Is the party over?

What’s gone wrong? Did the forex market change in some basic way? The most obvious answer is many new market participants came on the scene. In 1995, Parker Global had 13 managers in its universe (10 systematic and three discretionary), whereas today, there are 90 forex managers (59 systematic and 31 discretionary).

Other tracking services (such as Barclay Group, www.barclaygrp.com) cover a wider universe (106 names), although with similar results: The “currency trader” group is down 1.7 percent on the year as of the end of July (see “Managed Money”). Other institutional investment managers include exchange traded funds (ETFs), notably Ryder Investment’s Euro Currency Trust, which attracted $600 million since it was started last December.

Plenty of forex managers don’t report to anyone except their clients, and many commodity funds allocate some capital to currencies even if they are not pure-cur rency funds. Overall, trading volume in the spot forex market grew 57 percent to $1.9 trillion from 2001 to 2004, according to the Bank for International Settlements (www.bis.org/publ/rpfx05t.pdf), and the amount reported by the Chicago Mercantile Exchange for futures keeps hitting new record highs — $96 billion last Dec. 12, for example (www.cme.com/about/press/cn/05-162FXRecord16828.html).

It’s impossible to blame all the newcomers for losses without figuring out how they are ruining things for the rest of us. After all, a rise in the number of market participants can only mean a rise in liquidity, and that’s always a good thing. Maybe they are making the market less trended?

The trendedness of the forex market is one of its key attractions, along with its returns being uncorrelated with the returns on stocks and bonds. However, judging trendedness is tricky — or at least, statisticians argue about it. One trend measure is linear regression, which is the line that minimizes the distance between itself and every data point in a series. If the slope of the linear regression is positive, the trend is upward. The higher the number, the steeper the slope and the stronger the trend. A slope of zero means no trend (or, to be more accurate, a sideways trend).

The band on either side of the linear regression is called the standard error band. It measures the variability within the trendedness. The standard error bands are similar to Bollinger Bands except they show the variation away from the straight-line linear regression instead of a moving average.

Using the slope of the linear regression as the measure of trend, the first band shows a sharp uptrend lasting from March 13, 2005 to May 19, 2005. The slope is steep, measuring 0.00859. The second band (through late August) is much less steep (0.00046), and clearly shows less trend and less profit opportunity — in fact, price fails to top the early highs in this band, so you can’t reasonably expect the trend will deliver a gain at all, depending on where you got in.

As noted, a slope of zero happens very rarely, but it does happen. It occurred between Oct. 20, 2000 and June 7, 2002. However, if you were trading the euro using trend following techniques, you would not be able to complain that the influx of new traders was making the market less trended. The yen over a longer time frame shows a similar outcome. Don’t concern yourself with the actual numbers associated with the slope; they vary by whether the data used is daily, weekly, or monthly. What’s important is whether trends are becoming flatter. But in the yen, the 2005 trend was actually steeper than the two trends from the August 1992 to August 1998 period.