The proportion of spot FX traded algorithmically by corporates shot up to 28% in 2016, from only 10% in 2015, according to a Greenwich Associates study.
Across the wider FX market, corporate use of algos increased to 10% from 7%, states the report, written by Richard Johnson, vice-president of market structure and technology at Greenwich, which investigated long-term investor appetite for FX algos.
Growth was concentrated outside the most active FX trading corporates – those trading more than $50 billion per year in notional value – where algo use was already well established. For these corporates, algo use remained essentially flat, having declined to 23% in 2016, from 24% in 2015.
The most startling jump was for the least active corporates, trading less than $1 billion notional per year, where FX algo use rocketed up to 10% in 2016 from no use at all in 2015. For corporates trading between $10 billion and $50 billion notional it increased to 12% from 8%, while for the $1 billion to $10 billion segment it was up by 1% to 3%.
Observers in the FX algos business say the survey confirms trends they have observed themselves. Rather than revealing a surprising direction of travel, it provided interesting insights into the speed of travel – the rate at which FX algo use is catching on.
Johnson says: “The electronification of the market has yielded great benefits to institutional investors, but it has also created a sometimes tangled web of liquidity sources that require sophisticated technology to untangle.
“As a result, execution algorithms are increasingly penetrating the buy side and corporates, as they seek to improve their quality of execution.”
The average corporation in the Greenwich study was trading $20 billion per year. On average, “each basis point of improvement in trading performance translates to an additional $20 million straight to the bottom line”, Greenwich concludes.
FX algo use has grown for real-money accounts too, by about one-third, to 21%, from 16% year on year. Here the picture was reversed relative to corporates, with growth in the spot market practically flat, inching up from 20% to 21%. Growth was a relatively consistent 3% or 4% for managers trading different volumes, except for the $10 billion to $50 billion segment, which spiked to 19%, up from 4%.
Greenwich argues algos help traders achieve and demonstrate best execution, as laid out in the FX global code of conduct and Mifid II, which is the single biggest reason for their increasing use in FX.
While the rules only apply to forwards and options, it is easier for traders to apply the same standard across the board than treat FX derivatives and spot differently.
This healthy growth in FX algo use is encouraging banks to dedicate increasing resources to the business. Among the more recent entrants to the market is Northern Trust, which recently unveiled the FX algo suite it has developed in partnership with Chicago-based BEx.
Northern Trust offers clients a “liquidity panel” that aggregates pricing from up to 15 global liquidity providers, depending on the specific currency pair, including exchanges and Euromoney’s top 10 FX banks. Execution algorithms ensure minimal market impact for large orders.
Tim Linehan, head of operations at Silchester International Investors, says the liquidity panel ensured it saw “competitive quotes and narrower spreads without a corresponding increase in counterparty exposure, balance-sheet risk or operating complexity for Silchester and its clients”.
Rowan Gillespie, senior e-FX sales at Northern Trust, says: “We create bespoke algos for our clients, which we also believe is a differentiator from our competitors. It means we can work together with clients to create algos that replicate their existing trading strategies and create new strategies.”
However, the Greenwich report states that “while providing customization is appealing, it often translates into complexity, which is the opposite of what the buy side is looking for. When executing trades with significant notional value, simple and transparent orders are best.”
Greenwich also found that FX investors currently use algos predominantly for passive trading.
“If they want to trade an order aggressively, they will more often choose to interact directly with streaming quotes, without paying a commission as they would if using an aggressive algo,” says Greenwich’s Johnson.
In aggregate, passive algos attract 80% of the spot algo flow, with corporates particularly likely to use passive over aggressive algos.
Northern Trust’s Gillespie confirms this trend.
“We expect to see more diversified strategies as our business grows,” he says. “But for now the majority of our clients are real-money institutional accounts and they are typically looking for passive guaranteed execution algos that minimize their market impact.”
Northern Trust reports significant take up for its time-weighted average price (TWAP) and TWAP-Randomized algos, the latter of which randomizes amounts and time intervals to further reduce market signalling.
Hedge funds and fund managers are still heavy users of passive algos, but are slightly more likely to use aggressive algos.
Curtis Pfeiffer, chief business officer at Pragma Securities, says: “Since there are many types of hedge funds strategies, they use execution algos in different ways.
“A macro fund that takes big directional currency bets will be more likely to use aggressive algos, while a systematic quant fund trading many orders simultaneously, with lower liquidity demand, is more likely to use passive algos.”
The healthy growth in FX algos invites questions as to just how far such trading could eventually extend. It is already far more entrenched than in the fixed income market, which has so far proved difficult for algos due to the variety of contract types, maturities and other characteristics.
While algos have made some inroads into the US treasuries market – the most liquid bond market in the world – they account for a smaller proportion of flow than in FX or equities.
Whether FX algos could ever reach the levels seen in the US equity market, where their dominance is total, is another question.
Pfeiffer says: “Algos account for around 80% of order flow in the US equity market. Could FX one day get to that level? Sure, many things are possible, but we see it as unlikely because of the FX market structure.”
Johnson warns against making exact comparisons between equities and FX, but is confident algo usage in FX will increase.
“The 80% of volume traded in algos refers to total volume, including buy side and sell side, and is not directly comparable,” he says.
“However, consider that in equities, 89% of institutional traders use algos for about 33% of their flow. On the FX side, I do think we can see penetration of algos in institutional real-money accounts get to above 50%.”
Johnson adds: “Among corporates, we are unlikely to see broad adoptions, as they are not financial companies, but given recent increases in penetration I think there is further room to grow.”
Northern Trust is evidently banking on it.
“We believe around 15% of FX flows are traded algorithmically, so there is huge room for growth in this business,” says Gillespie.