Using FX Volume to Predict Market Moves
Posted by Colin Lambert. Last updated: August 11, 2021
The daily volume of FX trading helps predict returns on currency trading the next day, says a study co-authored by Lucio Sarno, professor of finance at Cambridge Judge Business School. The research, Learning from volume: Asymmetric information in the foreign exchange market, is soon to be published in the Review of Financial Studies journal, and as well as predicting returns over the next day, FX trade volume can also reveal asymmetric information among FX market participants.
“When volume is abnormally low for a given currency pair, we typically observe a return reversal over the following day,” says the study. Such reversal weakens and moves toward a next-day continuation of the same direction of returns when volume is particularly high.
The research also finds a high level of information asymmetry between FX market participants, with such asymmetry being independent of the volume, volatility and liquidity of currencies traded.
“Does volume contain predictive information about future exchange rate returns? How is FX trading volume related to asymmetric information about the drivers of currency returns? Can currency investors employ the information embedded within volume in designing their asset allocation strategies? Until now, answers to these questions have remained tantalisingly unobtainable,” the study says.
The authors say they developed a simple model tied to demand for foreign bonds (which are risk-free in local currency) that tests the relationship between FX volume and future currency returns. It then tests the model’s predictions based on data from CLS Group for 31 currency pairs over six years.
Under the model, daily currency returns are driven by the interaction of the previous day’s return and unexpected volume (deviation of volume from a 21-day moving average),’ the authors say in a release, adding the study finds the results to be similar across various currency pairs. This indicates that the information asymmetry does not depend on average levels of volume or the other factors, they say.
“The finding is important for industry regulators interested in the consequences of moving from the decentralised, dealer-driven foreign exchange market to a more centralised system,” says a policy-focused summary of the research written by the authors on the Centre for Economic Policy Research’s VoxEU.
“Indeed, the results contrast starkly with prior evidence from centralised markets. US equity market studies, for example, have found that the analogue of the interaction coefficient is, for most individual stocks, negative.”
The study also finds that this coefficient is positive and highly statistically significant when estimated using spot and forward volume, but shows little or no impact when estimated with swap volume – suggesting that market participants provide more private information using spot and forward exchange rates.
Is Asymmetric Information a Problem?
Putting aside the question as to whether regulators are actually seeking to move the FX market to a centralised model – something that is likely to prove very difficult given the number of jurisdictions involved – it would be interest see the results of the same analysis conducted in the first decade of this century.
Clearly market behaviour has changed with market structure and as such, when a trend starts, it tends to be quick, therefore more traders jump on the trade – hence why an increase in volume would be a signal. When markets are characterised by longer term trends, the top level results would no doubt still be the same, but there would, in theory at least, be less of a rush to jump into the position – everything might develop a little slower.
Effectively what the analysis seems to be saying is, market reversals rarely happen after high volume days, which is interesting, because anecdotally, a lot of traders talk mean reversion in FX markets much more often than they did previously. There is perhaps a threshold of volume at which the theory gains strength?
On the question of asymmetric information, this should hardly come as a surprise, asymmetric information largely exists in any transaction. The seller may have details of supply, the buyer of demand. In foreign exchange market terms the information skew is largely towards the liquidity provider, especially at the big internalisers, however there are times when the buyer has the advantage, especially if they are executing a larger order in parcels or sweeping the market.
The role of the risk warehouser in FX is vitally important to the buy side, and any move to strip the IP that is their information away leaves real economy hedgers at the mercy of anyone who wants to be top of book in the smallest amount possible.
This does not mean, however, that the more symmetric equity markets offer the best model, and as the authors point out in their blog post, asymmetricity exists in equity markets as well. In these markets, it probably lies with the aggressor in a trade as, presumably, they are the only ones who know the bigger picture in terms of volume to be executed, however a market maker, especially at the fast end of the spectrum, will probably have the ability to see moves earlier and cancel and reprice.
The suggestion that a centralised FX market is a target also misses one of the core facets of foreign exchange – the broad range of users. Whilst alpha seekers and fast market makers may like the centralised (but still fragmented) model, the real customers of the FX market – those using it to hedge exposures or make cross border payments would probably find it a cumbersome, onerous and unnecessary process.
Firms in the real economy typically want to execute their FX trades in the most frictionless manner possible, with a reasonable amount of best execution. Typically these trades are larger than normal, hence the need for risk absorbers or internalisers and therefore the totally transparent, information symmetric world, of the centralised market is the last place they would want to execute their trades.
Too often people think of the FX market as a multi-trillion dollar beast, which it is, but they ignore the breakdown within that. Yes, executing a spot 100 million hedge in EUR/USD can be done in a CLOB (there will still probably be slippage), but try doing the same in anything other than maybe half a dozen currencies.
There are simply too many currency pairs required for hedgers to retain a healthy central limit order book that can accommodate larger trades without market impact. Trading with a risk absorber or internaliser in that pair can reduce slippage, and that is why these firms still prefer to execute their hedges away from the public eye (and that is before they start thinking about cost of connectivity, collateral and other regulatory costs).
There is enough high quality information available from independent sources to ensure that a firm executing in the FX market is getting a fair price. The role of the risk warehouser in FX is vitally important to these firms, and many of these warehousers have potential market direction and access to internal liquidity pools as key components of their IP. Any move to strip that away leaves the real economy hedgers at the mercy of anyone who wants to be top of book in the smallest amount possible and then use the information they glean (legally) from that. To go back to the equities comparison, there is a reason so much volume is executed by certain firms off exchange and in dark pools.