For years, brokerage providers have been forced to address public perception over market making versus straight-through processing. Can a broker represent a counterparty's best interests if they aren't pushing each deal through to the underlying market in its entirety? We explore with Artur Delijergijevs, Quant Trading Manager at CMC Markets, why it's vital to delineate between popular internalisation of flow with rogue B-Book operations.
First principals - What's b-booking?
B-booking is a shorthand term which grew out of some leveraged market operators putting the business or flow where they felt the counterparties were guaranteed to get stopped out into a secondary, or "b-book" of liabilities. the idea was that by assuming close to zero risk, the broker could significantly bolster its profit margins.
However, this presents a significant conflict of interest, especially in over-the-counter markets where no centrally cleared price exists. Not only would an errant broker have visibility of its entire book, in turn enabling it to trade against accumulated client positions, but a blunter approach would be to simply slip an OTC price fractionally, resulting in either a margin call and/or a position being closed.
Understandably, tier-one regulators take a dim view of such behaviour, although it's difficult to police and is still used in offshore jurisdictions. It also comes with an inherent degree of risk - if a dealing desk makes the wrong calls, they can be left with spiralling liabilities.
Anecdotally there are suggestions that some of the early cryptocurrency platforms were unable to hedge their liabilities so as a consequence b-booked every trade. As the market kept rising, problems started to emerge.
So, all brokers should use an STP model?
Whilst there's certainly a transparency argument for using Straight Through Processing (STP) where each order is passed directly into the wider market, this comes with a range of issues. It removes a valuable efficiency for larger brokers, namely the ability to aggregate and match off highly correlated orders in the so-called internalisation of flow.
It also needs to be taken into account that every order placed via STP is ultimately ending up with a dealer at some point, presenting a range of sub-optimal outcomes, be that market impact, contention for liquidity or indeed presenting the opportunity for a broker to trade against the originating party.
What has popularised internalisation?
Some of the biggest prime brokers and tier-one liquidity providers have perhaps predictably driven the change here, and data from the triennial BIS reviews indicate that this has been going on for a decade or so. Improvements in technology - so the idea that large institutions have sufficient volume to create an "internal market" on a cost-effective basis - have been instrumental, but just as important has been the increased focus on reducing market impact, along with a better understanding of how costly this can be if done incorrectly.
We have discussed trade cost analysis before and the adverse impact of rejected orders - there may be some reluctance on the buy-side to always want to acknowledge this, but opinions are perhaps starting to change.
Does that mean there's a grey area - can risk management be a good commercial practice or end up looking like b-booking?
Arguably that could be seen as the case and for more than one reason. First up is closely correlated positions, so if two assets typically behave in a very similar way, a commercial decision can be taken to match off exposure against two similar, but not identical, instruments.
That however is astute risk management rather than b-booking and would need to be done within documented and audited risk protocols. Secondly, for the most frequently traded products, it can again make sense to have some surplus exposure on the book, allowing short term imbalances to be absorbed, again within the agreed risk metrics and all supported by the necessary capital adequacy requirements.
Critically however this is being done with the objective of reducing trading costs, rather than trying to exploit open client positions.
What about market impact?
It goes without saying that if an order can be matched internally, then there's zero market impact and that's something counterparties certainly ought to welcome. And even if the order can't be fully matched internally as the result of insufficient market depth being available, a split fill approach can be used and this, in turn, should ensure that the counterparty gets an overall price that is always better than would have been available in the underlying market in isolation.
Assuming the counterparty has sufficient clout, they will either be able to make a price at greater depth than is currently on hand or otherwise use their own balance sheet to ensure the trade can be placed on an anonymised basis.
And the ability to guarantee an order is filled?
Subject to the depth of the internal market, absolutely. There's no risk of contention when an order is placed like this, and by blending both internal flow as well as liquidity from the broader market, the counterparty ought to find that they are by virtue receiving a better price than would be available in the underlying market alone.
In summary, whilst there's plenty that's wrong with the b-book model, just because an order isn't immediately passed into the underlying market before being filled doesn't indicate malpractice.