Far from the Manhattan crowd

Over the last few years a confluence of technological developments in computing, combined with  new social and academic thinking in the worlds of decision science and behavioural economics, have led to some – primarily quantitative – hedge funds taking an innovative new approach to their research and idea generation process . Some firms refer to this new approach as an ‘external contributor platform’, others as ‘crowdsourcing alphas/trade ideas’, whilst others refer to the structure as being an integral part of their ‘Alpha Factory’^[1]. In this piece, the first of a two part paper, I discuss the forces that have made this possible, what it involves and potentially whether this could mean that today’s version of the rural worker in Victorian England, for example farm owners in the developing world, may have greater access to opportunities that until recently were unavailable.

Platforms, crowds, and a little behavioural economics

Platform businesses are taking over a range of industries: “Uber, the world’s largest taxi company, owns no vehicles. Facebook, the world’s most popular media owner, creates no content…”^[2] These businesses, which are largely thin-layered software interfaces, are winning, whilst established incumbents, which are more akin to what we have always seen as being ‘traditional’ businesses, are losing out.

These platforms are being assisted by the development of the crowd, a “term for the startlingly large amount of human knowledge, expertise, and enthusiasm distributed all over the world and now available, and able to be focused, online”^[3] . The crowd is generating the content, offering the cab rides and taking them, whilst in the world of crowdfunding it may even be pre-buying products so that it can get them designed and built.

The force that has enabled the power of the platform and the crowd is the freeness of the internet, its ‘no holds barred’ geographical range, and the speed at which it can reach anywhere with a good enough connection. The world has become a smaller place, with the internet enabling those in developing countries to engage with possibilities for work and progression that were previously out of reach. Samasource, for example, is a non-profit business that outsources digital work projects to workers in India, Kenya and Uganda. The company states that its goal is to “lift people out of poverty”^[4]; its mantra is “Talent is equally distributed, opportunity is not.”^[5]

In tandem with the development of the digital age the past two decades have seen the popularisation of the field of behavioural economics. Daniel Kahneman’s ‘Thinking, Fast and Slow’, a New York Times bestseller in 2011, introduced many to a range of psychological heuristics and biases that result in people placing too much confidence in human judgement. In addition, an earlier work by James Surowiecki, ‘The Wisdom of Crowds’, proposed the thesis that “under the right circumstances groups are remarkably intelligent, and are often smarter than the smartest people in them”^[6], putting forward the idea that a diverse collection of independently minded individuals, rather than crowd psychology per se, is likely to be more accurate than an individual, even an expert.

The growth in FinTech has also revolutionised many areas of finance – peer to peer lending, social foreign exchange platforms and wealth management robo-advisors have utilised the power of the platform and the crowd, backed up by the machine, to various extents. Hedge funds have long been users of the machine, but can they also harness the power of the platform and the crowd? If they can, will it disrupt the industry in the same way as Uber and Airbnb, for example, have disrupted their respective industries?

What is a crowdsourced trading signal platform?

Over the last few years a handful of quantitative hedge funds (WorldQuant, ) and start-up online firms (Quantopian, Quantiacs, QuantConnect, Numerai) have launched online platforms that allow the crowd to learn how to build predictive signals that can be used in a quantitative trading strategy. For such platforms, the crowd is primarily made up of data scientists, computer programmers, engineers, or anyone with the appropriate coding and mathematical raw skills that are required for someone to become a quantitative researcher at a hedge fund.^[7] Due to the reach of the internet this crowd is intended to be globally distributed.

The general structure of these platforms is relatively similar. Given the somewhat specialised nature of the work, guidance and training is provided. This may come in the form of recorded video lectures, self-paced tutorials and suggested reading materials. This support structure is intended to enable skilled quants make the leap to becoming fully literate in quantitative finance, a subject that has few high level university courses, particularly the further away one moves from major financial centres. The platform also typically provides access to number of free datasets, for example historical price data on securities, which can be used to build signals (alphas), as well as a ‘sandbox’ simulation environment to create backtests and a range of tools to monitor and risk manage the signals that are created. The backbone of the platform is, essentially, the basic structure of a quantitative hedge fund manager, although there are some notable differences. For many participants the major attraction will be the access to datasets and tools to develop their trading signals, as would be the case at a hedge fund manager. For others, however, it may be the intellectual stimulation and access to elements of a ‘quantitative community’ that the platform offers. Chat forums, online and in-person events, and online notebooks showing where others have succeeded and failed, help to create a sense of community and, most importantly, mean that the platform remains popular and active with a high number of registered users and an engaged crowd. The reward for participation varies from platform to platform. There may be an algorithm/signal competition with a financial prize, or the opportunity for the algorithm to be traded live within the fund or on platform capital, with an agreed percentage of the profits paid to the participant. In other instances, the algorithm may be linked or sold to institutional investors, or, in the case of the more traditional hedge funds, the participant’s contribution could lead to a lucrative job offer.

Who are the quants on these platforms?

These platforms have huge numbers of users. Quantopian, for example, reports that its platform has more than 200,000^[8]quants as members of its community. This raises the question, who are these people and why do they choose to trade through a platform, which will surely have some limitations when compared to working at an active quantitative research firm?

One of the primary reasons that many quants have embraced these platforms is that many of them enjoy getting quick numerical feedback on whether they have been successful or not, so financial markets can become appealing to work in even for quants who have traditionally worked on other sorts of problem.  Many quants don’t view financial workplaces as appealing, so this method of working may offer a sense of freedom, and it also offers variety – a freelance programmer, for example, could also be a freelance quant algo trader if working on a platform. Most importantly, these platforms provide global reach and freedom of location. In 2018 WorldQuant ran its ‘Alphathon’ competition in five locations in China, two in India, two in Romania, three in Russia and three in Vietnam^[9], whilst its overall community, alongside the communities of many of the other platforms, is truly global in scope. One of the more interesting examples is of a current WorldQuant part-time research consultant^[10] that is “from rural Taiwan and had earned an engineering degree but, because of political considerations and his desire to retain ownership of his family’s land, had instead become a farmer”^[11], before also working part-time for one of the largest quant firms in the world. This would not have been possible without the reach of the internet and the ability of web-based platforms to make access to opportunities more geographically equal.

Many participants are more akin to the intellectual hobbyist that genuinely has another full-time job; this does not necessarily mean, however, that they are making up the numbers in the community. Amongst its ranks Quantopian boasts Dan Houghton, ‘Burrito-Dan’, co-founder of London-based Mexican restaurant chain Chilango, whose algorithms have been selected for live trading by Quantopian. Houghton explains “It’s primarily intellectual curiosity but I’m fairly competitive and the chance of getting included [in Quantopian’s fund] makes me stay up for the extra hour working on a strategy.”^[12]

What effects is this new structure having?

There are clearly sound reasons why quants would choose this more flexible way of working in preference to a hedge fund office environment, and it is clear that the online-only businesses are committed to the use of the crowd in this way, but why are hedge fund managers going down the same route, rather than focusing on attracting the best people to work for them in New York or London? Obviously this is at least partially due to the potential power of the crowd, but the competitive landscape for hiring is also a major factor. Employees with quantitative and computing skills are in higher demand than ever before, with the battle for talent that was previously between Wall Street and Silicon Valley now also involving other asset managers and companies. This means that more imaginative routes to access talent, which is either part-time or located in more difficult to reach places, may be seen as an integral part of the hiring process for the larger quant firms that require a greater amount of talent. “Talent is equally distributed, opportunity is not”^[5] is clearly the mantra of this business model (though sadly it has come somewhat late for Hardy’s farmer in Victorian England).

Alongside the global crowd being accessed through the platform, and potentially hired as virtual research consultants, WorldQuant now has 25 global offices, the majority located in Asian and Eastern European cities that are not considered to be traditional financial centres. In addition, Trexquant, a firm launched in 2012, now has two offices in China and one in India. Indeed, it’s possible that newer firms will look to focus more on different regions that are currently relatively untouched by this model.

As much as there is greater global reach in hiring talented individuals with the requisite technical background, more could still be done to improve access to education for people with the right aptitude but who are restricted financially or geographically. In tune with this idea WorldQuant has set up WorldQuant University, which runs a free two year master’s degree program in financial engineering where over 50% of the students are from sub-Saharan Africa. There is no structural link between the university and the investment manager, with the aims of the university being more philanthropic in nature, though an improvement in access to education will clearly help increase the size of the crowd that has the requisite quantitative skillset to develop trading signals. Maybe this could pave the way for other hedge fund managers to build platforms, embrace the crowd, and ultimately educate people outside this initial crowd.