Vittorio Nuti

FX algorithmic execution – helping to give the buyside more control over Market Impact

June 2023 in Previous Features

Vittorio Nuti, Global Head of LD & FX Algo Trading at Deutsche Bank discusses some of the key issues regarding Market Impact.

What is Market Impact and why is it such an important consideration for many firms executing FX trades?

Market impact is the effect that a user has on the price of an individual asset by transacting. It can be categorised into temporary and permanent depending on the duration it takes for the asset price to recover. Information leakage and price volatility also form a core part of market impact consideration. It is important to consider market impact in its totality, as it forms part of the transaction cost of execution and market impact can significantly affect the performance of a trade. Firms, like Deutsche Bank, factor this additional consideration into the execution logic to optimise the outcome that we can achieve for our clients.

What factors are likely to influence the average Market Impact resulting from the implementation of a specific strategy?

Combining size and duration of orders to calculate the speed of execution, we can model the potential market impact based on market conditions. Influences on market impact will be mostly driven by the liquidity and volume of the asset in a given horizon. Less depth in a book combined with smaller volume expected can mean large and fast orders can have significant market impact. This impact can propagate and persist to leave not only price volatility, but a permanent impact on price if opposing volume doesn’t return.

How do leading FX algo providers like Deutsche Bank go about computing the level of expected Market Impact that their algos will be able to deliver?

For calculating market impact, we look at various indicators including the participation rate. This calculates the order volume as a percentage of expected volume, the higher the value, the greater the expected impact. Deutsche Bank has also created a PreTrade Tool that models average and expected slippage based on order parameters to allow clients to view their potential market impact. We utilise such impact modelling, along with assessments of current market conditions, to form decisions on optimal execution for our dynamic algos.

What sort of choices regarding Market Impact do buyside firms need to weigh up and consider when it comes to balancing the risk of it against other risks of their trading objectives and strategies?

The speed of execution forms an important metric in algo decisions for all firms. Market impact is often balanced against this along with subsequent performance against benchmarks, including risk transfer price. Using analysis (such as Deutsche Bank’s PreTrade Tool) can help clients pick dynamic algos that weigh up their need for faster execution against the minimisation of market impact. Establishing ranges for expected slippage and comparing this to the arrival price, firms can make informed decisions about how algo performance may improve execution over a simple risk transfer.

What sort of data and metrics can help firms to better understand what the expected impact of a given set of algo strategies would be, on average?

The speed of execution forms an important metric in algo decisions for all firms

As mentioned before, participation rate forms an initial assessment of how an individual order may impact the market if we have a timeframe. With different constraints we can form better models that calculate a confidence bound for the expected rate of slippage.

There is extensive academic literature on how to model impact but the basis builds upon a function of volatility and volume. More dynamic algos will calculate this impact and balance against the client’s desired speed of execution to find an optimal horizon. Firms should investigate slippage from previous orders and investigate algo strategies market impact models that dictate execution.

In what ways are the latest generation of FX algos able to be configured in ways that can deliver more control over Market Impact?

Algos are now more aware of not only the market impact other participants have but also the movement they are causing in the market. To have a simulated model for potential market impact of any given fill allows algos to be calibrated with better precision to minimise impact. This new generational logic enhances the overall rate clients achieve through this process. As providers, we continuously have more data on the market impact in the time periods around any fill we make. It is a case of being much more aware of any influence algo actions have on each pair at any given time.

What advice would you give to buyside firms who are potentially looking to use FX algos to help them manage and fine tune the impact of their trading behaviour?

Discuss with providers the logic behind their impact modelling and how their algos dynamically react to differing market conditions in order to minimise market impact and hence transaction costs. Assess the performance of previous algo orders by looking at how the algo performed against the average expected slippage and the confidence interval of slippage. Comparisons against arrival mid don’t always paint the fullest picture of the algo’s real market impact. It’s important to have a conversation with providers on how their algos operate and your desired outcome and allow them the flexibility to suggest options most suited to your objectives.

How can post trade toolsets be used to determine how effective an FX algo trading strategy has been in reducing Market Impact?

Post trade benchmarks and tools are a vital part of a client’s assessments on how algos have impacted the market during their execution. It is important not to view them in an echo chamber as there could be exogenous market events that impact execution outside of the actions performed by the algo.

However, over a large enough sample set, looking at performance against average expected slippage, performance against arrival mid and whether execution fell within the expected range of execution speed can form standards to assess a provider’s impact execution.

Changes in these benchmarks over time allow you to see how algo execution logic is altered over time by providers fine-tuning their responses to order placement and fill market impact. Although not a perfect metric, it does allow clients to form a comparison metric across providers.