Founder, Student One Causal Networks Private Limited
Artificial Financial Intelligence
My name is Shubham Sood and my forté is First Principles.
I started Student One to do one thing: run systematic searches on financial data and keep only what is numerically real.
My background is first-principles thinking. I treat market data as signals in time, not "content." Work by people like Ehlers and others in digital signal processing showed that if you transform price and volume into waves, you can reason about phase, frequency, and regime instead of staring at candles. That is the mental model behind this company.
I was also influenced by platforms like Numerai: you do not get raw names, only anonymized, rescaled data; you are judged only on the quality of the structure you extract. Student One follows the same philosophy. We do not sell data access. We sell the result of large searches over those waves.
In a two-trillion-dollar market, if you haven't checked four trillion combinations, then how do you even stay informed?
This is not traditional finance. This is wave mechanics applied to time series.
We do not study "price action." We study waves against price.
By quantifying price action in discrete time intervals and transforming them into wave representations, we can analyze financial data the same way physicists analyze signals: phase, frequency, amplitude, interference patterns, harmonic structure.
When you look at a candlestick chart, you see price levels. When we look at the same data, we see:
Position and momentum of price movement, not just "up" or "down"
Cyclical components extracted via Fourier-like transforms, revealing hidden periodicities
Signal processing to identify when the underlying wave structure changes—not by fitting models, but by measuring the wave properties directly
Multiple timeframes overlaying like quantum superposition—constructive and destructive interference of market rhythms
This is why we do not call this "technical analysis." This is signal analysis. The mathematics is closer to electrical engineering and quantum mechanics than to traditional finance.
Waves, Not Candles.
Price is the observable. The wave is the state function. We measure the wave, not the candle.
At Student One, the core work is:
Defining data theories for markets (what we consider a valid signal, a valid stencil, a valid regime)
Running heuristics and meta-heuristics to explore very large search spaces
Using combinatorics and search theory to decide which parts of the space deserve more compute
Using symbolic regression to discover deterministic relationships, not just approximate them
Symbolic regression is central here. It is one of the highest grades of machine learning in the sense that it searches over functional forms, not just parameters.
I have no formal finance background. No computer science degree. No quant PhD.
I studied hospitality management. I spent years in hotel operations—revenue forecasting, demand modeling, capacity optimization. Systems thinking applied to real-world constraints.
When I decided to enter financial research, I did not go back to university. I built the knowledge base myself: signal processing from academic papers, algorithmic search from first principles, market microstructure from practitioner literature, symbolic regression from open research.
I approached financial research from first principles: what constitutes a signal? What defines a valid search? What qualifies as evidence? This meant working directly from foundational questions rather than inherited frameworks.
Hospitality teaches you that systems must work under pressure, with constraints, in production. No infinite data. No perfect conditions. You optimize what you can measure and you ship when it is ready—not when it is perfect.
Self-taught means you do not wait for permission. You read the paper, you implement the algorithm, you test the hypothesis, you measure the result. I shipped this solo end-to-end—करने से होता है.
Coming from hospitality—an industry with brutal regulation (health codes, labor law, licensing)—I treat compliance as a design constraint, not an afterthought. Immaculate Compliance is not marketing. It is architecture.
I did not start by hiring a machine-learning engineer. I went the opposite way on purpose.
The difficult parts here are not a TensorFlow pipeline. The difficult parts are:
I learned and implemented those concepts myself before adding anyone else. Not because I couldn't hire, but because I wanted the core mathematics and the core constraints to be internalized, not outsourced.
Even the policies aren't generic. I wrote all 14+ policy documents myself—custom-built after reading hundreds of compliance papers from SEBI, SEC, FINRA, MAS, FCA, ESMA, and BaFin. Every clause, every threshold, every exclusion reflects actual regulatory reasoning, not boilerplate templates. I'm a bit of a lawyer too—except, I'm not. It's just de principiis and juris ratione.
People who want to join this now—in any capacity—have reached out. We receive interest from interns up to CTO/CFO level, but our small team is adequate for now.
Student One operates in a deep IP space. We do not maintain a public "team" grid on the website.
If you want to understand this company, do it through our methods, our policies, and our outputs. The math is the point.
There is no way I would ever operate in a grey area. We apply science and mathematics to a highly regulated domain—a pristine white area. In a world of black boxes, I made a pristine white box that could, in fact, teach something to a black box. Mathematical and machine learning fact.
If you are a registered asset manager, a family office with compliance infrastructure, a hedge fund with verified regulatory status, or an advisory firm with 10+ analysts, allow me to offer you a few reports as goodwill—no strings attached.
Reach out with your company email and corporate identity. We will show you what we deliver for a few weeks as a demonstration of our mathematical finesse on silicon.
The way I see Student One is very simple: we operate like a core facility in a proteomics lab. You bring the samples; we run the machines.
In biology, a research group might send protein samples to an external core facility to run mass spectrometry and get back hard numbers on protein abundance, post-translational modifications, or structural conformations. The core facility does not promise a cure, does not interpret clinical relevance, and does not speculate on downstream drug targets. It only reports what is actually there under a defined protocol.
Student One works in exactly that capacity for financial data. Your historical tapes are the samples, our engine is the instrument. Under the constraints you define, we run a systematic battery of tests and report every statistical phenomenon that genuinely meets those thresholds.
I don't declare, imply, or speculate on whether any of it will be profitable in a live market. You are paying for the computation and the measurements, not for a prediction.
Markets carry risk not only in price but in structure—in how often rare excursions occur, in how fast they unfold, and in how consistently certain geometric distortions repeat. Our mathematics does not predict these events; it simply enumerates the geometry that has already existed. These analyses enumerate historical risk signatures embedded in an asset's past behavior. They do not forecast, advise, or recommend. They simply reveal patterns of past excursion, recurrence, and directional stability observed under repeated structural conditions.
This is a mathematical computation service. The value is in the exhaustive search, the state-of-the-art machine learning models and techniques we apply, the mathematical rigor, and the compliance framework that ensures every reported historical phenomena has met a strict threshold.