A contract for difference is where the counterparties agree a
par price for a
publicly traded good, typically
shares in a
company, or an
index and one party agrees to pay the other if the price moves one way, the other to pay out if the price moves the other. The money paid is the difference between par and the
quoted price of the good on the
exchange at a specified date, or when one party calls for the contract to be executed, this latter being the more common.
Commonly, the "long" counterparty (the one betting prices will rise) is charged a "financing cost" for taking out a notional loan to purchase the security. In this case, the CFD is equivalent to trading on margin (trading with borrowed money).
Who uses them?
These arrangements allow risk-hungry
speculators to take
short or
long positions without having to pay the
transaction costs of actually buying a
security, which for tax reasons may be considerable. They are popular with UK consumers for the reasons given.
Typically one of the counterparties will be a company that specialises in contracts for difference, essentially a bookmaker. As these products are unlimited-liability gambles, the consumer will normally be expected to have an account with their counterparty, and will be subject to margin calls.
What are they good for?
Not a lot, as you really shouldn't
borrow money to gamble unless you really know what you are doing. If you really do know what you are doing, they can provide a cheap way to trade, and a cheap and easy way to take a short position. Unless you are wrong, in which case they can be very
expensive indeed. If you really want to do weird things with your payoff curves, you could combine these with a
straddle position (you buy a
put and a
call on the same security, or in this case you could just buy call if you are the short CFD counterparty, or a put if the long counterparty), thus limiting your losses - with a straddle you make money if the price of the
underlying security diverges greatly from that at the time of purchase, so if you are seriously wrong about the movement of the security, your straddle will finance part of the loss. If you were right in your gamble, both the CFD and the straddle will pay off. Or maybe your CFD pays off, but the straddle doesn't, or the CFD pays off, but less than the cost of your straddle.
CFDs, being private contracts have very little intrinsic effect on
price formation, as the value of CFDs traded is not known publicly, and does not necessarily cause any
trading in the
underlying security. If the "bookmaking" party buys and sells the underlying to cover their part of the liability, in effect executing the "consumer"-party's trade, then the CFD has the same effect on the market as if the "consumer" were trading through a normal broker.