Buyside focus : Next generation algos : Chris Hall

A DEEP DIVE INTO DATA.

Chris Hall examines the algo strategies buyside firms are pursuing to enhance performance.

Algorithm performance analysis by buyside trading desks has reached new levels of precision and granularity in recent years. A combination of message standardisation, technology investment and improved analytical skills is enabling desks to wring greater levels of execution performance out of their suites of algorithms, and exert greater levels of control over the execution process.

There are at least three reasons for this intensifying focus on execution analytics. The first is a natural progression resulting from the increased ownership of the execution process by the buyside trader. Initially schedule-based, mechanistic labour-saving devices placed solely in the hands of sales traders, algorithms are now highly sophisticated, finely calibrated tools, easily available on the desktop. To wield such instruments effectively requires a deep level of understanding based on solid, evidence-based analytics.

Second, the lack of transparency over how brokers route orders to venues and how they manage potential conflicts of interest are still causing concern on the buyside in the aftermath of ‘Flash Boys’ and various dark pool scandals. The result is risk committees and other stakeholders are seeking reassurance that trading desks are monitoring executions effectively. Third, regulators are imposing new rules to a similar end, demanding that asset managers demonstrate they have the processes in place to deliver best execution consistently.

Regulatory drivers

With MiFID II’s best execution requirements on the horizon, State Street Global Advisors’ (SSGA) London-based EMEA trading desk has taken a key step forward by reviewing its use of algorithms, based on a framework of ongoing execution performance monitoring. The outcome was a reduction in the number of brokers used from 17 to 10 as well as types of algorithm from 25 to 8, leaving traders to select from a matrix of 80. This was considered a manageable number that could be used with confidence and fully understood by traders, in compliance with the impending regulation.

Be33-Web-12In addition, the firm conducted risk-adjusted execution performance analysis to create a league table that would be used to allocate business over the following six months. Algorithmic performance was analysed against the relevant volume weighted average price or implementation shortfall benchmarks.

SSGA is currently going through its second full six-month cycle of this performance analysis process, but will shift in July from the original table of 10 core brokers to three leagues of four. To ensure a sound basis for monitoring algorithm and broker performance, it utilises its own internal trade analytics capability, but also uses a single third-party transaction cost analysis (TCA) provider to analyse all executions. The asset manager consolidated its FIX connections with ostensibly one provider to enable analysis of FIX messages and time stamps on a like-for-like basis.

SSGA’s analysis has focused on differences between providers, for example, the percentage of passive fills per order, and the impact this has on overall performance. This is useful in its own right, but also increases the real-time understanding of the routing logic of brokers’ smart order routers (SORs) for specific strategy types. “If we can see differences between the executions achieved via different brokers, for example, in the amount of flow that is internalised, or from the proportion of resting executions on particular venues executed by a VWAP algo, then we’re in a better position to have an informed discussion with brokers about their performance and the routing logic of their SORs relative to their peers,” says Dale Brooksbank, head of European trading at SSGA.

Don’t delay

While MiFID II’s best execution requirements have been delayed, asset managers have been on notice to step up their game since the Financial Conduct Authority’s ‘Best execution and payment for order flow’ thematic review of 2014. Brooksbank, also chair of the buyside trading committee of the Investment Association says, “The FCA’s review made it very clear that asset managers need to have the level of understanding that enables them to hold brokers to account for the execution quality they provide. We need to be asking brokers to explain how they manage potential conflicts of interest and we can best do that by having rigorous independent analysis that allows us to understand how they select and route to venues. We’re seeing more conversations on these topics across the industry,”

He adds that the big leap forward for the buyside is the ability – enabled by the FIX trading protocol – to analyse the fills at the child order level in real-time and their contribution to the overall market impact of a trade. “Having this granular level of analysis is good for both the sell and the buyside in that it helps us to improve our knowledge and fuels ongoing dialogue around service improvement,” he says. “The improvement in buyside performance analysis stems in large part from increased investment in analytics capabilities by asset management firms, in parallel with a significant up-skilling on trading desks.”

Be33-Web-11Further innovation is afoot. At Principal Global Investors, Huw Gronow is in the midst of a project to generate more detailed analysis of equity trading outcomes in order to ensure alignment between the execution decisions of his London-based trading desk and the intent of the firm’s portfolio managers. As head of equity trading for Europe and Asia, Gronow recently helped to oversee the implementation of an upgraded order management system and is now mulling further technology investments to support more detailed analysis of equity trading performance.

“We’re looking to close the feedback loop between the portfolio management function and the trading desk in order to optimise the implementation process,” he says. “ As part of this initiative, we’re aiming to systematically extract data from the fields in the FIX messages generated by each child order.

While the project was initiated before MiFID II’s best execution requirements were firmed up, Gronow sees beneficial overlaps with upcoming compliance requirements. “Asset managers need to collect a great deal more execution data to provide the analysis required to meet best execution requirements under MiFID II. We’d hope our existing investment in analytics and compliance efforts will be mutually reinforcing,” he explains.

At an industry-wide level, Gronow suggests there is a growing impetus behind standardisation of execution performance analysis, driven in part by a desire across the buyside for closer scrutiny of SORs. As co-chair of the FIX Trading Community’s EMEA investment management sub-committee, Gronow is at the helm of a new push to standardise use of FIX across the buyside. “If we can achieve consistency in the information that is returned to the buyside via FIX messages we are in a much better position to compare execution outcomes, across venues, brokers, algorithms, etc,” he asserts.

Dig deeper

As the usage of execution algorithms extends from equities to derivatives, so does the demand for in-depth analytics. According to Yuriy Shterk, head of derivatives product management at Fidessa, fast followers at large investment institutions, including some insurance firms, are joining the quant-based early adopters of algorithms when trading listed derivatives, thus increasing the interest in analytics over the past 12-18 months.

Be33-Web-10“The concept of TCA is beginning to appear in listed derivatives. Increasingly, traditional equity houses are demanding the same level of analytics when trading futures and options,” says Shterk, whose firm launched a derivatives analytics service last year. “Demand depends on sophistication. Some firms want to benchmark the performance of algorithms and brokers to determine future order flow, others want to understand the results of an execution in more detail in order to improve future performance. The more advanced firms are looking to use the analytics to enable the algorithms to adapt to circumstances, such as changes in market direction or volume.”

©BestExecution 2016
[divider_line]