문 닫은 Quantopian, 집단지성 알파는 허상?

1.
문을 닫았습니다. 아주 오랜 동안 성장하는 모습도 보았지만 결국 수익을 내지 못한 듯 합니다. 아마추어퀀트시대를 열고 월스트리트의 민주화를 외쳤던 ‘Quantopian’이 막을 내렸습니다. Quantopian의 시작은 2011년입니다. John Fawcett가 설립자입니다. 개인투자자들이 스스로의 알고리즘을 연구개발하는 환경을 제공하고 전략을 수익을 내면 플랫폼이 투자하여 수익을 나누는 모델을 지향하였습니다. Quantopian은 이런 모델을 ‘Cloud Sourcing Alpaht’라고 불렀습니다.

Quantopian was launched off the back of that dream. A platform that taught users about quant investment and gave them a platform to write and save their own code, Quantopian was supposed to be the first crowd-sourced hedge fund. For years, users iterated on each other’s code as a community developed on the platform, full of users with and without financial backgrounds. These users would pit their algorithms against one-another, and Quantopian would go on to use the winning equations to manage investor assets, giving the winners some returns.

Fawcett and Bredeche would go on to raise $48.8 million for Quantopian in the meantime. In 2016, Steven Cohen announced that he would be teaming up with Quantopian to the tune of $250 million, relying on some of the user models Quantopian managed and investing in Quantopian itself.

In February, the first cracks in the city wall took hold. For at least two years, Fawcett said in an interview, Quantopian’s low-risk, market-neutral strategy model hadn’t been yielding results. Fawcett and Bedeche announced that the company would be returning investor money and switching strategies in an attempt to keep things afloat, asking that users now develop models beyond the market-neutral ones the company had relied on for years.
How Quantopian’s open-source investment dream died중에서

2010년을 전후한 때 Social Trading이 유행한 적이 있었습니다 특정한 투자자의 실거래내역을 보고 “따라하기” 혹은 “Copy Trading”을 하는 경우입니다. Croud Sourcing Alpha는 이런 개념을 헤지펀드 운용에 적용한 경우라고 할 수 있습니다. 멋진 출발이었지만 기대에 미치지 못한 듯 합니다. 2019년 이와 관련한 Fawcett의 글입니다. 방향전환을 이야기합니다.

Quantopian’s mission is to level the playing field in quant finance. We believe anyone, anywhere, should be able to compete to find market inefficiencies.

I’m writing today to announce a strategic pivot that better serves our mission. Through 2019, in the face of challenging market conditions, many equity quant themes underperformed as valuations continued to rise. The key learning after running our market-neutral strategy for two years is very clear: a more diverse set of investment objectives will better unleash your potential. To do so, we will shut down our market-neutral strategy and pivot our strategy in two ways:

By building bespoke strategies for asset owners like the US corporate pension that is hosting our first third-party challenge.
By researching and defining numerous products that span from long-only to market-neutral, cover many geographies, and operate at a wide range of turnovers.

Both of these initiatives translate into more opportunities for Quantopian’s business to grow and for you to receive royalties for your intellectual property (IP). We refer to these initiatives in combination as our “on-demand” strategy.

This change reflects Quantopian’s continuing evolution based on our learnings as a business and our belief that this pivot is the smartest use of our resources. On-demand is where the best opportunities for Quantopian and our community lie.

Many of you have commented on how constrained the contest and the selection criteria are. Crowd-sourcing alpha offers a limited number of allocations and ones that fit only strict criteria given that it requires limiting risk exposures and adhering to other constraints. In contrast, we’ve found that there are many interesting strategies that you have created that fall outside these constraints — now we’re exploring how and where they best perform.

On-demand challenges provide a broader range of complex problems for you to solve and offer a wider range of opportunities to get your IP licensed. As a result, on-demand gives you more opportunities to have your work recognized and rewarded. A large US pension fund is hosting our most recent challenge and we are actively pursuing other partnerships.

We have the right data, operators, signal selection, and combination methods to execute our plan for on-demand. But most importantly, we have you, our community members whose work we see as on par with that of researchers in the best firms in the world. Your submissions to our challenges have been high quality, diverse and demonstrate your ability to research complex problems in the industry quickly and diligently. Third-party investors find this, the 12 million backtests and 6 million algos already produced by you on Quantopian, extremely interesting.

The potential partners we’ve been speaking to rave about the talent depth within the community. Your commitment to our mission, trust in our vision, and the countless hours you spend researching and developing your ideas are what make Quantopian unique and on-demand possible.

The contest will remain running as usual and top performers will continue to receive daily prizes. We’ve found that that the contest is a great tool for identifying engaged community members and instilling essential concepts that continue to come up in quant finance. We do not anticipate that workflow and goals for activity will change much. We want you to continue researching your ideas and submit them to our challenges.

In the short term, our pivot allows us to highlight the crowd’s ability to perform world-class research for some of the largest asset managers in the world through third-party challenges. Longer-term, our pivot allows us to identify the criteria for what the right future strategy entails.

Our strength comes from the diversity of our community and we are committed to bringing our members the best growth opportunities we can find — we believe on-demand provides them. We are staying true to our original purpose: leveling the playing field for everyone interested in quant finance.

We are extremely excited about this and we look forward to continuing to work with you as we take Quantopian and the community forward. The team at Quantopian is tirelessly pursuing on-demand leads and we’re excited to update you on developments soon.

전략을 선택하는 기준을 바꾸고 플랫폼판매도 해서 Quantopian의 꿈을 이어가겠다는 이야기입니다. 플랫폼 판매를 위해 FactSet과 전략적인 제휴도 하였습니다. 그렇지만 결국 기대한 수익을 얻지 못한 실패한 모델로 되었습니다. 폐업에 즈음한 입장입니다.

“’I’m sad to say that we will be winding down Quantopian’s free community offering over the next couple of weeks and on November 14th, the community platform will be taken down . We know this is short notice
Crowd-sourcing alpha was a moonshot and I’m deeply thankful that we had the opportunity to pursue the dream together for 9 years.”

2.
왜 실패했고 얻어야할 교훈은 무엇일까요? 미국내에서 비슷한 사업을 전개하는 곳중 하나인 QuantRocket은 3 Takeaways from Quantopian Shutting Down에서 3가지 교훈을 정리합니다.

첫째 다중지성을 자신운용에 적용할 수 없다.(You can’t crowdsource alpha)

이를 보여주는 Quantopian의 논문인 All that Glitters Is Not Gold: Comparing Backtest and Out-of-Sample Performance on a Large Cohort of Trading Algorithms을 인용합니다. 논문을 Quantopian을 통해 이루어진 전략(백테스트 전략)들이 실제 운용성과로 이어지는지를 분석하였습니다. 결과는 신통치않았다고 합니다.

Quantopian’s founding idea was to apply the concept of crowdsourcing to investment management. Crowdsourcing has yielded impressive results in a variety of domains, but investment management turned out not to be one of them. The demise of Quantopian’s investment fund was foreshadowed by Quantopian’s own researchers in a 2016 paper titled “All That Glitters is Not Gold.” The paper examined a large sample of backtests created by Quantopian users and found that backtest performance metrics offered “little value in predicting out of sample performance.” Quantopian spent the next several years trying unsuccessfully to work around this basic problem.

Download (PDF, Unknown)

둘째는 트레이딩소프트웨어회사로 살아남으려면 작은 회사를 만들어라(The trading software market favors small companies)

If Quantopian indeed pivots from managing investments to selling software, they will almost certainly need to become a smaller company. Before this year’s staff departures, Quantopian’s head count stood at around 50. Back of the napkin, that suggests an annual expense line of $5-10 million. Had crowdsourcing proved a good way to manage Steve Cohen’s billions, such an expense line would have been pocket change, and Quantopian’s management team could have collected fat incentive fees.

The economics of selling trading software are very different. The trading software market is quite fragmented, consisting of numerous small companies (typically 1-10 employees) rather than a few dominant players. There is a good reason for this. Traders employ many different strategies. There are numerous markets, data sources, trading styles, and execution frequencies, and there is competition among traders to be unique and find the edges others are missing. Trading is fundamentally exploratory, open-ended, and creative. Trading software is not a fungible commodity. It functions more like a planetary rover, the specific design of which affects what experiments the researcher can perform. No single platform can be optimized for all use cases, and therefore no one software company can monopolize the market. This is a boon to new startups, but it sets a ceiling for potential revenue. Large companies need not apply.

위 문단을 읽으면서 한국을 떠올렸습니다. ZeroAOS를 서비스하면서 들었던 생각, 현재와 같은 서비스구조를 만들었던 때가 떠올랐습니다. 다만 미국과 같은 자본시장에서는 트레이더들이 운용하는 전략의 폭이 넓기 때문에 소수집중하라고 하였지만 한국은 시장규모가 작기때문에 소수운용하여야 할 뿐입니다.

마지막 돈을 내면 서비스가 달라진다.(If you depend on software, make sure you’re paying for it)

돈을 받는 입장에 서면 이용자를 대하는 태도가 달라집니다. 위 기사에서도 언급하였듯이 Quantopian은 무료 사용자에 대한 고려없이 그냥 문을 닫았습니다. 구글도 그냥 문을 닫았지만 예고를 한 점이 다를 뿐입니다. 이런 태도는 무료이기때문에 나옵니다.

이번에서는 Robotwealth 운영자가 분석한 My Thoughts on Quantopian’s Closing입니다. Crowd Source Alpha 전략에 대한 의견입니다.

Quantopian’s closure speaks to a fundamental reality: unique alpha is hard, particularly for the individual.

Professional teams have major advantages over independent traders:

  • A team to collaborate and share ideas with. No reinventing the wheel.
  • Good training
  • Technology that facilitates research and development

Quantopian’s vision was to eliminate these advantages. It faced significant challenges:

  • The constraints on the alphas that it could allocate to. Alphas needed to be unique, relatively slow-moving, and trade liquid instruments. That’s a difficult game, played by some of the biggest and most well-resourced players in town.
  • The inherent competitiveness among users in gaining an allocation meant there was always a good reason not to share your best work. That makes it hard to realise a community that is focused on collaboration for common benefit.

두가지 도전중 두번째. 서로 다른 이해관계를 가지거나 경쟁자인 플랫폼이용자들이 공동의 이해를 위해 협력할 수 있는 방법을 만들어야 한다고 합니다. 그러면서 자신들의 Labs을 장점을 강조합니다. 제가 보기에 Quantopian과 달라보이는 점이 없긴 합니다만.

이외 QuantConnect도 입장을 내놓았습니다. 다만 남의 불행이 나의 기회이기 때문에 ‘옮기는 방법(Migration)’을 소개합니다.

Migrating to QuantConnect

한국에 Quantopian과 같은 사업모델을 택한 곳은 없습니다. 비숫한 개념의 서비스들이 있지만 완성도에서 많은 차이가 있습니다. 다만 사업의 성공 혹은 지속성이 서비스의 완성도에 종속하지 않습니다. 비즈니스는 비즈니스입니다.

Leave a Comment

이메일 주소는 공개되지 않습니다. 필수 필드는 *로 표시됩니다

이 사이트는 스팸을 줄이는 아키스밋을 사용합니다. 댓글이 어떻게 처리되는지 알아보십시오.