A Few Notes about HFT as a Business and the Virtu IPO

If Virtu’s IPO goes through as planned next week, there will be two HFTs on the public market. From my own experience, otherwise sophisticated people can get overexuberant about the business prospects of HFT. Here, I’ll try to give a bit of detail on some business risks that I think are often neglected. In addition to this blog’s normal disclaimer, I’m emphasizing that nothing in this post constitutes advice to buy, sell, not buy, or not sell a security or any financial instrument.

In contrast to the popular view, algorithmic trading is a very challenging business. This can be true even in the face of market inefficiencies that easily lend themselves to profitable trading strategies. I don’t have any specific knowledge of Virtu. But, in addition to the commonly cited risks like the Knight debacle or regulatory changes, there are obstacles to long-term success that are (I think) universal sources of anxiety.

Rapid Devaluation

The intellectual property of HFTs, by its very nature, devalues rapidly. One way HFTs create strategies is by analyzing patterns in market data. Over time, current strategies may leave detectable signatures that competitors can integrate into their own strategies, reducing profitability. Perhaps more importantly, the ultimate sources of trading revenue, an HFT’s counterparties (or counterparties of counterparties, etc) may adapt to the trading strategies, eliminating their edge. Both of these effects are fundamental to the operation of markets. One way to look at trading is that it consists of pouring information into the public domain in exchange for money. Eventually, you have dumped enough information into the market that your information source (in this case, your algorithm) is no longer of any value.

Virtu, of course, alludes to these risks in their S-1:

[A]doption or development of similar or more advanced technologies by our competitors may require that we devote substantial resources to the development of more advanced technology to remain competitive…
The markets in which we compete are characterized by rapidly changing technology, evolving industry standards and changing trading systems, practices and techniques.

The short lifespan of HFT algorithms (and infrastructure) might be manageable were it not for the mobility and negotiating power of employees. Clever, experienced employees are a rare resource and it takes many years for an investment in the next generation of talent to bear fruit. Virtu has ample resources (and public image) to compete for promising newcomers, but my feeling is that their current employee-related costs are lower than the competition. That could continue, but I think that there are strong forces that would bring it more in line with industry norms. Many commentators view the financial sector as a sort of worker’s paradise. I would argue that HFT is an extreme example of that; the loss or gain of just a few key employees could determine the fate of a company. A bank could potentially stomach even 100% turnover over a short period and still retain *some* value. I don’t think you could say the same for an algorithmic trading company, and the employees know it.

Virtu in 2013 reports a little under USD 80M in employee compensation, with a headcount of just 151:

Our low-cost structure allows us to maintain a marginal cost per trade that we believe is favorable compared to our competitors. Our scale is further demonstrated by our headcount — as of December 31, 2013, we had only 151 employees.

Optiver reported 738 employees in 2013, IMC reported EUR 176M in remuneration over 619 employees, KCG reported USD 437M in remuneration over about 1100 employees in 2014.

This suggests to me that Virtu employees are, in aggregate, at a mature stage in their careers. It would not surprise me if Virtu aggressively increases compensation in order to retain them.[1] The alternative may be that they go on a hiring binge with new recruits taking years to reach the productivity level of their current crew. The latter option involves letting some of the more expensive employees go to competitors, which may also increase the speed at which their current IP devalues. Even with non-competes and trade-secret covenants, experienced employees who leave undoubtedly take valuable intuition with them.[2]

Expansion into Wholly New Markets

The above pressures on profits could be counteracted with expansion into new businesses. There is discussion of opening up stodgy trading activities that are currently restricted to banks (e.g. swaps, spot fx, bonds/credit, MBS, insurance). If HFTs are able to get access to some of the core revenue drivers for banks, they may attain rapid increases in profitability, at least until competition starts to compress their margins again. This is, after all, the original story of HFT (and more broadly, “Financial Tech”): highly profitable but uncompetitive industries become electronic and open to new entrants, which can capture market share astonishingly quickly.

If this opening occurs, you have to ask whether current incumbents like Virtu will be the ones to capitalize on it. I obviously don’t know the answer to that, but it’s worth remembering that many HFTs are less than a decade old. Unlike other businesses, they have no customers and no brand loyalty. Their employees are savvy and often have savings sufficient to forgo a salary for an extended period. From my perspective, frontier markets offer plenty of opportunity to new firms. That is, if the markets aren’t just opened to those with large legal teams and a capacity to eat massive entry fees. Which, in my mind, is the upside potential here; If Virtu is able to gain access where others cannot, their profits could well expand many times over.

HFT is not the type of business where investors can even roughly estimate the value of a company years down the road. A big question is whether Virtu will survive long enough for investors to make their money back.[3] My guess is that Virtu’s current IP will be largely worthless in about 5 years. So really what an investor is betting on is Virtu’s management and its ability to cheaply retain, recruit, and produce effective employees. With Virtu seeking a valuation of about 2.6B on net income of about 190M, at their current rate they will need to go for about 13 years before paying back their investors. In HFT, that’s an absolute eternity.

[1] Virtu also seems to imply that they have implemented a greater degree of automation and efficiency into their business. This would allow them to offer competitive compensation to their employees, while still generating higher profits. If that’s the case, it seems like just a matter of time before other HFTs realize similar efficiencies and begin to poach employees, putting upward pressure on compensation.

[2] Virtu’s S-1 says:

We may not be able to protect our intellectual property rights or may be prevented from using intellectual property necessary for our business.  Our success depends, in part, on our ability to identify, recruit and retain skilled management and technical personnel. If we fail to recruit and retain suitable candidates or if our relationship with our employees changes or deteriorates, it could have a material adverse effect on our business.

[3] One commentator raises similar questions (paywalled):

Trading firms are difficult to value,” said Sang Lee, co-founder of Aite Group LLC, a research firm. “They need to explain not just how well Virtu is doing today, but what the firm will look like in 10 years. How are they going to diversify? Where is the momentum?”