haha no - haven't taken a punt on bitcoins... I still reckon you should find a way to short them at some point though
Size you're trading doesn't matter too much - costs can still be high - I wouldn't be opening a particularly big account, wouldn't be scalping but I also wouldn't be taking directional punts rather I'd need to be able to execute orders across two or more products... so really need to be able to join the queue and wait for a fill on at least one leg or have resting limit orders further up and down the ladder. bund/bobl/schatz in the morning maybe US treasuries in the evening... pnl per round turn is going to be less than the value a tick - they're way way too efficient and so adding an extra spread which *has* to be crossed would make it pretty much unfeasible to even attempt to profit.
Spread betting is fine for the odd punt... as you do have a whole variety of markets to chose from - if you're going to be betting on some stock one week an index then next, maybe some commodities etc.. then great. If you're systematically trying to 'trade' a particular product even over a few days its still very dependent on your average pnl per round turn or per bet... going back to the FTSE example say you were trading a similar size in GBP there as you had on yesterday in the S&P/ES - assume 10 lots for example perhaps, say you placed 10 bets across the course of a month... that's still £800 a month extra in costs or £9,600 across the year (and that is being generous assuming execution of market orders, stop orders etc.. are going to be the same and the spread isn't going to get skewed against you). For that to be worth while you'd have to have otherwise had a capital gains tax liability of greater than those costs of £9600 - assume a CGT rate of 28% on amounts above £10,600... a profit of £44,885 would leave you with a CGT liability of £9600 and a net Pnl of £35,285
So at a spread betting firm @£100 a point - over 120 trades across a year you'd therefore need to have made greater than £35,285 for the tax advantages of spread betting to outweigh the extra costs - this isn't necessarily unfeasible. So in that scenario it might be worth while for some people such as yourself- for most people though it likely isn't.
The majority of clients of the firm you're using trade at £3-£5 a point so the tax benefits are unlikely to apply... but they still have to put up with the higher costs (then again they're trading at a size less than the tick size of 1 lot). I guess the other scenario where its useful is where you're unable to fund an account to trade equities or futures etc..
Obviously decrease the frequency of the trades/bets while increasing the amount of pnl per trade and the costs become less significant... increase the frequency of trades/bets and you get to the scalping scenario from earlier @10 lots you'd be looking at £4000 a day in costs to overcome there so basically impossible/blow the account up in a matter of a few days. Then again there are finance charges to contend with too when increasing the time frame.
So yeah I do agree that they have some use - basically for people who are taking a few directional punts over a few days which - presumably what you did yesterday with that S&P short etc.. the tax advantage is there if you're very good at that - I don't think I'd be good enough personally which is why I'm not looking at forecasting/taking directional bets. A lot of people who do that sort of thing tend to have fairly hefty swings in their pnl too and the small sample size of trades mean its going to take a while to really establish if any positive pnl can be attributed to a particular edge or whether its just the result of variance/noise.