Algo Execution Path and Momentum Timing - Center of Mass

In this paper, we introduce the Center of Mass as a metric to summarize the loading path of an execution algorithm and evaluate the algo’s momentum timing capability. Algorithms distribute fills based on several factors. Market condition forecasts, including momentum, volatility, spread and liquidity, and benchmarks, dictate how fast and aggressive or slow and passive an algo behaves. A simple TWAP algorithm executes at a constant rate. In contrast, a more complicated opportunistic algorithm may capitalize on market liquidity, change execution speed according to short-term momentum forecast, and fill a larger portion of an order in a shorter time. The opportunistic algo may also detect unfavourable conditions and execute less aggressively. These behaviour changes may lead to different outcomes, and it is crucial to understand what is happening ‘under the hood' and how this can affect performance.

Please email contact@bestx.co.uk if you are a BestX client and would like to receive a copy of the paper, which is also hosted within our FAQ section of the UI.

BestX Algo Insights: Algo Density Intraday

Technology has revolutionised FX trading. It has been widely documented that increasing automation and electronification has reshaped the structure of the FX market. As algos continue to grow in popularity, it is essential to know when and how to best utilise them.

Our latest research will discuss a new metric - algo density which sheds light on the intraday pattern of algo usage.

Please email contact@bestx.co.uk if you are a BestX client and would like to receive a copy of the paper, which is also hosted within our FAQ section of the UI.

How long should TWAP algos run?

Our latest research looks at how the duration of TWAP changes with trade size and market conditions. We build a model that accurately reflects how TWAP durations change across currency groups and use this type of model in a number of interesting ways. In the article you will find:

• Full details of the model and how it is structured.

• How TWAP durations changed over the COVID crisis.

• How estimating a TWAP duration for a trade can build a new benchmark to judge decisions.

Please email contact@bestx.co.uk if you are a BestX client and would like to receive a copy of the paper, which is also hosted within our FAQ section of the UI.

How Does a Limit Affect Algo Performance?

Our latest research looks at the impact using a limit has on algo performance. By using double machine learning we find there is no causal link between using a limit and algo performance.

Please email contact@bestx.co.uk if you are a BestX client and would like to receive a copy of the paper, which is also hosted within our FAQ section of the UI.

Measuring an Algo Fill Profile – Centre of Mass

In our latest research article we present a new method for visualising the execution path of an algo and whether it executed more of the total notional at the start of execution period (front loading) or at the end of the period (back loading).

We condense this fill profile into a new metric, Centre of Mass, that can classify individual executions as either front or back loading. There isn’t necessarily any right or wrong as it clearly depends on the algo’s objective and the prevailing market conditions, but it can be useful to understand the behaviour when analysing an algo’s performance. This metric can therefore be used in conjunction with other BestX metrics (e.g. performance versus Arrival, market impact etc) to assess the performance differences across front and back loading algos.

Please email contact@bestx.co.uk if you are a BestX client and would like to receive a copy of the paper.

Assessing the Cost of Trading a Cross

Our latest research article attempts to answer one of FX’s long standing questions – is it better to trade the cross directly, or is it better to trade through the legs?

Analysing this problem empirically we find both intuitive and interesting results. For example, on average, we find that it is generally better to trade EURSEK rather than the USD legs, as many would assume. We also find that for some crosses, it can be more optimal from a pure cost perspective at least, that trading the cross (e.g. AUDJPY) can be better than trading the USD legs.

If you are a BestX user and would like a copy of the research article please contact us at support@bestx.co.uk.

FX Algo Performance during the Covid Crisis

In our latest research article, which is an update to one of our previous publications which analysed performance of different algo styles in a variety of market regimes, we specifically focus on algo performance in Q1 this year.

This period obviously incorporates the period of extreme volatility arising from the pandemic, and we compare performance to that experienced in more normal market conditions in Q4 last year. Interestingly, we find that the Get Done algo style when measured versus Arrival Price as a performance metric generally improved in Q1 compared to Q4 for EURUSD. However, Opportunistic algos appear to have underperformed compared to Q4, when measured versus the Interval TWAP benchmark.

We also find that the performance differences across regimes have increased in Q1, reiterating the need to make informed decisions when selecting algos to achieve best execution. In times of more volatile market conditions, the cliché of selecting the right tool for the job in hand is even more accentuated.

Please email contact@bestx.co.uk if you are a BestX client and would like to receive a copy of the paper.

BestX Case Study - An Introduction to Algos

Our latest case study provides an overview of all the different metrics that BestX can calculate for algos, with the aim to help clients identify which ones are more or less relevant to the performance, given the type of algo selected. There is no 'one size fits all' so having an understanding of what the metrics are measuring allows the client to decide which ones are most important for their trading objective and best execution policy. This is also crucial in order to ensure you are selecting the right algo for the right task, and judging them appropriately for how they are designed to perform.

If you are a contracted BestX client and would like to receive the full case study please contact BestX at contact@bestx.co.uk.

Algo Performance by Market Regime

Our latest research article is a further extension to our previous work on measuring and predicting liquidity and volatility regimes.

In this paper we investigate a practical application of the regime framework by analysing the performance of different algo styles within different regimes. As one would intuitively expect, choosing the right algo, or even style of algo, for the prevailing market conditions is an important decision component in the overall best execution process. We find that the performance difference across different regimes can be significant, for example, for the Get Done style, slippage vs arrival price can fluctuate by 2.5bps on average across different volatility regimes. We believe that trying to adopt a more rigorous approach to algo selection, that is both data driven but combined with trader intuition, represents a positive step forward.

If you are a client of BestX and would like a copy of the paper, please email support@bestx.co.uk.

Signalling Risk – is it a concern in FX markets?

"Should you find yourself in a chronically leaking boat, energy devoted to changing vessels is likely to be more productive than energy devoted to patching leaks."
Warren Buffett

What actually is signalling risk and is it something I should be worried about? This article will seek to answer both of these questions. As a topic, signalling risk has become more widely talked about in the FX markets following the move to a more order driven market, and the increased adoption of the use of algos. As discussed in a previous article, there are benefits to using execution methods such as algos , although there are also associated potential drawbacks. Signalling risk is one of these drawbacks.

In essence, signalling risk is effectively telling the market what you are about to do, perhaps inadvertently, and is also referred to as information leakage. For example, in a penalty shoot-out, there are occasions when the penalty taker effectively informs the goalkeeper of the direction of his intended penalty, as witnessed in the recent Italy v Germany quarter-final at the European Championships. Providing this signal in advance clearly gave the goalkeeper an advantage, and greatly increased the potential for a negative result for the penalty taker.

Information leakage is already a major concern in equity markets, with some studies indicating that an institutional equity order in the US now needs to run at a participation rate of less than 3% in order to prevent detection, whereas this rate was as high as 33% in 2007. This is supported by a separate study by Credit Suisse, which showed that VWAP performance starts to deteriorate when the participation rate starts to exceed 5%. As with a number of market structure issues, the equity markets can be a leading indicator of future developments within the FX markets, and signalling risk is no exception.

When executing an order over a period of time, such as via an algo, there is always the chance that signals to other market participants are provided before the algo is fully completed. Within FX, a good example of this risk is through the use of some algo types during the expanded 5-minute window for the WMR Fix. A paper published in April this year by Pragma highlighted this behaviour, where the authors found that ‘the rate change during the first minute of the window predicts a continuing rate change in the same direction over the subsequent minutes of the window’. Furthermore, they found that this behaviour was exacerbated at month and quarter ends. Such a pattern is potentially easy to detect and therefore allow other market participants to benefit from this knowledge.

Information leakage is not just a concern for those using algos over the WMR window. As liquidity has become more fragmented, and the depth of order books have become generally thinner, it has become easier for patterns in orders to be identified throughout the trading day. The increasing use of direct market access (DMA) via the use of smart order routers provided by liquidity providers and agency brokers contributes to the concerns as placing any order, especially if done naïvely or simplistically, can result in information leakage. How does this leakage contribute to a negative result to the original client order? Well, it is manifested within higher implicit costs through increased slippage. In order to demonstrate best execution, it is becoming increasingly important to take such factors into account when deciding how to execute.

Regulators are becoming increasingly concerned about the compliance risks posed by signalling, since signalling in effect invites front running and may lead to poor client outcomes and even market disruptions including liquidity air pockets and flash crashes.  Such events impact market integrity and harm the reputation of our financial markets.  Accordingly, where firms ignore the signalling risk posed by their algorithmic offerings or other execution methodologies, they may face claims for failure to pay due regard to interests of customers and/or failure to meet best execution requirements, especially where such signalling impacts client outcomes.  (See, e.g., FCA Principles for Business 6: ‘a firm must pay due regard to the interests of its customers and treat them fairly’).  In cases where signalling leads to market disruption, more fundamental claims such as failure to maintain orderly markets may be levelled.

 So what can be done about it? Simply deciding not to use order based execution is probably not the answer as such execution methods can contribute to significant cost savings as previously discussed, and therefore should form part of menu within a best execution policy. However, it is a risk that should be considered in such a policy and process, and the first stage in any form of risk management is measurement. Measuring actual signalling risk is obviously extremely difficult as it is not possible to isolate exactly what the response of the market is to any specific execution. In physics this is referred to as the ‘observer effect’, in which measurements of certain systems cannot be made without affecting the systems (exemplified by Heisenberg’s Uncertainty Principle). In the same way, it is difficult to know with precision exactly what the market would have done without your trade participating in it.

 However, it is possible to produce metrics which indicate the potential signalling risk that an order may have created, and hence, how easy or not it would have been to read by potentially predatory market participants. The key here is to compare apples and apples and use the same metrics, computed in exactly the same way, to allow fair comparisons across order types and providers. BestX have developed unique measures for this purpose, which allow users to compare the relative signalling risk across different order types. These metrics form part of the best execution suite within the BestX Post-Trade application, and use of such measures over time will allow users to mitigate the risk of information leakage. If a particular venue, or algo, consistently produces relatively high signalling risk in a given currency pair when analysed over a statistically large sample size, then informed decisions can be made, and justified, to alter execution choices and decisions.

 As Warren notes at the beginning of the article, it is probably best to change vessel once you discover you are in a chronically leaking boat. 

For further information on Signalling Risk and the available metrics, please contact BestX at contact@bestx.co.uk