade with the interbank market, but it does so selectively. Let's say a trader is consistently profitable. The dealing desk broker has the option of putting him in a different trade server that gets automatic execution and all trades are offset (sent) to the interbank market. If so, the client is happy and the broker is able to still make money on the spread. The other alternative for a dealing desk is to make life difficult for a profitable trader by providing slow and poor execution until the client leaves. This second option used to be more common in the early years of retail Forex and much less common now days.

By isolating a pool of traders that is consistently losing money, the broker can earn a much more handsome profit. If 100 consistently losing traders deposit $10,000 each, after 2-3 months, the broker will have earned in spread and profit potentially $1,000,000. Because the dealing desk broker is the final counterparty (or risk taker) for the client's trade, the loss for the trader represents the profit for the broker. Although not all dealing desk brokers want to see clients lose all their money, there is no denying that a losing client enhances broker profitability. This conflict of interest has made brokers with a dealing desk earn a bad reputation.

IMPORTANT: The prices quoted by banks will typically have a duration. This means that they are valid for the broker if executed within 1-2 seconds or even fractions of a second.

Why price re-quotes occur?

When a trader gets a message that the price has changed, he or she will probably assume that a broker has put him or her on manual trade execution. Although this is possible, it used to be much more common a few years ago when it was easier to scam traders. The most likely reason for this annoying problem is explained below.

A slow transfer of prices from bank to end-user will cause the broker to show stale prices on the platform. This will be apparent when a trader requests a trade, and the broker responds that it can't offer the quoted price because the price has changed. If this problem happens during low liquidity times, it is a sign of low speed of transfer from bank to client and back. If the trader has low speed of connection to the internet, this could also cause price requote problems.

If the price requote happens after a news announcement, then it is probably because liquidity providers have shortened the trade response time for the few minutes after the announcement. Neither banks or brokers have the desire to be responsible to traders for a given price for very long. In a fast moving market, holding on a price means assuming currency risk for big sums of money. This takes us to the other problem of slow price transfer: latency risk.

What is latency risk?

In simple terms, latency risk is the risk for the broker that it has accepted a client order before being able to secure a price that guarantees a profit. Some brokers offer guaranteed execution (what you see is what you get, or WYSIWYG). In order to not assume currency risk caused by latency, these brokers have to have ultra fast transfer times and/or charge a wider spread as a buffer to earn a profit from the spread.

Latency risk is not a risk just to brokers. If a broker does not offer a guaranteed execution and a trader requests a given trade (say a Sell USDJPY at 92.99), price latency in the broker order transfer line could expose the trader to substantial losses if the price is moving fast. Instead of filling the order at USDJPY 92.99, the broker typically has the ability to fill orders at the best available price. Instead of 92.99, the order could be filled at 93.40 or any price that the broker claims that it has received from its liquidity providers immediately after the client order was received. But there is an additional problem complicating the picture of execution during volatile periods. We explain that next.

Spread widening and trading during the news

When a news announcement is released, it has the potential for impacting currency prices dramatically, particularly if the release is much higher or lower than market expectations. This increased volatility is evidenced by a widening of the spread seen on trading platforms. The EURUSD may go from a 2 pip spread to a 5 pip spread, for example. Obviously, the larger the spread, the bigger will be a trader's to put on a trade. But it is useful to understand why the spread widens.

The spread widens because at a particular time, there is a volatile mix:

Imbalance of longs and shorts + bottleneck of orders + uncertainty in the market

The imbalance of longs and shorts refers to the fact that in an instant, what was a more or less balanced market with 1000 sellers (shorts) and 950 buyers (longs) may turn to a completely imbalanced picture: 50 shorts and 1900 longs. Everybody wants to buy and not many people are on the opposite side. Imagine a stampede on the supermarket where you can buy everything at 50% off as long as you go through one of the three cashiers in less than 2 minutes. Everybody would get as much as possible and would start to form a line, which leads us to the next factor: bottleneck of orders.

A broker accumulates orders, just like the cashier in our simple analogy. It stands as a match maker between the prices that it sees coming from the liquidity providers and the client orders that have arrived first. The fact is that during news announcements, there are almost inevitable delays during the first few minutes while the queue of orders in the system are processed on a first-come, first served basis. An order that is executed in 1 second during regular periods could be executed in 10 to 20 seconds during a volatile period, and at a very different price than requested.

The third element for spread widening is that currency markets take a few moments to assimilate all the good or bad of a news announcement. During this brief moment, there can be wild price fluctuations because of uncertainty. Trading during news announcement is extremely risky and an invitation to higher trading costs and unexpected outcomes; definitely only for the brave at heart.

Why was my stop loss filled away from my stated price?

During this period of volatility, prices will not necessarily move in a linear fashion 1.2303, 1.2304, 1.2305. A broker may see in less than one second quotes skip prices like this 1.2303, 1.2315, 1.2335, 1.2369. If a trader has a stop loss at 1.2340 and it was executed at 1.2369, it is because 1.2340 or 1.2341 did not trade at all during the time when his stop loss became active. He was given the first available price given his stop loss condition.

The backoffice systems - a reflection of the broker

As we have explored in this document, a Forex broker has numerous technical considerations to manage efficiently. Our point is that it takes four attributes to provide a professional trading experience to clients: stable/attractive technology solution, more than adequate capitalization, qualified personnel, and responsive management.

The broker cannot afford to be sloppy or cavalier with the trades it is responsible for, yet they are sometimes. More importantly, traders should avoid brokers that can't be transparent and forthright about their policies and technology, as well as brokers that show serious shortcomings in the four attributes described above.

We hope that this document has educated you on the general structure of Forex brokers and how traders and brokers can make money in Forex. If you have found it useful, we encourage you to tell others about the ForexDatasource.com Blog and Broker pages. We also encourage you to be a part of the Forex Datasource Trader Concierge Program. In this program, sophisticated traders receive one-on-one support to make informed broker decisions and timely alerts to protect your capital. Find out more at ForexDatasource.com.