This is a story of finding a universality, transforming it into an indexing innovation, persisting with it, bringing it to Canada, and revolutionizing DIY Investing.
$ 100 Million and 3 Years
The name AlphaBlock came from a brainstorming exercise about how to validate and protect Alpha (returns above the market) using blockchain. Using a smart contract, we built a performance attribution system that would build trust, allow for interaction and validate algorithmic alpha claims. The reason we wanted to do this was that as an “Alpha Innovator” (ai) industry expects you to show real money performance for 3 years and have a threshold of $ 100 million in assets before you can approach the large asset owners like CalPERS. There is simply no system for a small ai to get in front of an asset owner, while there are published studies pointing out the high likelihood of selected outperforming managers underperforming and vice versa because of natural reversion. [1]
Mumbai to Transylvania to Toronto
Mumbai, the financial capital of India taught me finance. With more than 100 years of financial history, you can only imagine the sense of generational knowledge a child gets from a parent. I was lucky to get a firsthand sense of it and also able to find myself in Mumbai. I had great teachers, who did not just teach me, but also showed me how the character was important if you had to understand markets. A flight to Romania in 2004 exposed me to a frontier market with similarities to 1990’s India. I decided to stay.
You need to be crazy, living on a ‘Transylvanian’ hill, cutting your wood, with the 2008 crisis behind you to believe you had enough quantitative background to build a meaningful global financial innovation. What started as a research paper challenging Mandelbrot and showing that fractals could exist in time was a great signal to begin trading pairs [2] but how do you go about building an asset management business around it, a hedge fund.
80% of global market technicians in 2010 were on the East Coast, and since more than a fifth of all global active trading happened from Canada, Toronto seemed like a great hub. Lake Ontario kept calling.
Smart Beta, Basis Point Wars, and Stress
It was while building systematic actively indexed strategies for asset managers in Canada, Hungary, and Romania that the idea of building a rule-based process to build a version of Smart Beta happened in March 2013. Could I scale the process to Dow 30 or S&P 100? The business school had made it clear, that S&P500 was the market, the holy grail. Everything in capital markets moved around it. It goes up, it goes down and leads the economy as a bellwether. So, recreating the S&P 500, making a better version of it was seen as a fool’s game. Beating the market, as a technician, was a fathomable idea, since you forecast day in and day out, and capture trends from every intermediate multi-week move. But converting that signal process into a new S&P500 was an overwhelming feeling.
Fortunately for me, Robert Arnott had already started the revolution in 2002 and I was just a decade behind him. From the inspiring beginnings of Smart Beta when Fundamental Indexing (FI) was launched and Wikipedia pages mentioned that FI outperformed the S&P500 on a risk-weighted basis by 250 basis points [3], it was clear to me that this was our chance to get into the big leagues as our process (methodology) was more scientific, unambiguous and delivered more than 400 basis points risk-weighted excess return.
I felt I was in a war of basis points. It’s not hard to express how an obsession with a process can consume you. It took me time to realize that doing what I was doing was not normal, and I had to give in to the process, and park the idea of a “normal life”. I had a purpose to convert a nascent process into a robust model, which could eventually find itself on a global stage. One of the many side effects was stress, consistent overthinking created a traveling lingering pain, a few months in the neck, a few months in the shoulder, and so on. Running helped to keep it at bay.
$ 1 billion Assets Under License
Back then in 2013, I believed that we could take away 1% of Bob Arnott’s business. He led Research Affiliates (RAFI) to the fastest growing $ 100 billion in assets under license (AuL) in 10 years. My belief and percentages have grown exponentially since then. Fundamental Index aura diffused in time as ‘Value’ factor underperformed and though brilliant in terms of the idea that the market capitalization (MCAP) is an inefficient way to build passive baskets, the MCAP’s intrinsic growth bias secularly outperformed the ‘Value’ factor investing.
As I read and researched later and wrote about Boulding [4], Brown [5], Bachelier [6] that I realized that the history of finance was checkered and refused to look beyond informational content. My vision strengthened that invariably the industry will have to look beyond its current scope and get more scientific. The 400-basis points risk-weighted excess return against S&P500 gave me comfort that I had something the world wanted. While I was maturing as a quant, the industry was maturing in terms of expectations and was reluctant to pay basis points for licensing old technologies that did not produce alpha.
UChicago to Nasdaq
During a University of Chicago event, a co-speaker offered to introduce me to Nasdaq, saying that they would be happy to work with you. In 2014, Nasdaq first calculated the models from 2004-2014 and then listed a basket, which we called RMIVG 20, the name was an acronym for Risk Management Index Value Growth 20.
I had a small confusion pre-launch, should we launch a small active 20 basket model, or should we launch the S&P 500 version. If I would have done the latter, I would not surely be writing this now. It’s amazing how life is when you are learning as you go. You have no clue where you are going. You are driven by a sense of exuberance and survival as it guides you to make decisions. Standing on the 50th floor at the exchange, looking at ground zero, I felt I had arrived. A listing would change everything, and assets will start flowing, but that did not happen. After about 7 years of listing, the experiment closed in April 2021. We had beaten the NQUSB (The Nasdaq Benchmark Composite Index) by an annualized 590 basis points on a risk-weighted basis.
RMIVG20 – Nasdaq Orpheus RMI Value Growth 20
From 2013 till 2019 we kept stress testing and validating the method under various global institutions. We listed 19 models on Bloomberg platforms while completing the arduous journey of building a delivery platform for tech which built automated smart beta portfolios and updated them end of the day.
The belief was that once we build the tech then it would build trust, assist in due - diligence, and asset managers and asset owners will flock to the idea and we would hit our $ 1 billion in AuL. I was still working on my logic that alpha should be a magnet for capital. “If you build it, they will come”.
The fact that a machine could build a hands-free process is all that should be needed to understand the integrity of an algorithmic process, to understand that there was no backtesting but backdating. Yes, all of the above-opened doors got us further into more grilling due diligence and customization and finally, assets started trickling in.
Tralio to MIT Fintech Award
Two years after the Nasdaq listing, while we were building Tralio (the automated smart beta builder and analytics platform), I ended up enrolling for the MIT Fintech Program and saw myself in the top list among around 800 participants globally. Awards can open doors, but they cannot build AuL.
Quoting my wise friend Rona Schehrer from Zurich, a seasoned asset manager, and a mentor to me, who always said, “In our business Mukul there are only three important things, Capital, Capital, Capital”. The idea that only capital can validate your alpha and not become a magnet for it was twisted logic to me. How can you influence $ 100 billion to follow you? The fact that it was going to happen $ 1 million at a time, quarter by quarter, made me realize that either I had to live longer to see those billions, or I had to conjure up another plan. Awards don’t create innovation-driven businesses, persistence, being at the right place at the right time, marketing, and access to the network do a part of the magic. Innovations need a certain complexity to thrive. Fortunately, society today has managed to create complexity, this is why innovation has got a chance to incubate, persist and eventually flourish.
New Entrant’s Curse
Assets are at the heart of asset management. It’s like a pie, you don’t share it with the new entrants, you collaborate with new entrants to extend the pie and hence the business. Since attracting capital is at the heart of asset management, many great alpha-generating ideas have floundered because they couldn’t attract enough capital. I call it Porter’s fort [7]. The new entrants can’t threaten the financial industry because assets are the passport. And you won’t waste your time building an idea hoping that will get assets. Hence, new entrants, have the monumental task of becoming incumbents. Even substitutes create incremental benefits. ETFs are substitutes for Mutual funds (MF), they have nibbled at the MFs for more than 30 years and have overtaken them but have not created any alpha innovations. How does the industry rivalry change, is a Harvard Business School case study? So here I was, a new entrant with an innovation, seeking assets, aspiring to become the incumbent.
Harvard Business Review: How Competitive Forces Shape Strategy, 1979
Porter’s Five Forces Framework
And because you don’t have asset power, you are a rookie. There is no bargaining power. You get happy with what you are offered. Some basis points. No basis points. Joy from an institutional validation. It is like a new actor trying to make his way into Hollywood, you have a chance only if you have the energy to persist. There is limited governance when it comes to ai. We are at the mercy of the new technologies to give us a level playing field. The code of financial industry conduct forces an ethical play, but that is still for the developed markets, where punishments are serious, but there is simply no classification or rules for nurturing ai.
I am the few extremely fortunate lucky new entrants to get backed by vibrant early-stage venture capital and by industry veterans, supported and appreciated. I had the luxury to complete the tech and get to a product.
But many times, I wonder about the lost generation of youth who spend countless hours coming up with something new to build a hedge fund. I was also lucky because of my love for research. Life in the stock market is highly taxing, leave aside eyes, sedentary life, family life, etc. conflicted by a belief that if you can become a golden egg-laying hen, you would have faster success. You can’t blame the asset managers too, it’s a fiercely competitive industry struggling to grow their assets while striving to differentiate. And with 9/10 asset managers underperforming [8] the market, there is simply not enough bandwidth left to mentor innovation.
Above that, there is compliance, which at some level is risk focussed and not alpha focussed. Alpha is like a perpetual motion machine, believed only in theory, not in reality. So, if there is a 100 basis in fees to share, there is the limited scope and nurturing for alpha-generating technologies to sustain and succeed. This is the curse of the new entrants which is hobbling the incumbents, they need the new entrants, but they are tied up with their problems, a tragedy of commons. There is another problem. Active managers get paid because their process is active. How can you get paid, if you employ a smarter passive process? Somewhere the industry has siloed itself into asset maintenance rather than understanding that the core fiduciary duty should be to create conscious performance innovation.
The entrant’s curse, in the end, is to scale or die.
No Purpose Without Persistence
When you are all alone, looking at the decade that passed, wondering about the tide that washed away all your sandcastles, you wonder why you are doing what you are doing? What is the purpose of all this? How can you breakthrough? A lot of times there is such a disconnect between the founder and the backers that it seems you are sitting at the bottom of the totem pole. Pulling yourself out of the abyss needs energy from another galaxy. Founders are such underappreciated souls, till they almost die of Malaria or bad marriage or being fired from their own start-up [9]. You need to rely on everything you have. Your physical energy, your mind, your spirit.
I am reminded of another story, when I was teaching at a business school in Mysore, a South Indian city close to the silicon hub of India, Bangalore. I made a friend at the local gym, who asked me if I had climbed up the stairs to the Chamundi temple on top of the Mysore hill. I said ‘No’ and we decided on a Saturday morning to climb up the stairs. I had no idea of what I was getting into. The idea of going on a hill trip was exciting. I started climbing the stairs, following my friend. The steps were carved out of rock and on occasions were large enough for two big steps. After a few hundred steps, I asked him, “how much more to go?”. “A little bit more”, he said. Again, after a hundred or so steps. “How much more?”, “Just a little bit more”. The fact that he kept going, brought out the competitive spirit in me. I could not show weakness, I should bring out my strength and keep going. I kept trudging along and reached the top of the hill, only to find out that there were 1000 steps and it was not easy to climb them without a stop on the first attempt. All I needed was a friend, and I could keep going. I had sunk a decade, a few million dollars into the project and I was not scaling but I had my friends. I had to keep going.
Lucky Montreal
Talking about road and hill trips. Montreal has been fortuitous for me. It’s like going to the temple. You go to Montreal and good things happen. I took a trip to Montreal in 2011 and the last call I made during the roadshow turned out to be the first Canadian asset management client for us. Another trip in 2018, got us a VC backing. And it was on a trip to Montreal in 2019, while brainstorming with my CTO regarding how to go to market, she said, she could work around the tech to give anyone the opportunity to build a unique portfolio. Why not let everyone build his/her portfolio? It was then we realized that we could take the technology to everyone and anyone. The algorithms did not have to stay captive in the tech but could start interacting with everyone based on his/her preference. This breakthrough moment opened up a plan to scale.
Taking Impact to the Investor
There was still something missing. How could we allow the users to play with our algorithms? How could we get them closer to the process so that they could trust it more? If we could bring the investor closer to a machine-based alpha, he could onboard and use the service without being told to do so. The capital could seek alpha. We had to build a sandbox where investors could play with the process and simulate it at their end, a codebase that could allow for total interaction. We had to operate in an open way with the investor. The impact could not happen if there was no trust. We also had the understand the academic challenge. Where was theory weak and how it could be systematically challenged.
The Nobel Prizes and Granger’s Challenge
Disruption is dramatic. The financial industry’s future relevancy [10] made our work a bit easier. SPIVA had made it clear that delivering alpha is difficult [11]. Behavioral finance was value investing [12]. Science was facing a replication crisis [13] and pension funds were struggling with the longevity risks. Our persistence had got us to be at the right time and right place. But then as Granger [14] said, “To build a method that consistently produces positive profits after allowing for risk correction and transaction costs and if this method has been publicly announced for some time, then this would possibly be evidence against EMH...Only if a profitable rule is found to be widely known and remains profitable for an extended period can the efficient market hypothesis be rejected...Benefits can arise from taking a longer horizon, from using disaggregated data, from carefully removing outliers or exceptional events, and especially from considering non-linear models.”
In simpler words, he said, that you could challenge a Nobel prize-winning and mainstream financial theory called the Efficient Market Hypothesis by building an alpha-generating innovation and leaving it in the public domain and see if it diffuses in time. If in this extended period of time the new theory sustains, then you have refuted EMH and have become the prevalent theory.
Real alpha according to Granger is not arbitrageable. I agree. It’s out there, but you cannot arbitrage it. It’s like the equity premium puzzle, not arbitrageable. Somewhere things are not arbitrageable because of human inability to time. If you can’t time the spread, you cannot lock it. LTCM [15] became part of history books, proving that timing is modern Science. If chaos was predictable, it won’t be called chaos. Some things just cannot be timed. If the alpha was real and significant, it won’t just vanish when you give it away to the public.
After working through new factor coefficients and expressing an enhanced version of the Fama and French factor model at the UChicago in 2013, I had two choices, continue my academic journey and spend the next 30 years defending my hypothesis, getting academically ridiculed by linear regression or not to worry about anything but about how to create impact? How to create alpha for everyone? How to take on Granger’s challenge?
Opening up the method for everyone was taking on Granger’s challenge and proving that the $ 100 bill can lie on the road and there is a consistent way to beat the S&P 500 even if everyone knew about the method. And in case it does work, and everyone knows about it, we have built a new benchmark.
If everyone is going to follow a more efficient method that does not mean the new method will eventually underperform the old method. Everyone knows about the S&P 500, has it started to underperform because everyone knows about it? S&P 500 and MCAP methodology is an inefficient way to build a benchmark and there is not one but many ways to beat the S&P 500 and build superior indices.
Sandbox 1.0
The Sandbox was built to take on Granger's challenge. Can we build an open process to beat the S&P500 consistently? Even though a sandbox is not an investment advice, it is an academic research that is assumed to be not investable in its current form and even if there is a claim, a paper claim, it should open up a method for the industry to wanting to build smart intelligent portfolios.
Having said that giving away a decade of research is not easy. Giving away a process that is replicable is difficult. And giving away a process that claims to do what has never been done in the history of capital markets can only be called a sandbox. We are talking about an apples to apples comparison. Top 500 top MCAP stocks, year after year, going back in time. Going back in time, maintaining the integrity of the Universe is not an easy exercise. We have prepared the process for 20 years and plan to extend the study to 50 years. I don’t know of Smart Beta academic models on S&P 500 going back beyond 1965. The idea is not to mirror S&P 500 but to show that the same baskets can diverge significantly.
The roadmap ahead after the top U.S. 500 and top U.S. 100 is to launch Europe top 50, Canada top 60, top 10 cryptos and continue with line extensions for all G 30 economies, slowly moving to other assets. The objective is to articulate that just by changing the weights of the components in a basket, alpha can be generated. This flies in the face of conventional financial theory and behavioral finance but that is the claim.
Universality, Score, and Holding Periods
The method assumes that the stock market or any other natural behavior is non-linear, complex, has the duality of forces and a multiplicity of periodicities. Natural behavior could be defined by forces of reversion – diversion [16], which can be studied for various periodicities. Things go up and continue to go up (rich get richer), things go down and sometimes continue to go down (poor get poorer), things sometimes are going up but reverse (rich get poor), and sometimes what’s going down, reverses direction and starts going up (poor get richer).
More such probabilistic states can be defined but from a simple perspective, information lives in these states before it has color, character, content. It’s the interaction of these states that create investment opportunities and risks.
If I have to simplify this further, the sandbox is a scoring system for stocks, based on relative price performance. Modern finance is built on scoring. Like you can score something on fundamental ratios, you can score it on relative performance or any other variable. The scoring system scores relative performance on a percentile score for different periodicities, like a running quarter to many quarters (1Q, 2Q, 3Q, 4Q all the way to 20Q). The portfolios are built on quintiles for three holding periods of 1, 2, and 3 years.
600 Basis Points
This annualized excess return number against S&P500 breaks all previous records and might sound ridiculous and it does place this claim up there, an active return for a passive process. But then I had no real reason to painfully taking you through my story if I didn’t have a real surprise for you. The fact that 600 basis points can be delivered on paper in excess of the world’s largest Index (S&P500) with a few trillion dollars in assets under license shows that financial theory never adopted a simple approach, it refused to experiment and iterate. We had to do three things to break the impregnable S&P500 thinking. First, we had to think relative not absolute. Second, we had to think group not component. Third, we had to think weights, not selections. The industry spent eternity to select, while before that it had to say, let me try to weigh this thing differently. I come back to the famous quote, “Simplicity is the most undermined investment technique”. Time and again we wake up and realize, how simple it was, and how complicated we made it to be.
The portfolio styles are the top quintile (TQ), the bottom quintile (BQ), all quintiles (AQ), the top and bottom quintiles (TBQ) together, and the rest quintiles (RQ) which don’t consider the top and bottom quintiles. The visualization file built-in python takes the ~ 7000 simulations from the generated ranking and portfolio code for 20 years of historical data.
The average minimum and maximum range for annualized excess returns for various styles suggest that RQ is a drag, TQ is overall the best, BQ is volatile and a skewed towards both extremes, and AQ which contains all 500 stocks can deliver more than 600 basis points excess returns against the S&p 500.
The probability distributions illustrate the ineffectiveness of RQ as it skews negatively, the positive hump for AQ and TQ and the extremities in BQ
The styles form a flower bouquet spreading in all directions in both tails when looked at from annualized excess returns and Information ratio together.
BQ torch illustrates the risk of holding BQ portfolio as they have extreme negative tails on excess returns and a low Information ratio.
The RQ or blend as the industry may define it is the drag on performance. If we underweight the core and double the allocation for the top and bottom quintile, the alpha jumps dramatically to ~ 600 bps of risk-weighted excess returns. The BQ is like a burning torch, it delivers but at the expense of negative outliers. Since BQ would have a lot of conventional value overlap, the only way to overcome the Value torch burn is by the superior selection, machine or human.
The AlphaBlock method as we may call it from now on should be tried with fractional allocation. As there is no other method to efficiently execute the systematic allocations both at an institutional and individual scale. Active investing may be seen as different from Passive investing but from my perspective, Active, Smart Beta, Passive is a part of a complex of investing styles. They all are a part of the investment spectrum, and like a set of spades, hearts, diamonds, and clubs, the investing styles compete and complement each other.
The Business Model and Next Steps
Any machine learning process that works on informational content, thinking of information like an orange to be juiced, is working with a significant disadvantage. History of fundamental research has already proven that time and again that information drifts, flits between relevance and irrelevance and size, the top factor of markets could be a proxy (Banz, 1978) [17]. So, if you are venture capital backing companies claiming to revolutionize investing, you now have a 600-basis points sandbox reason to revisit your allocation strategy.
The AlphaBlock method relies on the architecture of informational states and the probability that a certain component is going to continue to stay in a certain state or move to another state, a dynamic Markov transition matrix, which is chaotic and ordered at the same time. Another way to look at it is to assume, that stocks jump like electrons, one state to another, and the only way to see them is probabilistically.
Since the process is robust and universal in its behavior, we refer to it as a generality, which is domain agnostic. This is the reason our process can easily migrate to any region and asset class. Our approach is like understanding chaotic systems before getting into any idiosyncratic precisions, which remains an inefficient way to manage risk. Over the coming weeks, we will add more codebase to our GitHub like 1) Starting point bias. 2) Drawdown simulations 3) Modern Portfolio Theory portfolio statistics 4) Summary statistics 4) Performance attribution 5) Fractional shares and impact cost 6) Fama and French vs. AlphaBlock factors etc.
I am excited that we reached here. I feel a huge weight off my shoulders releasing this sandbox, something I had promised myself long back. And now to pack this process into an app to reach the ‘Uber of Asset Management’ stage, where we can take alpha to every investor, without having any assets of our own, is mind-blowing.
We launch our service for our first 10 B2B and first 100 B2C users this quarter in 2022, get feedback and iterate and open up to the next 1000 users, and so on. How will we eventually make money? Since the AlphaBlock method is an out of the box approach to investing and has a low replication cost, we can build models across regions, assets, styles and customize them uniquely for every global investor, which means catering to a global audience of around 100 million global investors with risk capital starting 100 dollars to a few trillion dollars. Asset managers, asset owners, investors need an investable process and guidance on how to rebalance portfolios. There is a need for education, analytics, and machine-driven systematic processes. Our library of APIs will either execute these instructions or offer decision support for a fee.
Technology, Noise and DIY differential
The allure is something like the $100 bill lying on the road, which behavioural finance says can not be lying there. Building a service that does not charge a 3% management fee and makes it easy to do it themselves (do it yourself) and manage their investments and get an additional 6% in excess returns takes the differential to an annualized 9% (6%+3%) excess returns. This DIY differential is our DIY business thesis.
Alpha was never about 250 bps, 400 bps, 600 bps, or 1000 bps. Alpha was always larger than an imaginable number. It is shocking that a simple informational architecture is an unexplored gold mine. The good news, the gold mine proves that we need to define a model before we do Artificial Intelligence. Nature is not artificial, she is intelligent. System intelligence subsumes stock markets. Stock markets should always have been a method for us to understand Nature, but we refused to get out of a frame because we had to stop looking at information for its content, an unknown way, without precedence. No one wants to walk alone on a road never traveled. But this is what research journeys are about, to take the road never traveled.
Machines will become more capable than human stock picking and with every passing day, the gap between human top skill and machine will reduce, till time machines break the stock market game and commoditize intelligence at large. Then it will be a war of stock market games, one machine against another. The end result would be broader system intelligence, which hopefully will take the civilization on a more enlightened journey.
We have built this sandbox so that we can change the way alpha is perceived and harnessed. Revolution can’t happen if we don’t scream at the top of our voices. The sandbox is our scream. The industry needs alpha. We want to assist pension funds from their ongoing pension crisis, collaborate with asset managers who are looking at systematic quantitative processes, advise other asset owners who are going more passive, guide alternative data users to overlay our algorithmic process to get make more sense of their data. Support the investor on the street, who is technology savvy but needs guidance regarding risk. Technology needs a purpose. Purposeless technology adds to the noise.
AlphaBlock Sandbox - Generate Rankings and Portfolios
We are looking for a lead investor for our ongoing fundraise.
AlphaBlock Technologies Inc. Pitch Deck
Florina Pal, Patricia Ratiu, Radu Tiric Ciprian, and Dan Todor contributed to preparing this Sandbox
Bibliography
[1] Goyal, Amit, and Wahal, Sunil, The Selection and Termination of Investment Managers by Plan Sponsors, November 2004, SSRN
[2] Pal, Mukul, The Time Fractals, April 7, 2010, SSRN
[3] Fundamentally Based Indices, Wikipedia
[4] Kenneth E Boulding, The Economics of Knowledge and the Knowledge of Economics, 1966
[5] Ball and Brown, An empirical evaluation of accounting income numbers, 1968
[6] Bachelier, Louis, Théorie de la spéculation, Annales Scientifiques de l'École Normale Supérieure, 1900
[7] Michael, E. Porter, How Competitive Forces Shape Strategy, Harvard Business Review, 1975
[8] Pal, Mukul, 9/10 Fail Logic, AlphaBlock
[9] Ashlee, Vance, Elon Musk: Tesla, SpaceX, and the Quest for a Fantastic Future, 2017
[10] Paul Kovarsky, Fabozzi: Finance Must Modernize or Face Irrelevancy, CFA Institute
[11] SPIVA, S&P Dow Jones Indices
[12] Charles Wallace, Using Behavioral Finance to Better Understand the Psychology of Investors, Institutional Investor.
[13] Ioannidis, John P. A.. Why Most Published Research Findings Are False. PLOS Medicine., 2005
[14] Clive Granger, Forecasting stock market prices: Lessons for forecasters, International Journal of Forecasting, 1992
[15] Roger, Lowenstein, When Genius Failed: The Rise and Fall of Long-Term Capital Management, 2000
[16] Pal, Mukul, Mean Reversion Framework, SSRN
[17] Rolf, W. Banz, The relationship between return and market value of common stocks, Journal of Financial Economics, 1978