Is Your Aggregator a Curse Rather Than A Blessing?
In 2015, I wrote an opinion piece about the inherent inefficiencies of re-aggregation of Prime of Prime feeds and I believe it is a good time to refresh this topic from a different perspective.
Costs of Re-Aggregation Far Outweigh the Benefits
- Increased probability of being picked off by clients
- Paying more to their bridge and technology providers than necessary
- Paying more in total to their LPs than necessary
The inherent inefficiencies and explicit costs of re-aggregation far outweigh the theoretical benefits; furthermore, the gains only benefit a broker’s Prime of Primes, technology providers and, for a short time at least, their clients.
Over time, the increased costs for brokers, and therefore reduced profitability, leads to them having to widen their own outbound spreads and increase commission structures, thereby disadvantaging clients in the longer term.
Despite what technology providers may say, in our experience, the larger and more professional the broker the fewer LPs they use, and the less likely they are to aggregate pricing.
It is the understanding of these key points that differentiate a good broker from a great broker and will, in aggregate, likely deliver a significant positive impact on overall P&L.
The Costs of Re-aggregation Are Meaningful but Not Always Obvious
Below I have outlined the explicit and implicit costs that brokers incur.
1. Reduced Book Revenue
Most retail brokers run a predominantly market making model, with STP business often being a secondary consideration. Deploying an aggregator directly reduces B book revenues in a number of specific ways.
Slower, less accurate price feed – The more complicated your structure, and the more LPs and technology steps that are involved, the longer it takes for the price to reach your trading servers.
As a result, brokers will be sending latent pricing to clients and are therefore vulnerable to being picked off by sophisticated or even lucky clients.
Re-aggregation, by its very nature, increases the probability of your price feed being behind the true market.
Price Latency – A major bank LP might send in excess of 500 quote updates per second in volatile markets. If a Prime of Prime takes quotes from 20 LPs, then they will be processing over 10,000 quote updates per second.
Each Prime of Prime therefore is processing a huge number of quotes to derive the best bid and offer, and each one is using different technologies and pricing configurations.
Each Prime of Prime also has a slightly different network and communications connection to the broker’s aggregation technology. This means that the prices processed by the re-aggregation software simply cannot be current prices.
If you then expand this by looking at the number of quotes that the re-aggregation software needs to handle, it is not difficult to see how this additional technology step, no matter how lean the software, must add meaningful additional latency to the quotes that end up going into the trading systems.
Adverse Selection – The adverse selection effect of aggregation means that a latent quote that is incorrect in the client’s favour (i.e. better than the best accurate bid or accurate offer) will always result in the broker losing money, but a latent quote that is not in the client’s favour will never be dealt on because it will never appear at top of book.
The combination of the additional technology step and the adverse selection of aggregating means that B booking on a re-aggregated rate is a perfect storm, because clients will be trading on latent prices (due to the technology), which are always in their favour (due to adverse selection).
Losses will be constant, and large in total, but will generally occur through small increments across many trades, rather than one big and highly visible loss; these losses are therefore very hard to detect.
Calculating, or even retrospectively identifying these losses, is not something that you can do easily but we have seen, time and time again, brokers moving from an aggregated to a single-provider setup resulting in a marked increase in profitability.
2. Explicit Costs
Re-aggregating Prime of Prime feeds will, in our experience, always lead to a broker paying more to their LPs. For example:
- Swap costs are higher when brokers have a short position with one LP and a long position with another (Prime of Primes who actively encourage end clients to re-aggregate are relying on this to make money – swaps often contribute the lion’s share of their P&L)
- Trading costs are higher, as the broker has to pay the spread to close long positions at one LP and short positions at the other
- Funding costs are higher where brokers are required to post margin to two or more accounts
- Overall costs from each LP are higher due to being a less important client to more brokers, thereby receiving wider spreads and swaps, and higher commissions from each individually.
Commissions and Connector Fees
Depending on a broker’s choice of re-aggregation technology, there will be commissions and connector fees to pay.
Many retail brokers use their bridge provider for aggregation and conclude, as a result, that they are not paying a fee (the commissions are no different if you aggregate or not), but the bridge provider will often be charging connector fees per LP.
This is an explicit cost that a broker incurs to use software that actually reduces profitability.