- WTF Is With The Yen?!

An earthquake, a tsunami, and a nuclear reactor meltdown all strike Japan one after the other. Thousands of innocent civilians are dead; there is an industrial disaster on a globally measurable scale; and there is a nuclear fallout risk for the entire country (maybe the whole region). There are thousands left without food water or shelter while the temperature falls below freezing. It’s a humanitarian crisis of unprecedented proportions. So what happens in the markets? The currency of Japan rallies dramatically.

This isn’t irrational – it’s counter rational. In economic terms it’s like diving on the hand grenade and then pulling out the pin. So what can be the cause of this utterly unexplained phenomenon? The talking heads are all saying its ‘repatriation’, which is in effect saying that it’s the unique character of the Japanese people that has made markets act as they have. The jury is still out, but initial data seems to indicate that this isn’t so. So instead the ‘explain it away’ caucus of analysts and economists are saying that it’s the unwind of the carry trade. But with the Japanese economic outlook from this disaster likely to pressure lending rates lower for longer and to lower the value of the yen, this doesn’t make a lot of sense either.

The explanation that makes the most sense to me is that this is being driven, like most unexplained things in the markets these days, by high frequency activity - in this case, in the currency markets.

Let’s think about risk management for a minute. For a traditional trader, the risk of a trade can be describes as the typical change in price of an asset over the holding period of the position. The way he limits his risk is by placing hard boundaries around those variables. Typically he’ll set a maximum loss as a percentage of the value of the trade and when the market crosses that threshold, he’ll exit the trade – whatever else may be happening.

But high frequency trading doesn’t always work that way. With high frequency trading the percentage return is often left as a floating variable, and it’s the duration of the trade which has a defined limit placed on it. So to use the prior example, rather than placing a 3% stop loss on the position they will allow the percentage to float, and instead place a hard boundary on the trade duration... closing any position after 3 seconds for example.

Since the maximum holding period is quite short, the risk is effectively managed for each individual actor. But when you have multiple participants all acting with the same risk constraints, they can end up trading consecutively. And that can move the markets dramatically in what seems like totally irrational (or at least arbitrary) way. The reason for this is that the outlook of the market over the next three seconds may or may not be influenced in a substantial way by its longer term outlook. Over such a short time frame there may be other issues which are more likely to have an impact.

One of them is inefficient trade flow and that can be generated by stop loss trades. When a stop loss is executed, the position is closed regardless of the effect it has on liquidity or pricing. High frequency trading system can detect this, and they will rush to exploit it. So by setting off the stop loss of a longer term trader, high frequency trading systems become more likely to drive the market further in that direction, setting off the next stop loss, creating more inefficient trade flow – and so on, and so on and so on. In a very short period of time this can impact markets in surprising ways – not unlike the behavior we’re seeing in the Yen.

For the high frequency trader this is only minimally important. They only look at the expected behavior of an asset over a very short period of time. But for the long term trader, the result is a cataclysm. While consecutive stop loss limits are struck, money is transferred from traditional managers with longer time horizons, to high frequency traders at a terrifying rate. And over time, longer term managers react to this phenomenon by lowering their average position sizes. But this has a minimizing effect on their winning trades as well. So when measured across the entire industry it has the effect of depressing cumulative returns.

This matches what we've seen in the hedge fund industry, and it makes the most sense to me.

What looks like wildly irrational market activity over the long term is actually a number of high frequency players acting consecutively, and each looking forward in time over the very short term. A good analogy for this would be to imagine driving a car while being able to see only a few feet in front of the bumper. So long as there is no major obstacle encountered, you can end up going a long way off the road before you realize anything meaningful has happened. And if everyone else is doing the same, before you know it the whole traffic pattern has veered off through fields and pastures – everywhere else except the road.

But so long as you’re trading a high frequency system, then you probably don’t care. You’re only worried about the ground you’re going to cover over the next very short holding period. Pavement, grass, gravel, it’s all the same to you. Whether the traffic pattern is in it’s expected (and arguably the most rational) place is irrelevant. And if an obstacle appears you’ll simple change direction – or change back. You’re totally agnostic over the long term.

I’ve been a defender of high frequency trading because it adds liquidity within the bounds of ‘typical’ market behavior, and can make markets much more liquid and efficient. But this admittedly one of the holes in my argument. It’s really all about tail risk. It seems to me like HF trading is turning 2 standard deviation events (something unexpected – but not unprecedented) into 5 standard deviation events (something so unlikely it strains the imagination) before anyone can do anything about it, or in fact before most people even really notice.

This too seems to match most recent market behavior, where every day there seems to be some new cataclysm. It's speculation on my part, I grant you - but it's worth thinking about further.