Portfolio traders face two challenges when trading baskets of stocks on dark venues. First, they need to maintain the cash neutrality of the list while maximizing their exposure to potential liquidity. Second, they need a framework to determine the priority in which they choose what securities to execute.
Dark aggregation algorithms are traditionally used on a single-stock basis and cannot easily be applied to multi-security trading because the random arrival of fills makes it difficult to maintain cash neutrality.
“A lot of portfolio traders are not participating in dark to the extent that their single-stock peers are,” Ben Polidore, head of algorithmic trading at ITG, told Markets Media. “With dark trading, I can't just say 'here's a buy, here's a sell,' send them to dark and assume they'll both get done at the end of some time period over which I'm optimizing."
"Whereas, if I use VWAP (Volume-Weighted Average Price) or another non-discretionary algo, I can run an optimization process that just takes chunks and sends them off to this system down the line and know that they'll all trade,” Polidore continued.
ITG’s Dark List algorithm is designed to address this problem by using probabilistic modeling techniques, i.e., stochastic processes, to estimate the likelihood that a basket of buy and sell orders will be filled on a dark venue.
“What we're trying to do with Dark List is come up with an approach where portfolio traders can maintain the constraints that they need to execute their mandate, but expose more liquidity to dark and trade in a more opportunistic way,” Polidore said. “That's quite simple for single-stock traders, but it's very difficult when you have hundreds of names that you want to trade at somewhat similar rates.”
Ben Polidore, ITG
Dark List optimally sources higher-quality liquidity from dark pools, according to ITG. Allocations are based on an optimization process that continuously learns from real-time fill and market data. Advanced statistical techniques estimate stock-specific minimum quantities allowing posted dark orders to avoid information leakage without missing out on liquidity.
“The objective here was to create a way for portfolio traders to trade opportunistically without violating constraints,” said Polidore. “We came up with a model first, then we had to make the model fast, which is the stochastic programming approach, and, then, once we have this we can do opportunistic trading instead of deterministic trading.”
Portfolio trading involves the execution of a basket of securities simultaneously. Compared to trading multiple securities independently, portfolio traders must consider a number of additional execution objectives in their decision-making process, such as minimizing risks, obtaining cash targets, and investing in multiple-asset classes.
This makes the use of algorithms for portfolio trading problematic. “Just like human traders, algorithms balance realized transaction cost against residual portfolio risk in order to minimize total trading costs,” said Wenjie Xu, head of algorithmic research at ITG, in a report. “A basket can be executed patiently with low transaction cost, but patient trading comes with the risk of future adverse market movement. In contrast, aggressive trading reduces risk, but incurs higher transaction cost due to increased market impact.”
Most sell-side institutions maintain a portfolio trading desk separate from active trading desks due to the different styles of trading: whereas active traders focus on a single-stock trade, portfolio traders focus on centrality of fills for the overall basket.
“Since portfolio traders are usually focused on risk factors and managing them, often they're not thinking so much about liquidity,” said Polidore. “They can trade it slower, so that's why it's a specialized practice because they're really thinking about risk factors and how to manage them, whereas single stock traders are usually thinking about liquidity.”
Feature image by AnatoliiLahutkin/Dollar Photo Club
