Articles
May 1, 2024
Originally published in FX Algo News, May 2024
Users of FX Execution Algos are becoming increasingly interested in their providers’ internalisation rates, as the buy-side seeks to optimise performance and differentiate between available offerings. It is crucial to understand the details behind headline internalisation rates to ensure best outcomes.
What is internalisation?
Internalisation is the Algo provider’s process of offsetting risk (positions) arising from a client’s Algo child orders with risk from other clients, such that the external market is unaware of the order and resulting trade. However, there are a number of different execution practices that facilitate this offsetting of client risk, making it a more nuanced activity.
Algo providers take differing approaches as to what they classify as internal liquidity and how that liquidity is generated. Importantly, there can be configurable parameters (or a more general opt-in/out) on the Algos that determine the available liquidity sources and trading styles.
Principal Liquidity (Internal Market-making Desk)
The Algo provider’s internal market-making desk generates principal liquidity which Algos can execute against or submit passive orders into. This may decrease or increase the dealer’s risk. It is important to consider what happens to new positions that the dealer acquires as all firms have risk limits that ultimately constrain the accumulation of inventory, even if those limits are large. Dealers will look for opportunities to reduce their risk through efficient hedging strategies or price skews, which can both cause information leakage.
Risk management strategies vary between dealers, with significant differences in the resulting signalling risk.
Franchise Liquidity (Client Flow)
Client flow can be offset in a variety of ways:
- Match against the dealer’s existing positions acquired from other clients. This may be part of a hedging or market-making strategy and so can overlap with the definition of principal liquidity. Both approaches are specifically risk-reducing for the dealer, so they should not result in interaction with external markets.
- Match directly via an internal orderbook. Information leakage can be managed by limiting the participants of the orderbook.
- Match over private bilateral channels with other clients, potentially including other market-makers. This carries greater signalling risk as anonymised order information may be shared more widely with other clients. This might be as part of price skews, or orders themselves.
Dark Liquidity
External venues offering ‘dark’ pools of liquidity are sometimes included under internalisation. Details vary by venue, but the key concept is that there is no visible orderbook to prevent information leakage, and trades are generally executed at a mid-market rate. The execution mechanics are important; if well designed they can offer counterparties fair matching on equal terms, and can alleviate some of the conflicts of interest associated with traditional forms of internalisation.
What are the benefits of internalisation?
Reduced market impact
This is generally the primary driver. The premise is that by internalising the execution, less information will leak into the market, therefore reducing the market impact of the trade.
Reduced fees
Execution on external venues often incurs brokerage fees, so internalising trades can save on that cost, resulting in a cheaper service.
Reduced slippage
Internal order matching and fills should be possible with lower latency and more consistent execution due to greater control over the systems, network, and execution mechanisms such as last look.
What are the challenges with internalisation?
Variable Market Impact
Each category of internalisation will have its own market impact characteristics, with varying degrees of signalling risk. Moreover, the market impact sustained from each method may vary significantly between providers due to differences in their implementations and the composition of their client franchises.
Even within each subcategory of internalisation, market impact can vary hugely. For example, offsetting risk with a hedge fund client is likely to look different to a corporate client who has very different reasons for trading. The source of the liquidity is significant, even if it is ultimately internalised.
By understanding the internalisation methods practiced by each provider, clients can evaluate their pros and cons. For some clients, skewing prices may be viewed as an effective way to reduce transaction costs and increase the urgency of the Algo. For others, it may carry too high a signalling risk and negatively impact the overall cost with worse execution.
There are unfortunately no shortcuts – internalisation rates alone cannot be used as a reliable proxy for low-market-impact without more detailed analytics and TCA covering both internal and external execution.
Conflicts of Interest
There is potential for conflicts of interest between the Algo desk and the principal desk. Careful data management and process design is required to ensure order information is not used by the principal desk in a way that results in worse execution for the client. Order routing must not be biased towards internal liquidity if it is not necessarily the best execution option for the client, and rates on internalised trades must be fair.
An FX Global Code review that included consideration of Algos concluded that transparency is the most suitable mitigant, consistent with the other FX Global Code transparency guidelines. Standardised disclosure information is encouraged via an Algo Due Diligence Template. ‘Conflicts’ is just one section, which covers fairness in execution, order priorities, and commercial interests.
In addition to understanding the disclosures, it is critical to take a data-driven approach to evaluating both internal and external execution performance.
Price Discovery
The trend towards greater internalisation of both Algo orders and risk transfers is understandable when it can benefit both clients (reduced market impact) and dealers (lower costs, greater spread capture, private information). This has led to a reduction in volume on primary venues in particular, which is where much of this business would have been executed historically.
Price discovery processes have had to evolve as a result, with less ‘lit’ volume on orderbooks. It may be the case that while internalisation can benefit the individual market participant, there are structural implications for the broader market.
Conclusions
Internalisation is a broad term which can cover many forms of execution and varies between Algo providers. It is worth understanding which methods each Algo provider is utilising and the nuances of their implementations by reviewing their disclosures and engaging with them directly.
While it can provide reduced market impact, trading costs and slippage, the observed effectiveness of internal execution should be analysed along with external executions using post-trade analytics and TCA. Granular data from Algo providers should allow this, but it may be necessary to combine it with additional external data analysis to improve the robustness of the conclusions.
Market impact is not the only relevant metric; users must consider what their own benchmarks are. There are likely to be times when it is necessary to sacrifice market impact for speed of execution, for example.
The optimal composition of liquidity will be different for each client, and different with each of their Algo providers. It is not as simple as maximising the internalisation rate – a transparent approach from Algo providers allows users to evaluate how they would like their orders to be executed to best suit their execution goals.
Navigating the complexities of FX Algos can be challenging. With their deep expertise and experience in the FX market, the team at Hilltop Walk Consulting bring unique insights. Partnering with clients to reach better outcomes.
