Buy-side Perspectives Issue 17 | Page 21

Which are your top 3 "DARK" algo providers? 4 The list of 20 suppliers has been filtered down to top 11 for illustrative purposes td. ing L Virtu ight: r Copy Therefore, the industry needs to find a new way to measure the value of a block trade. At Liquidnet we have developed a model called “Value of the Block” which uses the sum of three clear metrics; spread, market impact and incremental Alpha as metrics for evaluation. This model is open, transparent and uses metrics freely available to all market participants. Firstly, spread: The vast majority of executions on the Liquidnet platform, including human-negotiated trades, occur at the midpoint, therefore saving the participants 50% of the spread. This is the first significant saving of trading a block. Secondly, market impact: Executing an order via a single large block, in a dark venue with natural liquidity exhibits very little market impact (as highlighted by the price movement statistics quoted earlier), therefore it is fair to say that block trades save the participants the equivalent of the pre-trade market impact estimate. At Liquidnet we use our own internal market impact model based on the Almgren model, which is commonly used across the industry, even though any pre-trade model may also be used. Finally, incremental Alpha: Executing in a single block trade means the quantity does not have to be executed in the market over x number of minutes, thereby reducing any timing risk that may exist. This risk can be measured using a PWP at a certain percentage. Our default calculation is 10%, highlighting the out or under performance vs. this realised price. Combining the three measures gives a single basis point and dollar number that can be used to quickly and easily ascertain if trading the block was the right thing to do. As an example, the largest LIS execution in October (103m USD of EXPN.L) saved 2.5bps of spread, 32.2bps of market impact and 92.8bps of timing risk, with an expected duration of over 19 days. How do blocks improve algo performance? Blocks will not be appropriate for every algo order, however, when trying to achieve the arrival price, either via dark Winter/Spring 2020 K&K ult Cons l a b o Gl aggregation or a liquidity-seeking strategy, then trying to find block trades can be beneficial. We analysed all Liquidnet dark and Barracuda orders between 1 st July 2019 and 31 st October 2019 to see what difference block executions had on performance vs. arrival and the results were striking. Orders that executed >75% of the total order notional as LIS executions outperformed the arrival price (mid) by 1.65bps. Orders that executed <75% missed the arrival price by -2.12bps 3 . Due to the full access Liquidnet Dark and Barracuda have to the block market, i.e. connected to all four of the main conditional venues (Liquidnet, CBOE LIS, Virtu Posit and Turquoise Plato Block Discovery Service), these algorithms are fully exposed to the opportunity of block executions. The performance of these algorithms was recently recognised in a poll produced by K&KGC 4 which highlighted Liquidnet Dark as the most popular dark aggregation algorithm by buy-side traders across Europe. Barracuda was also in the top-3 liquidity seeking algorithms. Happy hunting With a robust model to measure the quality of execution and whether LIS block trades achieved Best execution now available, the hunt for LIS blocks is sure to continue. It is therefore important to ensure you’re fully exposed to the block opportunities that exist via your execution tools, in order to find liquidity and achieve Best Execution. 1 Liquidnet Analysis – Bloomberg dark MTF, on-exchange executions, 1 st – 31 st October 2019 2 Source: Big XYT Liquidity Cockpit – January 1 st to October 31 st 2019 3 Source: Liquidnet Dark and Barracuda algo orders between 1 st July 2019 and 31 st October 2019 4 Source: K&KGC, The Buy-side Perspectives 2019 The top 3 equities algorithmic trading brand preference ranking – December 2019 www.buysideintel.com 21