This article is more than 1 year old

EU: Let's cost financial traders $400m a day, because EVIL BANKERS. Right?

Wait 'til this one hits your pension fund where it hurts

Hurrah! The European Union has decided to save us from the perils of automatic trading! Also known as High Frequency Trading (HFT) or algo trading, this is simply the practice of writing a piece of code to do the buying and selling faster than a human being can possibly do it.

We've talked about the basics of it here before at El Reg.

Here's part of what is actually going to happen:

Parliament also introduced, for the first time at EU level, rules on algorithmic trading in financial instruments. As defined by these rules, such trading takes place where a computer algorithm automatically determines individual parameters of orders, such as whether to initiate the order, the timing, price or quantity. Any investment firm engaging in it would have to have effective systems and controls in place, such as “circuit breakers” that stop the trading process if price volatility gets too high.

That's not too bad. There's any number of markets that have circuit breakers in them, stopping trading if everything becomes chaotically volatile. In some markets they close if prices move more than 10 per cent while in others they keep on trading, but the basic idea of pulling the plug when it's all happening too fast for anyone (or anything, even) to keep up, then why not have a moment or two for calm reflection?

This next is extremely dubious:

To minimize systemic risk, the algorithms used would have to be tested on venues and authorized by regulators.

The life cycle of an algo is a few weeks at present; they're in an evolutionary arms race. Does anyone at all think that the authorities are going to be able to turn around approval of a complex piece of software in that sort of time span? There's a further problem here. What, actually, are the authorities going to authorise?

There's perhaps a conceptual problem here. What people are trying to exploit is correlations. X happens and Y thus moves in price. Therefore, if we can see X is happening let's buy (or sell) a slice of Y and realise a profit. But here's the problem with authorising such tactics. No one knows why prices move as they do. We're nowhere near the level of stating that the Coke share price goes down when the price of sugar goes up. We've moved way beyond anything that we can actually determine a causal relationship with. We really are just noting correlations then trading upon them until it doesn't make money for us any more. So what the heck can anyone authorise?

That the algo has to be tested... well, yes. For, as Knight Capital found out, deploying untested code can nearly bankrupt you. Or as actually happened to them, they plugged in a new algo and thought it was playing in the sandbox – but it was actually trading on the live markets. That blunder cost them $400m and ended with Knight being merged into another firm.

In for penny, in for a £1.000

However, there's one part of this new set of EU regulations that is simply insane. They've decided to limit the tick size in order to reduce the profits that might be available to HFT traders. Tick size is the price increment: how far does a price have to move before it is indeed a price change? Do we trade shares, in sterling, at £4.25, then the next available price is £4.30? Or is the next available price going up £4.26? Or £4.255? £4.251?

And there's a closely related concept: the bid/ask, or the spread. This is best understood as the fee we investors (or our pensions) pay for the ability to use the stock market to buy and sell. You can't buy and sell the same stock at the same moment for £4.25. You might be able to buy at £4.26 and sell at £4.24, though. That tuppence difference is the spread and it's a cost to us investors. Obviously, the spread has to be at least the tick size and is usually some multiple of it.

One more fact for you: HFT is one of the things that has greatly reduced the spread in the past couple of decades. It's not the only one, but most think it has been the major contributor. For HFT brings much greater liquidity. That's the other name for just more trading washing back and forth over the system.

More liquidity means several things: one of them being that those fixed costs are paid for out of more trades, so the margin on each can be lower. Another is that more trades means more competition, thus squeezing margins. But the effect has certainly been there.

Average spreads on the New York Stock Exchange have fallen from 0.2 per cent in the mid 90s to 0.002 per cent today, a drop of two orders of magnitude. That's 0.198 per cent of the value of each trade that we're not being stung for any more. With NYSE trading volumes at $200bn or so a day, that's $400m a day that investors are saving. That's real money, even in the financial world.

So, what have our distinguished Solons over in Brussels decided? That in order to prevent HFT traders from making money they're going to artificially increase the tick size. Thereby, inevitably, increasing the size of the spread. And thus reducing, or even eliminating, the one thing that HFT does that is clearly of benefit to us mug punters. Decreasing the amount that we must pay to buy or sell stocks.

God preserve us all from politicians who are ignorant, eh? ®

More about


Send us news

Other stories you might like