Like a growing number of long-only asset managers, Invesco contends it's no longer enough to rely on third-party transaction cost analysis in a marketplace that's becoming increasingly complex amid a sea of data, rapid-fire trades, and fragmented liquidity.
Third-party analysis is still important to Invesco – and for much of the buy side - since most participants see the value of having an outside entity analyze their trading costs, while acting as a sounding board for TCA-related ideas and advice. But in order to wring as much as possible out of each execution, Invesco says it's become equally important for firms to have their own customized TCA, which can provide better context behind each transaction, and ultimately help traders sharpen their strategies faster than what's possible when relying solely on outside analysis.
As a result, Invesco said it's now using OneMarketData's OneTick platform, not only for building custom TCA models, but also to run historical simulations and conduct quantitative research.
"We continue to use a third-party provider. It's always a good foundation to fall back on to make sure you're on track with the basics," says Dwane Bacak, the global head of TCA at Invesco Ltd. "But by including a customized solution on top of that and providing additional insight, you're able to test out ideas quicker and implement them in a customized solution where they show up as quickly as you can turn them around."
Within "a couple of days," Bacak says it's now possible to have such analytics embedded in their system. OneTick's platform will give Invesco more flexibility with the benchmarks it uses, a crucial feature that Bacak says will give a window into details that usually go unnoticed by third-party metrics.
“What’s easy to get lost is when you’re looking at these summed up numbers over a period of time,” Bacak explains. “When you’re looking at a particular trade, for example, maybe there’s an earnings surprise out there and you’re trying to understand how that trade did against different benchmarks. You can look at implementation shortfall, but that probably wouldn’t be the best benchmark. In that particular case, you can look at something like volume weighted average price (VWAP) because that takes into account that which would be more appropriate when there’s an earnings surprise in the name as opposed to just looking at one dimension of an implementation shortfall benchmark.”
Richard Chmiel, a senior vice president at OneMarketData, says the platform will also help Invesco get a tighter grip on fragmented liquidity, a problem he says is the biggest challenge that traders now face in optimizing their executions.
"You've got dark pools, you've got light pools. You've got brokers and broker algos. Deciding where to place an order continues to be the biggest challenge," Chmiel says. "What I have to do is use history as a way of predicting the future. The only thing I have is history. So I take history, models, smart quants, and a platform like OneTick and I try to predict the best way to handle an order."
Invesco's decision to use OneTick to augment its TCA comes as part of an overall trend among asset managers to use custom analysis alongside what they're getting from third parties. Bacak says high-profile events like last month's Knight Capital fiasco and the rise of high-frequency trading are part of what's driving the trend.
But he also points out that technologies like OneTick have placed such analysis within everyone's reach, something that wasn't always possible.
"The way the world used to work is you'd have to reach out to someone with a supercomputer of some sort that would allow you to do this type of analysis," Bacak says. "Now there are tools to enable you to create platforms that can provide you with feedback that helps you make decisions given the types of trades you're doing."
Chmiel adds that Invesco will even be able to write alpha generation models with the OneTick system.
"What you're looking for is ways to make money based on historical movements of stock or individual sectors of stock," he explains. "Given certain situations, how does a stock behave? How does a sector behave? They can write these very complex models that search for repeatable events. Once you have these seven factors aligned, these events occur 70 percent of the time."
He continues: "That is enough to build a trading model to make money. That's what they'll do."