AI and Best Execution: the Investment Bankers’ Dream Team

I don’t often blog about investment banking as it bores my readers who are cool and trendy retail bankers and geeks, mainly, but keep coming back to this article about JPMorgan’s Best Execution AI engine, LOXM.

LOXM’s job is to execute client orders with maximum speed at the best price, by using lessons it has learnt from billions of past trades — both real and simulated — to tackle problems such as how best to offload big equity stakes without moving market prices … JPMorgan believes it is the first on Wall Street to use AI with trade execution and said it would take rivals 18 to 24 months and an investment of “multiple millions” to come up with similar technology.

The reason I keep coming back to it is that best execution is a critical regulatory requirement, argued over extensively by the big bods of the European Union a decade ago when they were introducing the Markets in Financial Instruments Directive, MiFID. What the regulator wanted was a method to ensure that investment banks gave customers the best deal, but how to define what the best deal was became a big issue. Is the best deal the best price? Or is the lowest cost to process? Or is it actually getting the deal through the system in tact? Or is it ensuring the customer got the deal first, by being the fastest to process the indication of interest?

These four factors – price, cost, likelihood of settlement and speed – became the big debate of the first MiFID, and ended up being defined as:

A legal mandate that requires brokers to provide the most advantageous order execution for their customers given the prevailing market environment. Best execution encompasses several key characteristics that brokers must examine, track, and document when choosing how to route an equity, option, or bond order for execution. The broker must prove that, after examination of these characteristics, the broker utilized reasonable diligence in choosing how to route the order for execution. These key characteristics are the nature of the market for the security (i.e., volatility, communication availability, price, and relative liquidity); the number of markets examined; transaction type and size; and how easily a quote can be obtained.

This is a key part of investment banking operations and regulators have been tightening this definition more under MiFID II, which came into operation January 2018.

The new regulation mandates that “brokerage firms must take sufficient steps to ensure favourable execution of client orders as opposed to reasonable steps, and the added regulation that banks will have to publish on an annual basis their top five execution destinations by trading volume”.

Now let’s bring this back to reality. Most investment banks are trading billions of dollars a day on global markets and exchanges on behalf of their corporate clients, pension funds, institutional investors and more. Much of this is handled by algorithmic high frequency trading systems, that continually pour buy and sell orders into the markets based upon client instructions. Meanwhile, there’s some guy sitting somewhere who has to decide how to place the client order. He’s just got a buy-side order for $10 million investment in ABC Inc, and is thinking whether to place this on Nasdaq or BATS. Do we really think this guy can work out in a nanosecond which would be the best order route for the lowest cost at the best price with the fastest speed and guaranteed fulfilment?

No wonder LOXM is taking over and watch what all the other big guys do to react.