PM misalignment driving traders to ditch VWAP
Asia buy-side desks are following US peers by ditching volume-weighted average price as a benchmark for algo trading on the grounds that their interests become misaligned with portfolio managers.
Heads of trading at asset management firms and heads of execution at investment banks in this region all point to a shift away from VWAP towards use of implementation shortfall (IS) or participation weighted price (PWP) as an algorithmic benchmark.
While brokers in Asia still use VWAP for more than 50% of their electronic execution, estimates one Hong-Kong based specialist, this is falling and is destined to fall further. He says in the US, where the move away from VWAP has been going on for years, the figure is now about 20%.
One shortcoming with VWAP is that it can be manipulated by traders to shore up performance, and as such it is not seen as indicative of whether traders are genuinely alpha-generating.
VWAP data on single stocks – widely available on trading terminals – is calculated by adding up dollars traded for every transaction, and then dividing it by total shares traded for the day.
Of course, traders need to be able to show portfolio managers that the average realised transaction prices they achieved for the day either outperform or are in line with the benchmark.
Emma Quinn, head of Asia-Pacific trading for AllianceBernstein, explains why her team now uses IS and PWP instead of VWAP
“I could have a trader that buys a stock over 10 days and every day he could beat VWAP on that stock,” she says. “Over that 10-day period the stock could have risen by 10%.
“But if the trader had completed his order within the first two days, the stock could have risen by only 5% and his expected costs would be much lower than over the 10 days. So this trader might look great on VWAP, but against an IS benchmark those executions look terrible.”
Quinn suggests that IS has not gained as much traction in Asia as the US yet because it requires traders to take a view on a stock, whereas with VWAP the goal is to get the average, which some investment houses favour in order to reduce market impact.
But she notes: “It is less easy to explain to your portfolio manager when you get it wrong with implementation shortfall than VWAP. With VWAP you just need to pick what sort of volume constraints you want. With IS, as you are taking a view on the stock, you need to be more of a risk-taker.”
One key reason for using a VWAP benchmark is to ensure the trader executes orders in line with a stock’s normal volume through an average trading day (reducing market impact). But VWAP is also vulnerable to manipulation.
Ofir Gefen, Asia-Pacific head of research and algo consulting at ITG, says once the size of an order tallies up to over 50% of daily volume traded, that order has effectively “become the VWAP” – meaning the trader can easily manipulate it so the benchmark is calculated based on his orders.
Adrian Valenzuela, managing director for equities at Barclays, says that as the bulk of a day’s trading tends to centre around market close, a trader could hold back a portion of their order to take advantage of this additional liquidity and enhance their ability to outperform the VWAP average.
“For example, a buy order using VWAP where the stock’s last closing price is moved up markedly at the last minute could make the day’s buy orders appear ‘alpha-generating’ as they appear to be lower than the benchmark average,” he points out.
Gefen also notes that while a trader using a VWAP benchmark is implementing a trading decision for a fund their goals are misaligned, given that the fund manager wants to maximise returns.
“Implementation shortfall is a very deterministic benchmark and your target is set very clearly,” he notes. “Traders execute very close to the time they receive the order. They know what price they are aiming for and they try to add as much value as they can to improve the return on the portfolio while trying to lose as little alpha as possible.”
He notes that the use of IS in Asia has grown over time, with early adopters having started in 2006/2007. Now, he believes Asia is “in the beginning phase of crossing the chasm. I would be surprised if by 2017 IS is not the main benchmark of 60-75% of funds in Asia,” he reckons.
Separately, bankers say that brokers will struggle to win meaningful business in algo trading if they are selling commoditised, vanilla-type algos.
“An additional evolution in electronic execution is about bespoke product, algos that are customised to the needs of the client according to his specific benchmarks,” notes Valenzuela.