Andy Webb
Andy Webb

Business as usual: FX trading platforms help to achieve more with less

The current economic situation is unique for a combination of reasons, but one of the most prominent is the combination of a commercial slowdown with an exceptional decline in many bank balance sheets. While this has numerous implications, the effect on global FX trading volumes is particularly conspicuous. Andy Webb explains how FX trading platforms have a duty in such an environment to assist participants in achieving more with less.

First Published: e-Forex Magazine 36 / e-FX Industry Report / July, 2009

Liquidity in any market is to some extent a self-fulfilling prophecy, based upon collective ‘I will if you will’ confidence. As a result, a decline caused by fundamental factors can easily be exacerbated as participants lose faith and move to the sidelines. In the worst case, this behaviour accelerates into a downward spiral resulting in a drastic reduction in liquidity. 

Happily, the largest FX trading platforms have proved relatively resistant to this doomsday scenario. Nevertheless, it is incumbent on these platforms to do all in their power to facilitate the efficient execution of business at current liquidity levels. Apart from fulfilling the immediate need, any initiatives that aid traders in this respect by implication also boost future confidence and inclination to participate. 

Reading the numbers

The need for such action is apparent; the statistics speak for themselves. Figure 1 is drawn from a recent report by David Poole of ClientKnowledge and shows the general fall in buyside FX activity over the past six months based upon interviews with more than two hundred high volume FX traders. Figure 2 gives an insight into the sellside perspective, indicating a sharp decline in Tokyo market broker activity since October 2008. A similar situation prevails in FX futures and options; Figure 3 shows a commensurate decline in activity on the CME. While both FX spot and derivatives have shown signs of recovery more recently, overall activity is still well below typical levels for 2008.

Bank and factors

The reasons for this are easy enough to discern. The demise or retrenchment or absorption of AIG, Bear Stearns, Merrill, Lehman and ABN Amro have removed a large slice of liquidity from the market. As individuals these organisations were very significant participants, but collectively the disappearance of their proprietary trading, client business and market making activity has had a major impact. Those banks that have survived with state assistance have similarly reduced their activity levels, with some already publicly undertaking to withdraw either completely or partially from proprietary trading activities, including FX. 

Figure 1:Buyside Activity (Source: ClientKnowledge)

Figure 1:Buyside Activity (Source: ClientKnowledge)

Figure 2:Sellside Activity (Source: Bank of Japan)

Figure 2:Sellside Activity (Source: Bank of Japan)

Figure 3:CME Average Daily Volume - FX Products (Source: CME)

Figure 3:CME Average Daily Volume - FX Products (Source: CME)

Economic factors

While many have been quick to predict that regulatory capital will be switched away from proprietary trading to support flow business activities with lower capital requirements, the FX liquidity portents here are not particularly encouraging either. 

According to the 2007 BIS Triennial Central Bank Survey of foreign exchange and derivatives market activity, non-financial customers accounted for 17% of all FX market turnover. A significant portion of this activity is driven by global trade, which has been adversely affected by the global economic uncertainty. Figures 4 and 5 show the cumulative changes in import and export flows for OECD countries between August 2008 and February 2009. These figures are gloomy enough, but the situation is even worse as regards OECD trade with non-member countries; for example, Brazil’s import trade with OECD countries fell by nearly USD57bn and the Russian Federation’s export business declined by USD67bn over the same period.

Figure 4:OECD Cumulative Import Flow Changes Aug 2008 - Feb 2009 

Figure 4:OECD Cumulative Import Flow Changes Aug 2008 - Feb 2009 

Figure 5:OECD Cumulative Export Flow Changes Aug 2008 - Feb 2009

Figure 5:OECD Cumulative Export Flow Changes Aug 2008 - Feb 2009

Obviously if trade flows are down, the FX transactions associated with them are also down. As a result, banks are seeing a decline in their flow business as well as their proprietary activity in FX. 

Contiguous and liquid markets

Combine all the bank, regulatory and economic factors and the prognosis for FX liquidity is clearly less than optimistic. Furthermore, there is little that FX trading platforms can do directly to influence these factors and increase underlying demand. 

However, there is a great deal such platforms can and should do to enhance the ‘liquidity experience’ of FX market participants through consultation and innovation. One general point about the current liquidity conditions is that reduced total activity tends to highlight order traffic in a manner that can make the market seem less liquid than it actually is. For instance, many markets have a high level of interrupt and resubmit activity, which is less obvious (and of less concern to participants) when consummated trading volumes are high. However, when the level of completed trades declines on the back of the factors mentioned above, this interrupt and resubmit activity becomes more apparent. The quantity of orders being removed from the order book and subsequently amended and resubmitted can give the impression of ‘jitter’ when overall trading volumes are lower. An analogy is to compare the appearance of an early silent black and white movie with the latest computer generated imagery. 

This impression of reduced contiguity can have the effect of reducing participants’ confidence and their willingness to post liquidity. As mentioned earlier, liquidity is based upon collective confidence, so anything that can be done to boost the level of contiguous price action is likely to increase this confidence and potentially additional liquidity as well.

Keeping your place in the queue

One way of increasing the sense of contiguity is to reduce interrupt and resubmit activity. A common task for many traders is reducing the size of their quote or order. This could be for any one of a variety of reasons, such as the end client changing their instruction. Seeing that this involves pulling the current order and resubmitting a smaller one, the net effect is to reduce the sense of contiguity, as well as causing the trader to lose his/her place in the order book queue. For example:

  • The market is currently trading at 50-51 and a trader submits an offer for 10m at 52 on behalf of a corporate treasury client
  • The corporate treasury client becomes aware of a natural partial hedge between its business units and so decides that the order needs to be reduced to 5m
  • The trader pulls original offer and resubmits a new offer for 5m at 52
  • The new offer is now behind all other offers at 52 submitted by other traders since the original offer for 10m was pulled

The market as a whole sees an order being withdrawn before being resubmitted in smaller size, creating the perception of increased ‘jitter’ and short term loss of liquidity. This will be particularly acute when interrupt and resubmit activity takes place near the current traded price, as other traders will be attempting to hit bids/lift offers that suddenly disappear  (especially since this scenario is likely to be repeated across the entire market multiple times per trading session). 

One way to remedy this situation to the benefit of both individual traders, as well as the market as a whole, is to allow the reduction of existing order size without the loss of order book priority. In the immediate term, this approach makes any reduction in liquidity appear less severe and when extrapolated across the whole market will hugely reduce interrupt and resubmit activity. In effect, more liquidity remains in the order book for longer, thus boosting the sense of contiguity. In addition, the individual trader’s workload is reduced, as only two actions (place and reduce) are now required, rather than three (place, interrupt and resubmit) and the chances of efficient execution are improved as position is not lost in the order book queue.

Regulatory factors

The changing regulatory situation for banks has been a further negative for liquidity. In the UK, the Turner Review implies that increases in capital requirements for proprietary trading activity by commercial banks are inevitable1. (The Turner Review does make a distinction between outright speculative prop trading and market making activity conducted in direct support of customer service, but at present any concessions relating to the latter activity will be of limited consolation – see “Economic factors” below). Lord Turner was even more forthright in his testimony to the House of Commons Treasury Select Committee in February in which he anticipated that tighter regulatory controls would result in a “significant downsizing of proprietary trading.”

Other countries have proposed even more draconian measures in their sabre rattling against what are perceived as the perils of proprietary trading. For instance, Paul Volcker, former Fed chairman and key presidential adviser on the banking industry, was talking in January of this year of a complete ban on certain proprietary trading activities and regulating prop trading volumes in general. 

A similar tone comes from the EU, where its February report from the High Level Group on Financial Supervision in the EU made thirty-one recommendations, the seventh of which was to: 

“Introduce appropriate capital requirements on banks owning or operating a hedge fund or being otherwise engaged in significant proprietary trading and to closely monitor them.”

Anyone pondering the meaning of “appropriate” in that recommendation need only look elsewhere in the report: “If banks engage in proprietary activities for a significant part of their total activities, much higher capital requirements will be needed…”

While these mooted regulatory changes may not all become reality, the prospect of them is driving a mood of severe conservatism in many banks. In its ‘Global Credit Market Strategic Outlook 2009’ released in late 2008, Barclays predicted bank proprietary trading “will effectively cease” as regulatory capital would be transferred to other activities. This prognosis is beginning to look more than a little prescient, as a variety of banks – including many that had not suffered especially severely in the sub-prime blow up – have already announced their partial or total withdrawal from prop trading. 

Granted, much of the brouhaha over proprietary trading has focused on less liquid credit-related products, but this is clearly spilling over into other areas, so that an increasing number of FX trading babies are unfortunately being jettisoned along with their bath water.

Prime broking is another area that regulators have earmarked for tighter supervision and higher regulatory capital allocations. The knock-on implication for FX liquidity is apparent; as banks deleverage on the funding they are prepared to offer hedge fund clients, those clients are obliged to reduce their trading activity. Again, this doesn’t just apply to FX, but will inevitably have negative implications for FX participation.

Ergonomics

Reducing the number of actions required to achieve a trading objective in this manner increases trader productivity. In the process, this has the potential to boost liquidity; trades that might once have been passed due to workload become feasible. 

Ergonomics are obviously an important part of this process, particularly in the context of common trading behaviour. For instance, a popular strategy among many traders is to use bids and offers to cross the market at the very top of the order book. Leaving a bid in the market at the desired price is a more effective way of getting done than placing a buy order, because the phenomenon of stochastic resonance makes it probable that someone will give the bid price and the deal will complete. By contrast, trying to hit the bid/lift the offer runs the risk that the market moves before the trader can complete.

However, if a trader has a bid in the market at (say) 70 when the market is 70-71 he/she will need to amend his/her bid to 71 in order to cross the market. That involves changing the order and then sending it, which requires multiple keystrokes. Therefore any mechanism that allows traders to partially automate this sort of common task adds value. For example, a single keystroke combination that allows traders to amend and submit their bid/offer in one hit would accomplish this. At a stroke, their productivity increases (along with their chances of being filled) with this ability to automatically bid at the offer.  

The corollary to this is any mechanism that allows traders to combine orders into a single trade message; again the objective is to improve their workflow and by implication also the liquidity and tradability of the market. A short term directional trade with a profit target is an obvious example. The traditional approach with these is to submit an order, await confirmation it has been filled, and then submit a separate exit order. In a volatile market where the trader is using jobbing type trading strategies, this process can significantly reduce the likelihood of success. There is always the risk that by the time the trader submits the final profit target order the market will already have traded at that desired exit price and moved back away from there, potentially into loss. 

By contrast, if the trading platform can offer an automated contingent order facility, this process can be condensed into a single order message, which encapsulate both entry and exit prices. As a result the trader has the certainty that as soon as the initial order is partly or fully completed, the appropriately sized exit order will automatically be made live without any action required on their part.

Price flexibility

A relatively common scenario is for customers to give a trader an order to work along with an element of execution discretion. The downside to this in a conventional market is that the trader can end up having to interrupt and resubmit potentially multiple orders in order to get the trade done. Furthermore, there is the risk that the time taken to interrupt and resubmit can result in increased slippage if the market keeps moving in the interim. In a worst case scenario the trader ends up chasing the market with a string of unsuccessful orders and exceeding the limits of his/her discretion before being able to get a fill. 

In order to avoid this, some traders will risk submitting the new order before interrupting its predecessor. While this may potentially reduce slippage if the market is moving away from the trade, it has the considerable downside that if the market moves back towards the trade before the original order is interrupted a double fill will result.

One possible solution is for the trading platform to provide functionality that allows any discretion the trader has to be encapsulated in a single order. For instance, the order might show a 55 bid, but have a hidden component, only visible to the matching engine, indicating that the trader had two pips discretion and would therefore be prepared to pay up to 57. If the market is currently 56-58, then as soon as an improved offer was made the trader’s order would be filled at the improved price. 

Such a mechanism would reduce unnecessary order traffic and improve the chances of the order being filled close to the desired price without the risk of the market moving out of reach or double filling. Yet again, this improves the overall appearance and feel of the market to all participants in terms of uninterrupted liquidity.

Conclusion

In these challenging economic and market conditions the key to making the best of the available liquidity is to provide a more deterministic trading experience; ultimately traders need to have the sense that the outcome of attempting trades is predictable and reliable. A critical part of achieving that is to deliver more contiguous markets with a higher level of consummated trades in relation to order traffic. 

A realistic indicator of how successful any FX trading platform is in delivering this against a backdrop of fundamentally weaker demand is the response from its participants in the aftermath of a substantive news event. For example, if a major economic report comes in a long way from predicted levels, is this greeted by a surge in trading activity or just a flicker? 

As remarked earlier, liquidity is very much a confidence game. Minimising unnecessary interruption of orders and providing more conterminous liquidity gives traders the encouragement to commit. As a result, quotes at the top of the order book become ‘stickier’ and all market participants benefit. Trading venues that cannot deliver this will struggle to avoid a downward spiral, those that do will have achieved a virtuous circle.

Visibility and certainty

In addition to the level of interrupt and resubmit activity in a market, another factor that strongly influences perceptions of its liquidity and tradability is the minimum period of time that a quote can be displayed. This is particularly important in markets where there are both machine and human participants. The challenge here lies in finding an equitable balance between the two. 

A pragmatic line of action might be to link this to average human hand/eye reaction times. Most academic research in this field seems to suggest that for tests such as catching falling rulers or clicking a mouse in response to a colour change, response times are typically in the 200 to 250 millisecond range. Current less formal research appears to corroborate this2, with a median reaction time of 215 milliseconds generated from a sample of more than 1.5 million.

Given that experienced human traders are likely to have slightly faster reaction times than the average, but also have to react to more complex numeric information, a benchmark of around 250 milliseconds for a minimum quote display time appears reasonable.

A related matter is the actual tradability of prices displayed on the screen. An important issue here is the point at which an order is deemed interrupted. A trader might interrupt at his/her keyboard but their quote might still be visible at that time on another trader’s screen. The second trader might attempt to trade against the quote but not be matched as the quote was deemed already interrupted. The logical route here is to make the determinant of interruption the midpoint – i.e. the matching engine. If the matching engine is about to propose a match then rationally it should also be the point of determination for any order interruptions. This provides greater certainty and consistency of trading experience, reduced deal times and maximised trading opportunities.

2009 – Banks go back to basics

By Jake Smith, ClientKnowledge

  • Chasing flow from any source is no longer a successful FX strategy 
  • Banks are focussing on providing services to key clients and increasing efficiency 
  • Market liquidity and volumes are starting to drop in 2009

Over the past nine months the FX market has undergone significant change in terms of liquidity becoming more sporadic, driven to a large extent by the credit crisis.  In the past, market makers distributed prices across multiple channels in the understanding that there was a correlation between increased FX flow and trading profitability.  Deep liquidity was available to almost all clients whether they chose to trade on the phone, on single dealer, or multibank platforms.  This has been a successful strategy for the banks, but it will come as no surprise that market makers are no longer willing to chase flow on this basis. After years of growing FX volumes, prudent risk management has become the dominant strategy, and banks are adopting a “back to basics” approach, focussing on providing services to core clients. 

Banks are now reviewing in detail the profitability of clients, countries, regions and the FX business as a whole.  FX market makers have spent a number of years using technology to make price distribution more efficient and reducing the human interaction required to book vanilla type trades.  As a result, trading data is automatically captured (in most cases) and organisations now have the data to enable them to analyse the profitability of client trades.  This is not the analysis of sales profitability, rather how client trades impact trading profitability.  Through Managed Models, ClientKnowledge is working with 10 banks, providing them with information on trading profitability, and allocating that to clients and client segments.  This information allows them to review client trading in the first instance (and then, if they choose, to implement real-time liquidity management algorithms).  Many banks are now modifying their prices and ONLY chasing client flow where the data proves they have previously made a trading profit.  For those where they have lost money they have modified prices and included a sales mark-up.  

This is also happening with liquidity provision to trading platforms, as banks review the trading profit or loss allocated to platforms and individual clients trading over those platforms.  Of course, other factors are taken into account, such as the overall client relationship, but the focus on profitability and risk management seems to be dominant.  As a result, many banks are unwilling to provide the levels of liquidity seen a year ago, not just in relation to trading platforms but across all channels.

This lack of liquidity could be seen to play into the hands of aggregators, as clients look to increase the depth of book available to them.  Of course there is a cost to aggregating liquidity, and in reality the advantage is in aggregating prices from multiple sell-side organisations, not necessarily increasing the depth of liquidity from any one bank.  Aggregators have seen an increase in business over the past four years and will probably continue to do so as organisations seek to find the best price and available liquidity.  However, the apparent lack of liquidity is no guarantee of this trend continuing, as it is predicated upon a market makers willingness to provide prices to its clients; something that is dependent again upon the analysis of client profitability.   

One of the major trends we are seeing in the market is for the sell-side to focus not only on core clients, but also investing in efficiency and driving out costs in the sales cycle.  This includes initiatives to fully automate pricing, trading, risk management, algorithmic development, order generation and settlement.  In addition, banks are developing real time risk and position management systems.  Clearly, technology enables this process, but only if the existing infrastructure can support the new technology.  As an example, in our recent work with sell-side trading desks we have requested their trading data, usually for at least three months.  In many cases the banks have had to embark on a project simply to produce a set of data, without errors and including timestamps.  There are also banks using technology to maximise and retain spread, employing algorithms to maximise the internalisation of trades, within their own risk parameters.

What is clear is that in 2009 banks are becoming more risk averse, and in doing so market liquidity and volumes are starting to dip.  Whereas volatility buoyed trading volumes to record levels in 2008, signs in 2009 show that this is no longer be the case.  In order to maximise revenues the banks are no longer chasing flow from any source, instead focussing on the basics, servicing key profitable clients and increasing efficiency, thereby reducing costs.

1 The Turner Review – “A regulatory response to the global banking crisis” March 2009, page 97:  “A regulatory regime for trading book capital (discussed in Sections 2.2 (ii) and (vi)) that combines significantly increased capital requirements with a gross leverage ratio rule which constrains total balance sheet size.” 

http://www.humanbenchmark.com/tests/reactiontime/stats.php