The Missing Billions

James Sinclair

What is happening to the largest FX platforms? Many commentators have remarked on August’s dismal FX volumes, but that is not what I am referring to. July and August were quiet in the market as a whole. The issue is the long term trend for the largest forex platforms, EBS and Reuters. Viewed in absolute terms, the forex platforms appear relatively steady but if compared to CLS Spot volumes, we see a very different pattern. CLS is, of course, the central settlement system for FX between banks, including prime brokers. In effect, I’m using CLS as a surrogate for market volume net of internalization.

Figure 1 shows the major platforms that publish volumes in absolute terms:

Figure 2 shows the platform FX volumes added together and compared with CLS, all of them indexed back to January 2010. A yawning gap is emerging – volume is going increasingly to other platforms not on this chart:
Largest ECNs &CME vs. CLS Spot Relative to Jan 2010
Viewed in even more dramatic terms, figure 3 takes EBS, Reuters and CLS forex volumes indexed back to March ’09 and shows the difference between CLS percentage increase and EBS and Reuters’ percentage loss compared to March 2009. In effect, this represents an opportunity cost: had EBS and Reuters each kept up with CLS’ increase in FX volumes all three lines would be the same. As it is, from March ’09 to July ’13, EBS appears to have lost 77% and Reuters has lost 43% of forex volume compared to CLS. EBS reported March ’09 volumes of $147bn/day and Reuters of $120bn/day. That would imply that $165bn/day has gone somewhere else.

Cls vs. EBS & Reuters, indexed to mar 2009
The short answer is that the major platforms made a series of changes that favored removers of FX liquidity without corresponding benefit to those who post FX liquidity. In particular, the largest platforms provided faster price distribution and started publishing full depth of book. They previously showed only top of book and an indication of depth.

These were understandable moves – automated models are triggered when they see new market data and the platforms probably expected more transactions. However, these changes also mean that posters of forex liquidity are more exposed. If you are a poster, and you are off-market, the market will find out more quickly; in a thin market, anything you post will be visible more quickly and cause the market to move away from you. This is particularly true of EBS and Reuters which were used by so many banks to price their systems: if someone posted a price on EBS they could find they moved the market away not just on EBS but on many bank platforms and, in turn, on those platforms where the banks make markets such as Hotspot. Many manual and automated traders began to post on the largest platforms after they posted on other platforms.

EBS’ move to deci-pips in 2011 added to their problems. A differentiator of EBS versus single bank platforms is that you can post prices and potentially avoid paying the spread as the market trades in a narrow range. EBS’ price-time priority rules mean there is an incentive to post early, be at the top of the time stack when the market reaches your level and be taken. The problem was that, after the deci-pip move, it effectively cost one-tenth the price for someone to jump in front of a poster who had diligently waited.

Again, EBS’ move to deci-pips was understandable and indeed they did appear at top of book in aggregators more often after the move – favoring removers and giving short term benefit to the platform. However, the posters soon realized that the risk-reward of posting early was no longer as favorable. So they posted later and less. EBS partially reversed their move, going to half-pips, in 2012 but the missing liquidity did not return.

In a presentation I gave to P+L New York in 2011, I showed examples from other asset classes that forewarned of such issues.

So where did it go?
Indeed, this is the $165bn question. The answer is that it has dispersed widely. At first blush, it appears some went to the CME who had strong secular growth before these changes anyway. One platform that has benefited appears to be FXall:
FXall vs. CLS Total
I indexed against CLS total FX volume above because FXall also covers forwards and options, but the graph showing CLS FX Spot is very similar.

Currenex and Integral do not publish volumes, but I have evidence they benefited. The banks and non-bank marketmakers have differing views as to how much they can really account for the gap.

The net answer is that liquidity gravitates to platforms that display proportionately less market data and are less liable to be regarded by other platforms as reference prices. In other words, posters go to platforms that are ‘darker’, ones where they will not immediately be visible to predators and will not move the market.

Which is logical – while the largest platforms in FX were rushing into the light, the largest platforms in equities were running into the dark. Evidently the equities platforms had a reason.