Articles
February 1, 2024
Originally published in FX Algo News, February 2024
Execution Scheduling algos
Execution Scheduling algos methodically distribute trade execution over a set time period in order to balance two types of risks: market impact risk and volatility risk. So let’s explore their utility and the critical nature of timing in their performance.
Market Impact mitigation: By dispersing trades, these algos mitigate the market impact, which is the adverse effect large orders could have if executed simultaneously. This is particularly relevant for sizeable trades that, if placed at once, could move the market unfavourably against the trader.
Volatility risk management: However, when spreading trades over a period of time, the corresponding increase in volatility risk must also be considered. This is the risk that the market moves unfavourably against the trader as a result of underlying volatility in the market unconnected to this particular order. This must be balanced with market impact.
Significance of the duration parameter
The performance of Execution Scheduling algos is heavily influenced by the selection of the duration parameter, which determines the period of time over which trades are executed, and contributes to the balance of risks. Selecting the appropriate duration requires an understanding of market dynamics and the objectives of the trading strategy, and often falls to the qualitative judgement of an execution trader.
The Quantitative versus Qualitative dilemma
Algos are lauded for their systematic, quantitatively-driven approach to trading, which often contrasts with the more subjective method of setting the key duration parameter. When this critical element is determined arbitrarily or based on the trader’s discretion, it introduces an element of unpredictability into the algo’s performance.
This can lead to a scenario where the algo’s performance is not fully governed by its quantitative framework, diminishing the control and predictability that may have driven the adoption of algos.
Benchmark considerations
In the context of FX algo trading, the selection of an appropriate benchmark is crucial and can often determine whether Execution Scheduling algos are the right strategy.
Mismatch with transaction goals: Execution Scheduling algos are often benchmarked against average prices over a set period. However, the average price over an arbitrary period may not reflect the true objectives of the trader, particularly when the FX risk is realised at a particular point in time.
Targeting a point-in-time benchmark, such as the WMR fixing, requires a strategy that accounts for market dynamics and liquidity at that specific moment. The uniformity of Execution Scheduling algos does not always offer the flexibility needed to adjust to these dynamic conditions, which can lead to suboptimal execution when compared to this type of benchmark.
Tailored execution strategies: Research by organizations like Siren FX indicates that for larger trades or those targeting a precise benchmark, a different execution strategy, such as starting earlier and increasing execution rate closer to the benchmark window, might be more effective. This considers market movements and the impact of trades in a more nuanced manner than the one-size-fits-all approach of Execution Scheduling algos.
The role of Execution Scheduling algos in FX trading
While Execution Scheduling algos have a place, their suitability is not universal and is heavily contingent upon the approach used to select parameters, and the chosen benchmark for success. Firms must carefully consider these factors to determine the solution that can deliver the best outcome for their specific needs.
Differentiating among FX algo providers
Most market-makers provide a suite of Client algo offerings, however, the assumption that all these offerings are created equal could lead to overlooking critical nuances that might impact outcomes. Understanding the potential differences between these offerings is essential, particularly as they may align differently with the diverse use-cases of market participants.
Variability in strategies and models
The diversity of models offered by algo providers is a primary area of differentiation that can significantly impact success. As discussed in our previous article, the three major algo strategy families—Execution Scheduling, Arrival Price, and Market Impact Minimisation—form the foundation of most algo offerings. However, the presence and implementation of these strategies can vary considerably among providers.
Range of strategies: Not every algo provider will offer all three core strategy types. A provider’s focus may lean towards one strategy over others based on their market philosophy, technological capabilities, or client base.
Model specificity: Even within providers offering the full suite of strategy families, the underlying mathematical models used to drive these strategies can differ. These variations stem from distinct research inputs, proprietary techniques, and model philosophies.
Model-driven results: The choice of mathematical model is far from trivial. It can dictate the algo’s behaviour in different market conditions and ultimately determine its performance. For instance, two providers offering Arrival Price algos may produce different execution outcomes based on how their models interpret and respond to real-time market data.
Timing and liquidity awareness: The ability of an algo to discern and act upon fluctuations in market liquidity is crucial. It should have the capability to understand liquidity patterns and anticipate changes, allowing it to navigate through market microstructures effectively and choose the most opportune moments to execute child slices of an order.
Use-Case alignment: Different use-cases demand different model attributes. A model that excels in a high-liquidity environment might not perform as well in a market characterized by volatility and thin liquidity. Traders need to match their requirements with the specific strengths of a provider’s model.
Customization and adaptation: Some algo providers offer customization options within their models, allowing traders to fine-tune parameters to better suit their trading objectives. The ability of a model to adapt to a trader’s specifications can be a significant differentiator.
Different liquidity pools
Diverse liquidity sources: The expansion of liquidity pools, including the growth of dark pools and mid-matching services, has provided more options for executing trades. These pools can offer different prices and depths of liquidity.
Specialized pools for algo trading: Alongside traditional liquidity sources, there are now pools specifically designed to cater to passive and algo executions. These pools often offer features like reduced market impact and anonymous trading.
Internalisation: algo providers who are also significant market-makers may possess the added advantage of accessing their internal liquidity. This internal pool may provide more favourable conditions for client algos, potentially reducing transaction costs in addition to the increased liquidity.
Smart Order Routing
The functionality of Smart Order Routing—deciding the best path for each order slice to navigate through the many available liquidity sources—can significantly influence the execution outcomes of algo orders.
This decision-making process takes into account factors such as price, liquidity depth, and the potential for slippage, aiming to improve trade execution. By intelligently routing orders, the execution quality in terms of both price and speed can be optimised.
Aggressive vs. Passive order placement
Deciding whether to place an order passively or to take (aggress) an existing order requires a balance of potential benefits and associated risks.
Capturing vs. Paying Spread: Passive order placement is traditionally associated with capturing the spread, as it involves placing an order at a price that is not immediately executable, hoping another market participant will trade against it. Conversely, aggressing orders—taking liquidity by fulfilling existing orders—typically involves paying the spread, which can be a straightforward but costlier approach.
Risks of Passive orders: While passive placement can be beneficial in terms of costs, it carries its own risks:
- Execution uncertainty: There’s no guarantee that a passive order will be aggressed by another market participant, which could result in the algo’s child slice not being executed within the desired timeframe.
- Market Impact concerns: The mere act of placing a passive order might influence market sentiment or reveal intentions, potentially driving liquidity away and worsening the eventual execution price.
The decision to place orders aggressively or passively should not be static but dynamically adjusted based on real-time market data, balancing cost, urgency and market impact. An algo that can intelligently blend order styles based on the prevailing market conditions can significantly enhance execution outcomes.
Conclusion
Market Participants should carefully evaluate the offerings of different FX algo Providers before deciding which providers to use. While having an FX algo offering is now a fairly standard part of the services offered by an FX market-maker, the offerings themselves are far from commoditised.
While fee levels will always play a role in comparing one provider against another, the differences in available strategies, model performance, available liquidity pools and smart order routing have the potential to have a far greater impact on execution outcome than a small difference in fees.
