Co-founder of Superalgos.org, an open-source project building a Collective Trading Intelligence.
This piece explores how future trading collaborations may build and deploy trading intelligence given the proper technology infrastructure and a fitting design of incentives.
The author is a co-founder at Superalgos.org
A few weeks have passed since a group meeting at the now-famous subreddit hijacked the world’s attention with a collective action that is likely to go into the history books. The crowd joined forces to squeeze hedge funds out of their short positions against several companies.
The story is a lot more complex than that, but that’s what you’ll get from the mainstream media.
You may also read about the abusive behavior of certain hedge funds; how the system is rigged and needs more regulation; how the little guy tried — but ultimately failed — to stick it to the man; how Robinhood took arbitrary actions and betrayed their users costing them billions; how most people in the crowd ended up losing money; how Wall Street may change with the advent of coordinated actions by retail traders…
And it’s in that one last department that I’d like to focus on, because the mainstream analysis is falling way short.
“No, Wall Street Bets is not a revolution” — wrote the Financial Times three days into the craziest price action seen in decades, dismissing the movement as “people [that] have found a way to get rich quick, and are doing so. Nothing more, nothing less.”
A more thoughtful John Egan wrote a three-part article series on Forbes analyzing the nature of the events from the perspective of digital communities, exploring the idea of “How A Mob Became A Movement”. He concluded: “That momentum and energy could be directed, by the right leadership, toward continued financial disobedience. Alternatively, it will dissipate leaving us with little more than memes, Reddit Karma and memories of the time that a smoothbrained subreddit took on the system and almost won.”
CNN reports how hedge fund managers are running around trying to figure out the setup to monitor what happens on Reddit: “hedge funds especially need to know if their short bets are being discussed on WallStreetBets, indicating a potential short squeeze in the works.”
What everyone seems to be missing is that this is just the beginning! The crowd at WallStreetBets was using a tiny percentage of their potential!
What if the crowd didn’t rely on such a rudimentary communication tool? Messages in Reddit get buried under hundreds of posts in a matter of seconds, never to be seen again!
How would such a crowd interact if valuable pieces of intelligence could rise above the noise of a karma system that rewards epic memes over insight?
How efficient at research and execution would a crowd become had they access to the proper technology infrastructure?
How would have things turned out hadn’t the crowd been relying on a centralized platform (with a questionable business model) that left them hanging out to dry?
It’s not rocket science to come up with a better collaboration infrastructure, one that would elevate the crowd one step closer to their true potential.
However, I will take the mental exercise much further than the next evolutionary step. Let’s imagine what kind of setup could truly level the playing field with hedge funds.
Before WallStreetBets, big players would have agreed that collaboration in this context is an absolute impossibility. These players tend to see the markets as a zero-sum game. That is why trading intelligence is developed in silos, in absolute secrecy. In their eyes, it’s a purely competitive game. Such is the entrenchment of this belief system that even within firms quants work in small groups competing with each other.
The landscape wasn’t that different in the retail segment. Traders find it hard to envision a collaboration — particularly among strangers — because issues of trust become insurmountable. We witnessed such issues with the events by WallStreetBets: the game theory at the individual level conflicts with the interests of the collective. The individual was incentivized to take profits and sell the meme stock at the top, while the collective goal was to keep holding to tighten the squeeze further, at least until put options betting against the stock in the future expired out of the money.
There is no way out of the conflicting interests of individuals and the crowd when attempting a collective trade. It’s an ill-conceived collaboration!
Traders don’t need to trade in unison to collaborate. Instead, the collaboration should go towards developing the capabilities required to level the playing field.
By Evdokimov Maxim at Shutterstock.com
It is not only access to huge amounts of capital that tilts the table in favor of big players. It’s the asymmetric access to information, the huge gap in data analysis capabilities, the access to state-of-the-art algorithmic trading technology.
The smart form of collaboration is to work towards increasing those capabilities. The goal is to multiply the odds of the crowd at generating competitive trading intelligence.
It’s not about the match, but the championship. It’s not about squeezing hedge funds out of their shorts. It’s about competing on a level playing field, every day. It’s about beating big players in the long run.
Even before civilizations came to exist, tribal peoples traded goods and services. Free markets are an emergent phenomenon — no one invented them.
In the modern context, we can look into how the big players do what they do. The consensus is that information is king.
The ultimate predictive tool would be a model that considered every variable affecting the global economy, but simulating such a complex system is completely out of reach. Science has managed to model with some degree of certainty how the weather works, and knowing the laws of physics has a lot to do with such an accomplishment. But we’re not even close to understanding the laws that govern human interactions at the multiple levels the global civilization is organized.
That said, we all have intuitions on how the economy works. We all have an imperfect model. We can sense what is going on, take information in, interpret it, and make predictions as of what may happen next. With experience and the hindsight of past predictions, we may further develop and improve our model.
Needless to say, the more information you may be able to sense, analyze, and feed the model with, the more intelligent the model gets. And also, the more predictions you make and verify after the fact, the more you learn about which predictions work.
This is one of the departments where the big players excel.
By Immersion Imager at Shutterstock.com
Their first edge derives from the sophisticated data analysis technology infrastructure they develop in-house for their own use. That is how most math and computer science PhDs end up working for hedge funds.
The second edge derives from how they put all these predictions to work. They too develop trading automation technology, deploying unlimited numbers of trading algorithms, each working with multiple variations of multiple strategies, feeding from multiple data models, on multiple markets.
Big player’s capacity to analyze data, make predictions, improve their models, and take action on their predictions is several orders of magnitude higher than any of the individuals in WallStreetBets.
But how do big players compare with the collective? — Pretty much in the same way.
Wouldn’t a big enough collaboration be able to generate as much intelligence? — Yes and no. Not in the case of WallStreetBets.
For a collective intelligence to emerge, a crowd must be able to seamlessly aggregate the individual intelligence of its members. Individuals in the crowd must have frictionless access to the intelligence created by the collective, so that they may build on top. And they also need to be able to circle back to the collective any intelligence they may produce.
This is clearly not the case with WallStreetBets. By exchanging information over a subreddit, they’re effectively wasting the collective’s potential.
Intelligence is in the information resulting from processing data. It is in the models deriving from the information. It is in strategies that describe how to use the information in the models. It is in the execution of those strategies.
Needless to say, such types of assets may not be transferred effectively as text messages on a subreddit post.
By Phonlamai Photo at Shutterstock.com
To effectively share trading intelligence, a collective must first be able to produce the intelligence in a standardized format, so that everyone in the collective may be able to interpret it and put it to good use immediately.
Data processing, models, the description of strategies, and their execution must all follow a standard.
This leads us to the next question…
How may a collective produce intelligence in standardized formats? — By having its members use the same tools.
To be effective, the collective requires the same kind of sophisticated technology infrastructure that hedge funds produce for their own use.
In other words, for a collective to level the playing field it must empower its members with the same kinds of tools that big players build to compete with other big players.
If you wish to compete with the pros, you must think and work like a pro.
What’s the right model for kick-starting such a project? — Nothing new… it’s been around for decades. It’s tried and tested, and has taken over the Internet: Open Source Software.
The core competency of a Collective Trading Intelligence is the capacity to mobilize the crowd to develop its own data science and trading automation infrastructure, designed and built from the ground up with collaboration in mind.
You may think that it’s hard to bring people together to start such an ambitious project. You’re right. It is. But it’s not impossible.
All it takes is a small team of visionary entrepreneurs and clever design of incentives.
Incentives, Incentives, Incentives
Let’s say you have a team of visionary entrepreneurs.
Why in hell would they start an open-source project to empower the next WallStreetBets generation? — They’ll need a business model, right?
By Sergey Nivens at Shutterstock.com
If you don’t think this bit through carefully, you’ll likely end up with the next Robinhood: a sweet pro-little guy marketing speech, but unable to deliver and unwilling to take a stand for users when the going gets tough.
A centralized company with bills to pay, investors to answer to, and incentives different from those of the crowd is a recipe for disaster. Such an organization will never be able to stand up to the pressures that will rain down once the big players start feeling threatened by the collective.
What you need is an all-encompassing business model. One that not only does not interfere with the goals of the collective but guides them. In fact, the main design requirement of the business model is that it must align the interests of all the parties involved: the project, users, and the collective.
If there’s a sector in tech that has thoroughly explored the space of incentives, that’s crypto. We must look into the experience in the sector if we are to find a solution.
One of the many brilliant aspects of Satoshi’s creation is the design of incentives of the Bitcoin network. Satoshi figured out a mechanism that would incentivize every actor, every stage of the way, from inception to mass adoption. He built the software, released it, and boom! It spread like wildfire.
How much money did Satoshi raise from angels and VCs to pull this off? — Zero dollars.
No A and B Series rounds? — No.
No IPO? — Nope.
That is how brilliant the design of incentives of Bitcoin was from the beginning. The open-source project was bootstrapped from zero to the current trillion dollars market cap by adopters of the network.
If you are thinking that a token-based economy may be the solution to the conundrum, you are certainly right. Let’s do some basic exploration of possibilities.
With the prospects of a token economy, the team of visionary entrepreneurs are incentivized to start coding. They know that if they build a vibrant economy around the project, the token that will lubricate the pipes of the machine will eventually be worth something. A quick look into Coin Market Cap reveals that a Top 50 cryptocurrency is worth at least $1.5 billion. The founders will keep some tokens in payment for their work. So far, a typical founder’s incentive scheme for a crypto project.
So, they start coding.
What exactly? — That’s a good question.
Shall they start building the ultimate collaborative data science and trading automation technology infrastructure? — Yes and no.
Yes, in the sense that it is the long-term vision, and whatever is built must be built with that in mind.
No, because spending ten years developing a massive infrastructure is not a reasonable plan. Like with any tech startup, it’s important to get to a usable product as fast as possible. Then, iterate and evolve.
An early-stage product attracts users, helps to achieve product-market fit, and — most importantly — attracts contributors.
Contributions is a keyword as it’s how open-source projects grow and accelerate development. The philosophy is quite straightforward: a technical user finds the product convenient, thus is happy to contribute to making it better, for his or her own benefit, and also in appreciation for the work done prior to his or her arrival. That is the common thread in open-source communities.
Now, what happens if contributions are incentivized by the same token the founders get? Boom! Contributions explode and development accelerates to neck-breaking speeds!
Because the project is building software to process data and automate trading, users will be leveraging the product to create their own trading intelligence. Remember, intelligence is in processed data, models, strategies, and automation. And of course, the platform helps users develop these assets in a standardized format. As a consequence, assets are portable, meaning that users may start sharing data, models and strategies at any point.
Some people are naturally inclined to cooperate and will start collaborating with other users, spontaneously sharing whatever they build. This shared intelligence may be included with the software so that new users have something to start with, making the platform more attractive.
If this happens spontaneously, how much shared intelligence do you think the project may amass if it was incentivized with the token too?
At this point, you have a user base that is developing trading intelligence for their own use. At the same time, a subset of the user base is slowly getting invested into the project’s vision of a token economy around the software they are helping build.
Standardizing the format of intelligence assets makes assets portable. Still, a solution is required for the seamless transferring and deployment of intelligence.
How do users access the intelligence of other users? — Directly! The software should be able to establish peer-to-peer connections with other instances of the software. A decentralized, permissionless, and censorship-resistant peer-to-peer network is born.
By Yurchanka Siarhei at Shutterstock.com
Users may now choose to expose intelligence to the network, while other users are free to consume and build on top of it.
Picture the following scenario. Alice processes data and makes the data set available at her node. Bob uses Alice’s data set to train his Machine Learning algorithm and exposes the resulting model at his node. Carol takes Bob’s model, compares the output with a model of her own, and produces a signal when both models agree. She exposes the signal to the network, and Dave uses the signal as a trigger for one of his trading strategies. Once the strategy is back and forward tested, he exposes the trading algorithm to the network too so that anyone may run it.
This is how a collective may collaborate to produce trading intelligence that may — given sufficient time and evolution — compete with big market players.
Notice how the collaboration does not depend on users acting as a herd behind the most popular celebrity in the group. It doesn’t depend on users trading in unison. On the contrary, it offers an environment in which valuable contributions rise in popularity in proportion to the number of users building on top of those contributions. Intelligence exposed to the network become building blocks for more sophisticated intelligence.
Users may choose to work in silos, cooperate with partners within closed groups, or participate in multiple open global collaborations that may coexist in the network, each with their focus, interests, and goals.
All that intelligence must be worth something, right? — Indeed!
Intelligence may be exposed to be consumed free of charge, or for a fee through subscriptions.
Take a minute to think about the ramifications of the above. Let’s explore a few of them…
The possibility of monetizing intelligence adds a new layer of incentives to participate in the network. It allows users to take part in a distributed business involving the commercialization of intelligence products and services.
As intelligence evolves within the network, it’s a matter of time until turn-key trading services emerge. These services would become available to unskilled users — investors — who can’t produce their own intelligence, but who would be willing to pay a fee to hire trading bots to trade on their behalf.
This leads to a massive retail adoption cycle that boosts the economy and establishes a positive feedback loop: as demand for intelligence increases, the incentives to produce more and better intelligence grows; the latter results in superior intelligence that, in turn, attracts more investors.
The thread that ties everything together is the token, which is required as the medium of payment in the network. Users that wish to consume intelligence in the network must acquire the token to pay for their subscriptions.
Remember who owns the tokens? — Everyone that contributed to developing, deploying, and bootstrapping the network.
This is how everyone’s incentives are aligned. This is how the founders, contributors, and the collective share a common long-term goal. The business of the collective is to produce value in the form of intelligence in the network so that the token they earn for their contributions becomes valuable too. Remember that a Top 50 cryptocurrency is worth at least $1.5 billion, so there’s plenty of value to incentivize early adopters to participate in the collective enterprise.
Notice how the business of the collective exists in parallel to the businesses of each user, without any conflicts of interests. Users adopt the software to increase their own performance, that is, to pursue their own goals. They learn about the collective endeavor and figure it’s an obvious opportunity for a side gig, as it’s in perfect alignment with their personal activity anyway. So they contribute and become invested. In the long run, you end up with a huge crowd bootstrapping the economy that will benefit everyone.
The retail investors’ adoption wave leads to a Cambrian explosion of innovation, with network participants creating ever more sophisticated products and services.
It’s a matter of time until increasingly larger firms start paying attention to the opportunity and begin moving their operations to the network. These dynamics lead to the wave of institutional adoption. Pretty much what we are currently seeing with Bitcoin.
The end game is an open, decentralized, permissionless, and censorship-resistant financial services marketplace. Under such a paradigm, anyone may participate on either side of the counter on a level playing field enabled by the freely-available, open-source technology infrastructure that supports the network.
The marketplace will disrupt most of the rent-seeking mechanisms plaguing the current financial system.
By democratizing access to market information and trading automation, entities that were using their edge for nefarious purposes will lose control of those levers, as they will be kept in check by an equally capable positive force at an algorithmic level.
There won’t be a need for next-gen WallStreetBets people to band together and risk their capital in YOLO trades to keep bad actors in check. Instead, the multiple collectives in the network will build the algorithmic intelligence that will keep an eye on the sort of behavior they wish to eradicate.
You may be wondering how I came up with this elaborate plot two weeks after the WallStreetBets event.
Well… I didn’t.
What I just described is the work of Superalgos, an open-source project a group of friends and I started in 2017. The plan is in motion, and half of what you’ve read is already a reality.
The Superalgos Client is a powerful trading intelligence system integrating all crucial aspects of data processing and visualization, machine learning, visual strategy design, testing and debugging, trading execution, and trading bots deployments. The software has been in perpetual open beta for a year and is currently in its 9th beta version, with many hundred weekly downloads.
Superalgos built-in charting system.
The community contributing to developing the software and the intelligence the software ships with is growing and the dynamics described above in response to incentives is working like a charm. New users discover the collective business as they learn to use the system and are happy to jump on board.
The Superalgos Network will be launched in the mid-term, once a large-enough user base is producing sufficient intelligence to start the network’s bootstrapping phase.
The current focus is on cryptocurrency markets, as they are relatively more open and less gated that traditional markets.
We’re still at an early stage. There’s no better time to join the collective and be a part of a movement that will forever change how humans and machines engage with the markets.
Create your free account to unlock your custom reading experience.