A Contract For Differences is an agreement between two parties to exchange money based on the behaviour of a market asset. The two parties can be two investors, with a broker sitting in the middle who has matched the trade, and as a stakeholder; or the two parties can be one investor and the broker.

Suppose John Doe wants to put $10,000 into XYZ Corp: instead of buying 1,000 XYZ shares at $10, he makes a contract with Henry Boot, who wants to short XYZ Corp (via a broker) under which Doe will pay Boot the amount by which the shares go down, and Boot will pay Doe the amount by which the shares go up. In a matched transaction like that, the broker wouldn't need to hold any underlying stock; when the transaction is between one investor and a broker, it is usual for the broker to buy the underlying stock. If the number of transactions handled by a broker (or a network of brokers) becomes large enough, then the brokers' 'book' may get close to being balanced, reducing the need to buy underlying stock, thus reducing the broker's financing costs and hence the commission he needs to charge an investor.

Usually, a broker will allow trading on a 5:1 margin for CFDs, and positions are marked to market daily. The investor can specify a term in advance, or not, as he chooses. In a 2-way matched bargain, if one party wants out before the other, the broker then takes on the risk, and hedges it or not as he chooses.

CFDs are like individual stock futures contracts, but they are more flexible. What are the advantages?

• No stamp duty;
• Spreads tend to be lower than on the underlying shares - when there is a 2-way matched bargain, there doesn't need to be a spread at all, just an agreed strike-price;
• Speed and ease of execution, because there are no custody problems or share certificates, paper or otherwise;
• 24-on-7 trading.

Problems include what to do about dividends, the usual dangers of margin trading, and likely lack of liquidity in a crash when you most need it.

CFDs have been around for professional investors and institutions for some time - they are a kind of sophisticated stock borrowing - but have only just started being available for retail investors, and no doubt it will take a while before the market for CFDs has enough depth to be really useful.

But, imagine:

ECNs are rapidly joining together in groupings to offer after-hours trading in a wide range of stocks (and now, bonds) from different markets across the world.

At the same time, the 'legacy' stock exchanges are falling over themselves in a headlong rush to create seamless trading across exchanges and time zones.

Are they too late?

What happens if the ECNs merge with the commodity and currency trading brokers who offer CFDs? Who needs the legacy exchanges and their brokers?

There is a comparison with the horse-racing industry. On-course betting makes up only a tiny proportion of the daily turnover of betting shops and major international bookies, but until very recently, the starting price for each horse was set simply according to money arriving to bookies on the course, who were therefore vital to the whole process. Bookies were always able to take some account of off-course money, but it is only recently that communication systems have become good enough for the whole market to be reflected in the starting price. As a result, on-course bookies have become a picturesque irrelevance.

We need the legacy exchanges for a while longer to set the prices according to which electronic trading is conducted - but not much longer. Once the ECNs have joined together enough to create an international pool of liquidity which is larger than the pool available to the old system, and once instruments like CFDs mean that the market has become independent of stock, it will be goodbye.

Alongside dematerialisation of the exchanges will presumably also come dematerialisation of company financing: instead of selling their shares to the public, the price discovery mechanism of the CFD market will 'rate' companies and they will be able to borrow (or not) according to their ratings. What is the stock exchange but a price discovery mechanism alongside a casino? Once the Internet has replaced both functions, there will be no need for the silk hats and that silly hammer any more.