One of the things I plan on doing in the book I’m looking to develop about real trader performance is to see if there is any difference in returns based on the currencies people trade. Consider this blog post a bit of a teaser of what’s to come in that regard.
I’ve got just shy of 4 million trades done by a group of traders from all over the world through a number of different brokers between July 2008 and May 2013. After trimming out the most extreme 0.5% of them as being potentially erroneous, when I tabulate things I find that on average they lose 0.00487% per trade. This is looking only at the change in the exchange rate, so it doesn’t factor position size in at all. For you stat geeks out there, the T-test on a hypothesized mean of 0 produces a t-value of -24.3664, so clearly statistically significant.
Here’s a look at the distribution of those returns. The green line indicates a normal distribution. These trades follow the fairly classic pattern in the markets of being highly clustered around zero with long tails.
To put thing in terms which some may find easier to understand, a -0.00487% return would equal about 0.6 pips in EUR/USD if it was trading at 1.2, and about 0.5 pips in USD/JPY if that rate was at 100. In other words, we’re talking a very slight thing here.
In fact, that average loss is less than what you would expect given the usual spreads. I also looked at returns if you exclude an estimated bid/ask spread. My estimates were taken during the London/NY overlap, so in a high liquidity period. That means they should have been at about their lowest. Unfortunately, there was no way for me to actually figure out what the real spread would have been for each trade. I think my estimates at least are reasonable, though. When I add the spread loss back into the trades they average a gain of 0.0098%. The spread obviously makes a big difference.
Now, lets have a look at how the collective group of traders did trading the various major currencies using the same kind of comparison. The table below provides the figures. Not surprisingly, the majority of trades in the sample involved the dollar, with the euro not too far behind.
|Currency||Number of Trades||Avg. Return||Avg. Return ex-Spread|
What’s quite interesting here is that even before I account for the spread cost, the average returns for trades in the pound and the Canadian dollar are positive. That means trades in those currencies are actually beating the spread. (Again for the stats minded, the p-values for all the above are 0.00 with the exception of average return for GBP where it comes in at 0.0032).
These are really rough numbers, though. They shouldn’t be taken as a strong suggesting that you focus your trading on the CAD and the GBP, and perhaps avoid the JPY. We need to account for things like variation in individual trader skill, time of year, and things like that to really get a proper understanding. That’s all stuff I’ll be doing for the book.
I’ll try not to bore you with too much econometric analysis! 🙂
P.S. I do find it interesting that the two most active currencies, USD and EUR, have by far the lowest average return ex-spread. That would seem to say something about the efficiency level of those markets.