If you read Part 1 to this post (from December 15, 2016) then you know that at least as of the end of 2015, financial reports from HFT bellwether Virtu Financial illustrated strong and even increasing profitability. Our surprise from these impressive figures came from the countervailing hypothesis that HFT was already well past its prime (given the evidence of prop shop closings and consolidations over the past 5 years or so). Apparently, Virtu didn’t get that memo.
However, upon closer inspection of the most recent quarterly reports – which as of now yields details over 11 quarters starting in calendar Q1 2014 (March) and ending in calendar Q3 2016 (September) – even this bellwether may have seen its best days.
Exhibit 1 (below) is one perspective of what this recent turn of fortune looks like:
Some translation: After spiking in Q4-2014 and peaking in Q1-2015 at an annualized (adjusted net trading) revenue per employee (RPE) of over $4.1 million, trading revenue as of the end of Q3-2016 has returned to somewhat less astronomical levels last seen during the period 2011-2013 (of roughly $2.7-2.8 million RPE). Now, bear in mind for a minute that whether we are talking about RPE’s of $4.1 million or $2.7 million, both of these levels are more than 10x the average productivity level per employee of the largest global banks ($255,000 per employee in 2015), for instance – and still also nearly 2x the nearest observable comparable company (KCG) which for 2015 generated a still incredibly impressive RPE of $1.6 million (based on a combination of market-making and global execution revenues, whereas Virtu is almost purely market-making).
Our read of these figures: First, with this picture as our backdrop, it is important to make the point right now that – whether it is true that the trading strategies with the highest turnover (like Virtu’s) have finally peaked or not – the “technical leverage” being displayed here is extremely impressive. 148 people generating approximately $700 million in revenue in a year is basically unheard of. And, though typically not portrayed in this way (on an RPE basis), it is precisely this level of technical leverage that attracts so much attention.
Secondly, we are not going to try to solve – once and for all – the good vs. evil debate on HFT here. But, I do think that we can add some perspective given our modeling methods and unique views on technical leverage. Here’s one zinger to try on for size: Speed – or, increasing speed – is a requirement. We all know this. Being the fastest is a requirement to succeed with the highest turnover trading strategies. But, what else might be required to stay ahead? Oddly enough, I think it comes down to who you know… Let that sink in for a minute and then I will try to illustrate and explain. In addition to speed, success at the limits of trading speed comes down to speed plus who you know.
Here’s the main clue: Strength of profitability is localized in HFT. Virtu is a New York firm and earns the lion’s share of its revenue from New York area matching engines – and mainly in US equities. In any other category – from European equities to listed options and fixed income – Virtu makes good money, but nowhere near the dominance it seems to have from New York matching engines. Look at the rest of the membership of FIA’s PTG (Principal Trading Group) members . Think Jump Trading = Chicago, CME, futures. Think IMC = European options. Think DRW = fixed income. Although not a member of PTG, also think Wolverine Trading = Chicago, CBOE, options.
Persistent success at the limits of trade update frequency is not only about being the fastest, it’s also about possessing knowledge about the most detailed aspects of market structure, and even down to the components used in and the configuration of the hardware where a particular matching engine resides. To get to this level, it seems to come down to deep localized knowledge (i.e. – who you know).
Now, as for what happened to get to these (and other) numerical pictures – and what might happen next – we will save that for another day…