I often receive emails from individuals who are interested in joining High-Frequency Trading (HFT) firms. They are sometimes confused as to how to go about applying for roles and are unaware of the technical skills necessary to obtain a job. I've written this article to explain what HFT is, what type of skills are required to get hired and who to approach when looking for a career.
Be aware that HFT is an extremely technical discipline and it attracts the very best candidates from the fields of mathematics, physics, computer science and electronic engineering, often at the grad school level or with years of industry expertise in a niche area. Obtaining a role in an HFT firm, while often highly lucrative, will take a significant investment in terms of study and effort.
What is High Frequency Trading (HFT)?
The term 'HFT' covers a wide range of activities in algorithmic trading. For the purposes of this article it means executing trades at extremely high volumes over extremely low latencies. "Low latency" in this instance means acting upon information received from market data feeds and then finalising a trade on the sub-millisecond scale. In fact, the "bleeding edge" of the top HFT firms measures trade latencies on the sub-microsecond scale. This latency is only set to decrease as more sophisticated customised hardware becomes available.
HFT is an extremely secretive discipline. It is very hard to find out information about how HFT firms operate. Job postings, vendor marketing pages and the odd internet article do provide some insight, however. Ultra-HFT (UHFT) is also very distinct from other forms of algorithmic trading. It is highly technically driven and extremely quantitative. There is almost no discretionary input once an algorithm has been deployed (until it becomes unprofitable!), which is in stark contrast to lower-frequency systematic trading where there is often some human judgement mixed in.
It is also a very competitive and often-times disheartening environment. Months of research can be thrown away in a day if an exchange modifies its architecture, a new regulatory environment surfaces or a competitor is able to exploit a process at a rate faster than you are. For this reason it suits highly technical, disciplined individuals who crave autonomy and a collegiate environment of extremely capable people, while acting under a decent amount of pressure.
An Entrepreneurial Mindset
Many HFT firms are relatively small companies, often with a low headcount (~20-25). This means they possess a strong entrepreneurial culture and a meritocratic mindset. Any HFT firm will be questioning what you as a candidate can bring to the table that doesn't already exist in the firm.
Given that the bonus pool is shared by many employees (albeit in a weighted manner!), you will need to demonstrate an ability to generate revenue (either directly or indirectly) that exceeds your salary and bonus share, otherwise it clearly isn't worth hiring you. This means you need to possess a set of unique skills that the firm doesn't currently include, in order to even be considered for a role.
The flip-side to this process is that often you will be able to "create your own role" within the firm. The firm might not even be hiring, but if they feel that your skills in a particular area are strong enough they may create a position for you. The meritocratic approach of HFT firms usually allows significant autonomy in your projects. Thus if you wish to work with extremely smart and capable individuals, in a self-starting environment, then HFT is probably for you.
Such roles often come with longer hours than many might be used to. 60-70 hour weeks are not uncommon when project deadlines need to be met. The fast-pace, intellectual stimulation and compensation generally outweigh the workload, however. This may or may not suit your desired lifestyle!
Paths into HFT
There are a few paths into HFT, but most of them require extensive technical skills in one or more of the following hard sciences such as mathematics, physics, computer science or electronic engineering. Individuals often join HFT firms via:
- Grad School - Many HFT candidates are employed straight from grad school in the relevant area. This is simply because it is easier to assess a candidate's ability based on their doctoral work/modules, publications or (perceived) quality of their school. Thus if you are really set on a career in HFT, then carrying out research into low-latency systems is likely to be a good way in. It is also not unheard of for the best undergraduates to be hand-picked from the top technical schools (MIT, Stanford, Cambridge, Imperial) and then "trained up" on the job.
- Industrial Expertise - Experts in a particular low-latency industry such as telecoms are often brought in for their particular domain knowledge. Although, generally in order to function well in the initial domain these individuals have a deep technical background (usually academic) anyway. Scientists working on High-Performance Computing projects (such as the data centre at CERN) or other national super-computing laboratories are also highly sought after, by virtue of their "big data" experience.
- Financial Exchange Experience - Any individuals with insight into the inner workings of the exchanges being traded on will be highly sought after as they are likely to be able to help carry out research into new algorithms that can make use of the exchange architecture.
One common misconception is that it is a hard requirement to possess an extensive background in finance to apply for HFT roles. Most HFT firms are actually indifferent to your knowledge of finance, assuming that you have extensive technical expertise elsewhere that they can make use of.
Common Skills
The roles at an HFT firm are quite diverse. Nearly everybody in the firm will have a highly technical background and will be capable of independent research in that field (i.e. is likely to be academically trained). Since HFT is essentially a "technology sport" many will have backgrounds in computer science and electronic engineering or low-latency expertise from backgrounds in other industries such as telecoms.
It is also becoming more common to find individuals with deep expertise in certain types of hardware such as Graphical Processing Units (GPU) or Field-Programmable Gate Arrays (FPGA).
Essentially, any skill that can in some fashion reduce the latency of the trade lifecycle or increase execution speed of algorithmic calculations will be found in HFT. Examples of such expertise includes:
- Exchange Architecture - Extensive knowledge of how trading exchanges operate is a common skill found among high frequency traders. Being aware of how the order book operates, as well as the intricacies of the technology stack in a particular exchange can put you at a distinct advantage.
- Processor Design - HFT involves substantial volumes of calculations over an extremely short time scale. Being aware of any means of advancing the execution speed of these processes will be advantageous. Deep experience with hardware and processor design, especially on systems other than commodity x86 hardware (alternatives such as GPUs and FPGAs) is useful.
- Low-Latency Networking - A large part of the trade lifecycle latency comes from the networking stack. Experience optimising packet processing, writing custom networking modules or use of Infiniband switched fabric networks is also attractive.
- Regulatory Understanding - Being highly aware of the trade regulatory environments such as Regulation NMS in the US and MiFID in the EU is essential to HFT operations.
- Kernal Optimisation - The common theme among these requirements is reduction in latency and increase in execution speed. Thus it is now commonplace to rewrite parts of the software kernal in order to gain a speedup. Experience in Linux kernal modification is beneficial to many HFT firms.
- Online Algorithms - While I've mentioned latency/execution I've not talked too much about HFT algorithms. These often consist of "rolling" calculations for means, variances and linear regressions. Thus any prior history of high-throughput time series analysis expertise is beneficial.
- Programming Languages - While many UHFT firms have moved towards custom hardware (both for processing and networking), some less latency-sensitive systematic trading firms still make use of multithreaded C, C++ and Java (with custom garbage collection). Extensive experience in any of these languages and parallel processing will be attractive to certain firms.
As can be seen, these skills are often deeply technical and require either a grad school level of involvement or years of industry expertise in certain technologies. If your skillset intersects with any of the above areas then you should find that you'll be able to score some interviews with HFT firms.
Looking for a Job
As with most quantitative roles in finance the best way to gain a job is through recruitment agencies. The top equity-based HFT firms are generally located in New York and London. Chicago is also a large hub for commodities/derivatives HFT. The good recruiters are often relatively well-versed in the domain and will be able to advise you as to whether your background is suitable. Be aware though that the bar is set rather high! You will likely have to work hard to find a role and it could take some time.
While direct application to such firms is possible, the tricky part is figuring out which firms actually take part in HFT! Often, if you are well-known in your particular technical niche, the firms will try and recruit you directly. Thus it can be advantageous, if you are really keen to join such a firm, to try your hand at publishing work, attend/talk at some conferences and generally raise your profile.
If you have any questions about HFT or quant careers in general, feel free to email us at support@quantstart.com.
