TCA or not to TCA - Should (transaction) cost usually be an Algorithm's first execution consideration?
MiFID II has now firmly placed Transaction Cost Analysis (TCA) as a key execution consideration from the regulator's prospective, but for the buy-side clients in FX as well as in Fixed Income, market behaviour & market structure suggests that the Overall Quality of Execution is still what first matters
Some key considerations, among others, for an algorithmic order execution may include:
- Cost of execution, including known fees, spreads as well as price slippage (for the benefit of below discussion, I refer to these considerations as TCA)
- Order fill ratio achieved (until cancelled)
- Time taken to fill the order (until cancelled)
- (Perceived) Market information leakage (market movement directly correlated/ attributable to the specific client order being executed)
Optimising an execution algorithm for all the above four factors is what we strive for, but the fact remains, if a Client requires 100% fill on a large order, then there is bound to be higher execution costs, and depending on the size of the order & algorithm used for execution, longer (than desired) time taken to fill the order, invariably leading to some market information leakage.
A good example to bring to the table is one of the most liquid instruments in the FICC market - EUR/USD Spot
At a given time, a typical client has a choice of executing the EUR/USD Spot order on 100+ different venues, including Central Limit Order Books (CLOBs), Pricing Streams venues with (or without) last look (RFS/Q), Dark pools, Single Dealer Venues, and certain aggregators not to mention via voice with Voice Brokers, as well as Dealers directly.
Comparing the two more liquid venues types - CLOBs & RFS/Qs - Client can either:
- Place a Limit order on a CLOB, and subject to a Price - Time priority rule to achieve execution, or,
- Place an order on a RFS/Q stream, and subject to the last look rule of the stream provider, achieve execution.
The benefit of using limit order on a CLOB venue is known transaction costs & little to no price slippage. However, the downside being no guaranteed execution at that fair price & spread. Client can off-course submit an aggressive order in the order book, and this usually will lead to an execution, but then there is no "perceived control" on the transaction costs. Indeed aggressive liquidity seeking on order books is one of the last options for a client, usually used in illiquid markets.
In the RFS/Q market, although client may need to wait up to 100ms due to last look rules for the execution to be confirmed (10ms is where most rejections should happen), execution is usually "thought" to be highly achievable (75% or less overall fill rate will usually get the stream provider off-boarded from most markets). The downside being higher execution and slippage cost, specially due to the last look window when the stream provider may reject the order.
With the FX Global Code signed & implemented by the majority of venues and Liquidity Providers, last look rejection reasons range from either client running out of Credit Limit with the LP or the market moving against either the LP or the Client's benefit.
Worth noting is a research published by a FX venue that showed that they attributed $15 cost to $1 million order at 10ms last look window & $25 cost to $1 million order at 100ms last look window, in the RFS/Q market
Even with arguably higher transaction costs, market participation data shows (see BIS 2019 Triennial Survey) that majority of clients (D2C) increasingly using (disclosed) RFS/Q stream venues (compared to participation in undisclosed Limit Order books, where volumes have been dropping over the last six years).
Add to this historical trend, recent market volatility and more crucially in the Fixed Income market, some asset prices dropping by 10% or more intraday for the last couple of weeks, transaction cost (compared to a benchmark alone) is not client's first consideration.
To conclude, although transaction cost is indeed crucial, but overall quality of execution is still at the forefront of client's order execution strategy and must be prioritised by the execution algorithms by giving the client a range of choices for the execution optimisation strategy
#tca #algo #algotrading #algorithmictrading #electronictrading #efx #fx #fixedincome #productstrategy #productmanagement #businessanalysis