For what I understand, any market maker makes money from the simple fact that he always buys at the bid price and sells at the ask price. Bid price is naturally always inferior to ask price.
You, the market participant, always buy at the ask price and always sell at the bid price. You have the bid-ask spread at your disadvantage, the MM has got it at his advantage. The more people trade, the more they gain.
Spread betting firms are not unlike market makers in general, with an important difference. They in principle don't buy and sell financial instruments. They accept bets on both sides of the event, and they "match" the side against each other, pocketing the spread. If there was a guarantee that the two sides of the bet are always balanced the SB firm would have no capital requirements, so to speak.
SB firms have, at any given moment, on any given instrument, as a result of the bets of their clients, some money at stake short and some money at stake long (actually they have at stake only the difference between the two).
Let's imagine that on aggregate, there are at a certain moment 100 betters active on a certain instrument and they collectively have at a certain moment £140 per pip long, and £140 per pip short. The SB firm just cashes the spread and has no risk at all because they are market-neutral.
Now let's imagine a few minutes later they have £140 per pip short, and £190 per pip long. The SB firm is, on aggregate, £50 pip short (the aggregate of the punters is £50 pip long and the SB firm is the other side of the bet).
The SB firm will - according to their own risk management rules - possibly bring this imbalance "to the market" to cover their risk. They will open a position of £50 long on the instrument, in the underlying "real market", in order to find themselves again market-neutral.
They will go on monitoring the aggregate result of the bets, and keep themselves "covered" in general if the imbalance goes beyond a certain threshold.
Naturally all these market operations have a cost. The SB will probably not cover any minimum imbalance instantly, they will cover balances beyond a certain threshold, and they will act with a certain time delay, and choose to keep at home a certain level of risk.
That's why SB firms, for what I understand, do NOT welcome high frequency traders. Those players who open and close a position within a few seconds, maybe with a large stake, will make it difficult for the SB firm to bring the imbalance to the market. The SB will be playing against the punter and, basically, they don't want. They basically want to bring the imbalance to the market, and live out of the spread on the punts, which they pocket continuously (what they bring to the market is obviously normally a small part of the total of the stakes as long and short position would tend to balance themselves on aggregate).
It is on the other hand a known fact of life that the vast majority of inexperienced day-traders tend to lose money. The SB firm could treat this as a statistical "high probability bet" and decide not to bring to the market the aggregate positions of the (supposed to be) inexperienced traders knowing that, on aggregate, it is better to be on the other side of those transactions.
My point of view: if you bet £1 per pip with round-trips that take minutes, or hours, to complete, the SB firm will have nothing to lose from your behaviour. They will actually be glad when you win. Remember you pay the spread (to the SB firm) also when you win!.
The idea that the SB firm wins only when you lose is totally wrong. If you pay £500 as margin, and lose it in a month, and stop trading, even if the SB firm did not cover your punts, they will have cashed £500 and that's it. And they cannot rely for their long-term survival as a firm on the fact that there will always be a fresh supply of inexperienced traders.
If you pay £500 as margin, and you win let's say £2000 per year, and you go on betting, you can easily calculate how much the SB firm gained by summing all the spread from all your trades (good ones and bad ones) and that will be a nice amount. Remember: the SB firms pockets basically all the spread (they don't pocket only that percentage of punts that they bring to the market). They gain at each trade, whichever the result for the punter. And if you are successful you will go on making money to them. The long-term interest of the betting community and of the SB firms are, in fact, aligned.
If you play £50 per pip in rounds that last for a few seconds the SB firm will begin getting nervous. Besides, the nature of the SB business is such that the quotation often cannot be "instantly" equal to the underlying market. Sometimes the underlying market (in Forex for instance) does not exist as a unique bid-ask price couple. That will lead to price discrepancies, price uncertainties, or time delays that will be very relevant to a "scalper" and that can cause big losses to a SB firms if the punter exploits the imperfections and delays of the SB price formation process to game the system.
If the punter earns from this price discrepancy the SB firm will find a way to stop him. If the SB loses he will write some conspiracy complaint in fora like this one. Mostly, the punter should understand that a SB firm is not there to benefit the punters.
If one bets large stakes for a very short time, the place to be is the market itself, a real Direct Market Access.
If one bets small stakes for a decently long round-trip duration, then SB firms can offer tax-free profits (if any), a large number of instruments tradeable from the same account, a relative protection (from what I gather) from anomalous market price spikes due to "fat finger" mistakes, small stakes and a general convenience.
Another important difference: when you use DMA you pay a smaller spread and a fixed commission. When you use a SB firm you don't have a commission but you pay, I say this in general, a larger spread.
That means that for small stakes the SB firm can be cheaper, and for larger stakes the DMA tends to be cheaper.
Finally, some SB firms offer you "more-or-less fixed spreads" and "protected stop-losses" that might be useful for certain styles of trade and I don't think they are available with DMA (correct me if untrue).
In very short terms: it is very sensible to begin with a SB firm, relaxed strategy and small stakes. If and when you one feels he can make a living out of day trading, and/or wants to trade at high frequency, with higher stakes etc. then DMA is definitely the way to go.