Aggregation & Spreads: The Race to Zero

Barry Flanigan, head of Electronic Trading Solutions (ETS) at IS Prime, talks to eForex magazine about the pitfalls of purely focusing on spreads.

As the eFX market has developed, the main (and in many cases, sole) metric that many firms have been looking at when assessing their counter-parties is spread - Barry discusses the issues that this can cause including execution, book P&L and long-term performance with E-forex in their April 2020 release. 

The ‘race to zero’ has been a fact of life since e-FX started in around 2001, but as the market has been heading towards choice [zero] spreads throughout 2019 in majors, we’re seeing vibrations as the industry struggles under this breaking point. Downstream, Prime of Primes (PoPs) have been battling it out with their peers in a bid to offer the tightest possible spreads, which has been the focus and ultimate tool to wrestle business from the prized retail space. Initially the equation was simple: more market makers = more competition = tighter spreads.

However, the law of diminishing returns has finally led us to an important problem that can no longer be ignored.

In defence of retail brokers, who are in many ways pure, unadulterated marketing machines, headline spreads are understandably one of the premium weapons in their arsenal to grow their market share.

At the click of a button on comparison sites like myfxbook.com, retail traders can quickly view one broker’s offering compared to another. Therefore, having the number one spread offering is seen as a necessity by brokers in order to grow, and in many cases, to survive. In this article, I want to explore the effects that spreads have on both retail brokers and liquidity providers, and whether we can debunk some of the myths that “only spreads matter.”

In the above example, the slowest rate is a ‘winner’ on the offer, which means the client is buying significantly cheaper than the fair market represents

Broker comparisons

One of the main issues with the ongoing obsession with spreads is that there really is no 100% true basis for comparison between liquidity providers, particularly if you are making the decision solely on spreads alone, and not least without asking the correct questions.

Most technology venues throttle quotes into client systems. For example, MT4 can only handle about 4 updates per second, whilst many aggregators handle 100s or 1000s of updates per second, per pair. Picking which quotes get published and which ones are dropped can be extremely important (i.e. is it the update with the 4 best bids and offers, randomised, the most recent quote received?). There are ways to manipulate this value and one needs to be mindful of this when deciding which technology layer to do the recording. Quote sessions (pricing) are obviously different to trading sessions (the session where the trade is placed, and the actual market price is received). Assessing spread recordings in isolation takes the focus away from the other, arguably more important variable: execution. Furthermore, with modern technology, it is fairly easy for brokers to deploy the ‘bait and switch’ model.

This can be done in more ways I care to count, but might include:

a) Putting a negative mark-up on quotes yet slipping execution i.e. actual spread averages are lower and therefore more appealing, but at the cost of client experience. Not only is this damaging for the broker’s reputation but it creates an operational nightmare in dealing with the obvious impending slippage queries from clients.

b) Forcing spread caps to keep averages artificially low (a suppression of real market rates to a predefined cap). This is usually done around EOD, when spreads naturally blow out as market makers re-price for the new day, take stock of risk and inventory, and reset the connection for a new trade day.

c) Aggregating as many quote sources as possible. This tends to deteriorate execution metrics (including slow execution speeds or increased rejection rates) or is damaging to true market makers by causing unintended market impact.

d) Filtration levels are another important configuration that affects spreads but can alter experience. Most brokers have a setting to prevent passing on wide spreads, preventing stopping out clients from erratic spread spikes caused by market jolts or illiquidity that doesn’t “make sense”. Whilst filters can be valuable to retail brokers, they are open to abuse by superficially suppressing wide quotes and this impacts real spread averages. The other downside of this is that if a client wants to trade, there will be nothing available, arguably at a time of need when de-risking from extreme price moves.

There are many other techniques that less-scrupulous brokers use to artificially bring in their spread, but it is important to recognise that there may be more than meets the eye. It should also be noted that these technology configurations can be used for good. Filtrations can be extremely beneficial when used correctly and are an important safeguard against extreme volatility, as evidenced by the SNB crash of 2015. Finally, as many of us know from our regulator(s), Best Execution requirements are starting to accept that price is only one factor when analysing overall execution and duty to clients.

Adverse selection 

A major by-product of the race to zero in the retail broker space is the continued push for brokers to aggregate Prime of Primes or just infinitely add liquidity sources.

Rather counter-intuitively, it seems that the smaller the broker, the more likely they are to try to aggregate multiple providers, as many do not realise the long-term negative impact that aggregation can have on their businesses.

Again, we understand why this is the case - they are trying to compete with the larger firms whose relationships and economies of scale allow them to demand the spreads they need to win new business. What these smaller brokers might not realise is that in the long-term this is to the detriment of their firm as a whole - both in terms of their spreads widening due to “adverse selection”  or “the winner’s curse” (something covered very clearly in a series of videos that Deutsche Bank put out last year) and indeed the overall impact on their own P&L.
In short, the more providers they aggregate, the more likely that one of their LPs [or rather the underlying LPs] will have a price that is incorrect, most typically by being slow or latent. This virtually ensures the LP pricing this flow will incur “inception losses” when taking this trade, as it is slow to the party of a moving market.
 
Far more concerning for the retail brokers are the ones taking risk on their own book. Adverse selection by adding “bad liquidity” with a single-minded drive on tightening spreads leads to “incorrect risk” being taken on their internal book. 

Statistically, brokers with hybrid books trend towards an internalisation rate averaging 70-95%. By failing to understand the dynamics of bad quotes in the liquidity pool, this cost is predominantly taken on by the retail broker itself. In many cases, this can go completely unnoticed as it is notoriously difficult to detect, yet directly impacts profitability. This is further ensured given an underlying liquidity provider with slow quotes will pick up more volume irrespective if they have poor absolute spreads, and therefore may actually appear to be a valuable addition to the broker.

Brokers and their book P&L

The general attitude still remains in the industry that “all retail traders lose money”, mostly irrespective of market spreads.

Statistical observation suggests something entirely different. Over the long term, clients tend to only lose their transaction costs. The impact of tightening industry spreads has caused most brokers to show deteriorating returns on a dollar per million basis over the past few years.

In fact, prior to the market volatility resulting from the horrific COVID-19 impact, broker DPM yields had reached all-time lows. This has subsequently spiked but it is almost all due to wider spreads and not a deterioration of the retail market making “bad” trades (alpha trade signals remain relatively neutral).

To add to the marginalisation of P&L due to spread compression, we  also observed that brokers’ ability to yield large DPMs have been affected with the introduction of negative balance guarantees, and leverage restrictions have removed conditions that had previously skewed profitability towards the broker.

The stats don’t lie, and quite recently one of the largest retail brokers in the industry moved to a spread capture model and away from the traditional warehouse/b-book model based solely on client drop.
If not spreads, then what?

I’ll acknowledge the above topics are dangerously close to dismissing the value of excellent pricing and spreads. While we [IS Prime] have been striving for excellence in the marketplace, with spreads being at the epicentre of our product offering for the past five years, I believe that this is simply a by-product of a well-managed broker.

Technology is at the core of any good broker and underpins their entire offering. New competitors come and go with regular occurrences, and the “new kid on the block” gets privileges that include the benefit of the doubt with LPs (aggressive pricing) and new age technology. However, without the aptitude of improving the technology, managing liquidity correctly, and resisting the urge to grow the business with easy to win flow (i.e. toxic flow), the company can run a short lifetime.

Price consistency runs very closely off the back of technology. Looking at spreads day-by-day can create a lot of noise, chasing the tail of where the best pricing can be sourced. History would suggest the best brokers excel over long periods and maintain relative competitiveness through easy and arguably more importantly, through difficult periods of pricing (as we are experiencing right now). This is only really achievable through solid, mutually beneficial partnerships with clients and market makers over time. In fact, with the current volatility, marginal spreads haven’t even been a consideration to clients. The most important thing has been the right configurations and settings to ensure a fair and steady supply of liquidity so that clients can trade. 

Client support is another huge, undervalued factor. A good Prime of Prime should provide detailed knowledge on the trades to assist you with client and regulatory transparency, strong technical support in areas such as connectivity and performance optimisation, and market intelligence to help retail brokers to improve performance and deliver their product to market. They need to provide so much more than just the ‘commoditised’ product that liquidity is often seen as these days.

To learn more about IS Prime and our products and services, please feel free to reach out:

Read or download the article in PDF format by clicking the image below: