Better Than Average

Share

By Belinda Fong, AES Sales, Credit Suisse
In an ever-changing market place, we keep a close eye on trends and changes to better direct our business. For that we trove through a vast amount of data on a daily basis. Strategy selection is one of the key metrics we look at. There is one statistic that our clients are oftentimes surprised by: the popularity of VWAP. Last year, amidst widening spreads and dwindling volume, a whopping 53% of AES flow in the region was VWAP. Ever since its inception at the turn of the century, the humble strategy that aims to help you be average remains the cornerstone of any algorithmic trading suite. It is still the go-to strategy of choice for a passive over-the-day outcome that is mathematically the “Minimum Cost” way to trade.
Misconceptions of VWAP
For something used so frequently, there are nevertheless a couple of common ideas about VWAP that are misconstrued. The first thing is market impact. For a given timeframe, VWAP is the slowest possible way to trade for completion. It is also preferable over TWAP as it follows the well-established volume smile, so you are not too aggressive during the middle of the day, unnecessarily increasing market impact. All well and good if your size is small and can stay hidden. However, in the current low liquidity environment, most trades are of sizeable ADV percentage if not multi-day; as the ADV% of your order increases, so does the impact. Our numbers show that once you go above 5% of participation your performance on VWAP starts to deteriorate. But where does the impact come from? Most VWAP strategies run on a historical volume schedule with limited real-time consideration. This means that even though the bid only has 1,000 shares, a buy VWAP order that needs to trade 10,000 shares right now will go right ahead to post and subsequently cross the spread to keep to schedule, thereby impacting the price and signalling to the market.
Do price points matter?
The second key point is that the price you trade at is in fact irrelevant to the implementation of VWAP. This can seem counter-intuitive at first. One of the questions traders often have around over-the-day executions is, why did I miss VWAP if the market consistently came my way? I have been buying as the price got lower, should I not have beaten the benchmark? The answer is, of course, no. What drives VWAP performance is being able to accurately match the intraday volume profile. There is a fair amount of literature on price and volatility in the search for alpha, which has somewhat diverted the conversation from volume. Volume is one of the most unpredictable aspects of markets. The life of a trader would be vastly different if he or she knew how much a stock was going to trade on the day. You can think of a full day VWAP order as a series of smaller orders of 5-minute periods. How much is allocated into each bucket makes up the trading curve. Assuming that you match the average price of each period, you can still slip against VWAP if your 5-minute buckets are not distributed correctly over the duration of the whole order. VWAP execution can essentially be reduced to a problem of choosing an optimal trading curve and minimising volume slippage.
A dynamic and adaptive solution
Current VWAP tactics slice orders according to a stock’s historical intraday volume profile. In general the trading curve of a VWAP order is determined on order arrival and will remain static for the rest of the order life time. A departure from heavy reliance on historical data is inevitable for a smarter way to trade this ubiquitous benchmark. To accurately implement VWAP, algorithms will need to adapt and dynamically adjust your participation rate on the day. In this evolved form, VWAP algorithms will comprise of a historical component and a dynamic one derived from current market conditions. Additionally, trade signals can be captured and used to make adjustments to the trading curve to minimise the drift in volume trajectory versus the actual day’s volume.
The opportunistic framework
Much can also be learnt from the advance of more opportunistic algorithms for intelligently seeking liquidity. Encapsulated in our ongoing work on the Opportunistic Framework, an underlying structure for tactics that adapts to an expansive set of trading environments through smarts around the real-time order book; opportunistic tactics that are agile in their reaction to changing real-time conditions will help keep impact low regardless of size. These algorithms also have spread capture as their prime objective, given high spread cost in the region. This is achieved by dynamically posting at multiple levels to maximise queue priority and improving near-side fill rates. When it comes to paying the spread, tactics will only get involved when opportunities are right. In response to the faster and volatile nature of current markets, we have recently augmented and fine-tuned the behaviour to be more agile. Interval VWAP figures for our flagship strategy within this framework supersedes traditional algorithms with little deterioration as spread widens.
Opportunistic VWAP
Bringing the Opportunistic Framework together with dynamic scaling of the intraday trading profile, we get a VWAP algorithm that has the best of both worlds. Starting the day according to historical profile, it will use real-time statistics to continually adjust future participation buckets as the order progresses. A trading envelope will bind the dynamic trading curve depending on the liquidity of the stock while allowing discretion for best performance. For stock that has poor profile stability, this allows us to capture unpredictable volume. The dynamics of individual slices would inherit opportunistic behaviour, with the algorithm seamlessly weaving them into the order book reducing signals and spread cost to a minimum. Spread-crossing decisions are driven by proprietary quantitative logic and closely monitored for optimal results. With little volume slippage, this is a nimble and agile way to trade VWAP that we are confident will consistently produce significantly above average performance.
We’d love to hear your feedback on this article. Please click here