With all the talk of new and exciting FX trading models, there is often the discussion of which model is preferable for hedge funds. Does it make sense to talk in such sweeping generalisations? Given that the term ‘hedge fund’ can mean a multitude of structures and strategies, any answer is nebulous at best. But let’s look at the landscape of hedge funds strategies and FX trading technologies to see why that is the case.
Hedge funds are mostly defined by their strategy. A short and non-exhaustive list of strategies includes global macro, market neutral, active investor, statistical arbitrage, and perhaps even technology arbitrage (more on that one later). Sometimes, trading activity is the implementation of an investment strategy. At other times, trading is the investment strategy.
Technology has been at the heart of price improvement and operational efficiency in the eFX revolution. Such technology goes back well over ten years. Arguably though, industrial strength eFX solutions for investment managers started to appear around 1996. These were single bank solutions that dealt with a very real problem of how to eliminate the very lengthy faxes that accompanied a phone trade. These faxes contained the detailed breakouts of the aggregate amounts agreed on the phone and were manually typed in to the bank’s risk system. One of these faxes could take a sales person off the desk (and thus not generating revenue) for the better part of three or four hours. With the arrival of the single bank systems – still very much used and very much valued today -an investment manager was able to aggregate all his breakouts electronically and transmit them to the bank for efficient pricing. Banks could examine the aggregate of the breakouts and price on net positions through dollar legs or another nominated currency. The efficiency gains were such that clients started to demand this type of electronic trading with all their counterparties. In 2000, we started to see the first of these systems in the market with multi-bank capabilities.
As people saw the potential of electronic FX, venues began to appear which appealed to additional market segments such as corporates. These were typically bank branded, timeout-based automated price feeds on single currency pairs for a specific value date.
Up to this point, the technology mostly mimicked existing business practices, but that did not last long. Trading models and venues arose which aggregated banks’ pricing streams and commingled them into one price stream showing depth sometimes without showing the source of the rate. Central limit books were offered against those prices. Anonymous portals with embedded prime brokerage allowed participants to blur the distinction between price maker and price taker. Of course, let us not forget all the interest now surrounding algorithms in FX. Finally, and perhaps most interestingly, APIs continue to arrive into these systems allowing automated access to price information and execution capabilities.
It would be a mistake to assume that each new iteration of technology to the FX market was an improvement that benefited all participants. Each iteration did however appeal to an underserved segment or even created a whole new segment of market participants. We can now confidently say that anyone who wants to participate in the electronic FX market can now do so with little cost to entry.
But back to the original question of is there a ‘best’ model for hedge funds. That question itself is ambiguous. A more precise question may be whether there is a best model for a given hedge fund strategy. Any answer gets back to the previous theme of whether the purpose of FX trading is the implementation of an investment strategy or whether FX trading itself is the investment strategy. Also, any answer will depend on how a hedge fund views its relationship with its liquidity providers.
In a global macro strategy setting, the investment idea is placing bets based on shifting monetary and fiscal policies across countries. The implementation of that idea involves trading in equity, fixed income, and derivative instruments in those countries. This trading style will usually result in currency needs going in different directions across the funds under management. Whenever I see trading that involves allocations across many currencies, I think portfolio trading. Portfolio trading allows for the efficiencies of netting across currencies and even across different currency pairs. This less obvious netting avoids the payment of the bid/ask spread, resulting in better execution.
In market neutral strategies, the investment idea is to take long and short positions either within an industry or sector (traditional long/short) or even within a specific company (such as convertible arbitrage). The implementation is through trading specific securities in the same (perhaps foreign) market. Any currency needs are generated by purchases of securities in foreign markets. Usually a single or only a few currencies are involved. If these amounts are natural blocks of a single currency pair to the same date or at most a couple of dates forward, then dealing on streaming price or an anonymous venue may be the best currency method. The decision will rely on the relationship between the hedge fund and the price provider. I will explore the importance of this decision later on.
Active investor strategies involve taking concentrated positions in a single company to force management to add value in a specific way. Like market neutral strategies, the FX requirements (if any) are usually very concentrated in a single pair.
The investment idea with statistical arbitrage is that the laws of financial physics eventually reassert themselves. If long established statistical relationships wander too far from the norm, gravity will pull them back. This is a wide mandate with nearly countless implementations depending on what exactly is being looked at as the statistical relationship. The foreign exchange requirements to implement a statistical arbitrage idea can vary from blotter based to trading single blocks. So if the FX requirements end up being multiple currency pairs across multiple value dates for multiple funds, then blotter trading could offer distinct advantages. If the trading comes down to a single currency pair with no allocations, a simple streaming quote may do.
We can safely say that technology arbitrage exists solely because of the existence of electronic FX venues. As the name suggests, technology arbitrage’s investment idea is to take advantage of the embedded weaknesses of a specific technology or trading for profit. The implementations of this idea (because trading is the investment idea) go from the mundane to the quite clever.
As venues and trading models have been created, the distinction of price maker and price taker has blurred in some portals. This blurring along with APIs into these various systems have allowed for a number of trading strategies. Typically, any one strategy is short lived until someone catches on. But there always seems to be another weakness to be exploited. The most common exploit has been venue latency arbitrage. This involves simultaneously buying/selling on one venue and selling/buying on another with a known stale price sometimes even with the same counterparty on the other side! This is known as ‘sniping’ and for the most part has been eliminated, although with the use of anonymous portals some form of this still exists.
Another technology arbitrage strategy which I have come across recently is what one person calls ‘flicker pricing.’ This involves making a price on a portal via an API and updating that price faster than the physical screen can display it. These updates occur at a high frequency each second. Pity the poor individual on the other side who thinks he is getting one price, but often ends up getting another. The differential may not be significant on its own, but taken across the user base the pips add up quickly.
Clearly, trading strategies like sniping and flicker pricing do not make the users of those strategies particularly popular. However, one could make the argument that they are doing a service by pointing out inefficiencies in specific technologies that need to be addressed. Regardless, this brings up one final question about models and which to use. What do hedge funds want out of their relationship with liquidity providers in the FX marketplace?
For their part, banks are starting to take their relationships with their clients much more seriously. In fact, defining who is a client is a new focus by some banks. If someone deals with the bank in a branded way (via the telephone, a bank’s own portal, a multibank venue where the bank’s price is clearly identified) then the person is a client. However, if someone deals with the bank though an anonymous venue (you never really know who is on the other side) or an opaque one (you know only after the trade is done), that person is a counterparty. The distinction is very real.
A hedge fund who is a bank’s counterparty can expect at most a relationship (or non-relationship as the case maybe) that is dictated by market conditions. During periods of extreme market stress or when an internal statistical model has gone awry, you cannot always expect your counterparty to answer the phone. A hedge fund who is a bank’s client can expect trust and service. Trust in terms of how larger deals are handled (or even accepted) by a bank. Service in terms of the research, market colour, guaranteed pricing, and technology offered by the bank.
In the end, what model or models a hedge fund chooses to use really depends on the underlying strategy of the fund and how it views itself in relation to the market in general and its liquidity providers in specific. The term ‘hedge fund’ is as much a catch all phrase these days as ‘alternative investment’ was previously. Just as there are numerous ways to squeeze alpha out of the market, there will be numerous ways for the FX market to accommodate these strategies. There is no one FX trading model to rule them all nor should there be.