Author: Alejandro Reynoso
Chief Scientist, DEFI Capital Research
External Lecturer and Research Fellow, Judge Business School, University of Cambridge
There are books that explain finance, and there are books that explain intelligence.
This series does both — and then dares to show they are the same story told in different languages.
AI for Financial Practitioners is not a textbook in the traditional sense.
It is a living, thinking ecosystem — volumes where mathematics meets imagination,
where economic systems behave like neural networks, and where the reader does not just learn but experiments.
Every idea in these volumes can be executed. Every model can be observed in motion.
Each chapter includes one or more interactive Google Colab notebooks, turning concepts into simulations
and reading into discovery. In the digital edition, hovering over a notebook title launches it directly —
revealing liquidity flows, portfolio entanglement, and reasoning agents learning in real time.
Tag: course-volume1-v01
Download PDF (Release)
The opening volume descends into the machinery of intelligence.
It begins with the neuron — the smallest atom of reasoning — and builds upward to networks that forecast, generate, and adapt.
From dense layers to CNNs, LSTMs, GRUs, transformers, and diffusion models, the reader witnesses the gradual assembly of a financial brain.
Each notebook is a miniature laboratory: small enough to understand, powerful enough to astonish.
By the end, the question shifts from “What is AI?” to “What new forms of reasoning can finance sustain?”
Tag: course-volume2-v01
Download PDF (Release)
If Volume I teaches you to think, Volume II teaches your models to talk.
Here, intelligence becomes social — autonomous agents negotiate, cooperate, and compete across digital markets.
Readers learn to construct agentic collectives capable of executing trades, evaluating risk, and reasoning about regulation.
Multi-agent reinforcement learning, A2A orchestration, and tool selection become tangible through a symphony of interconnected notebooks.
Finance is reimagined as an ecosystem of interacting minds — every simulation feels like an economy learning to think.
Tag: course-volume3-v01
Download PDF (Release)
The frontier where economics meets physics and biology.
Markets are no longer numbers but living geometries. Agents become molecules; portfolios become wavefunctions; liquidity becomes energy.
Chapters explore quantum encoders, Hamiltonian portfolios, evolutionary reasoning molecules, and immunological defense systems.
Through Colab notebooks, readers visualize market phase transitions, quantum interference, and neural ecologies —
a union of theory and experimentation that asks what happens when finance itself becomes self-aware.
Tag: course-volume4-v01
Download PDF (Release)
Why this volume matters. Volume IV codifies a discipline: building small, auditable, governance-native learning systems that you can understand end-to-end.
It is written for researchers, students, and professionals who need systems that are not only strong but transparent, reproducible, and controllable.
Two-part structure.
- Part I — Conceptual & Mathematical Foundations. A precise vocabulary for training under scarcity: gradient field interpretation, structural convergence, representational folding, loss curvature, counterfactual supervision, and multi-agent learning dynamics. Each chapter doubles as a diagnostic manual — not just how models learn, but how they fail and how structure restores them.
- Part II — Ten Executable Notebooks (Colab). Building AI from Scratch: A Practical Guide for Financial Professionals. Pure NumPy/Matplotlib implementations (no heavy frameworks) that make every update and parameter observable. Each notebook follows a consistent arc: Motivation → Intuition for a few core equations → Minimal implementation → Visual diagnostics → Governance notes.
Notebook lineup (1–10).
Minimal Objectives & Gradient Fields · Representational Folding · Ambiguity & Error Correction · Supervision via Multi-Agent Architectures ·
Temporal Windows & Distributed Memory · Contrastive Reasoning & Intervention Design · Distributed Fine-Tuning & Emergence ·
Multi-Agent Supervision & Structural Learning · Rapid Alignment with Minimal Tuning · Generalization Failures & Structural Debugging.
Design philosophy. Minimalism here is not reduced capability; it is maximal interpretability.
In an era of billion-parameter models, mastering the smallest viable system is an act of epistemic clarity and a fiduciary responsibility in finance.
Tag: course-volume5-v01
Download PDF (Release)
Governance-native learning in practice. Volume V operationalizes rubric-based preference optimization (GRPO) for institutional finance:
structured rubrics, KL-regularized updates, escalation policies to larger models, and auditable logs that meet fiduciary standards.
What you build.
- A compact reasoning core trained via RPO with explicit rubrics and failure taxonomies.
- An orchestration layer that supervises multi-agent evaluation, safety checks, and recap prompts.
- A governance trail: experiment manifests, loss/provenance logs, reproducibility checks, and release artifacts.
Executable suite. Ten Colab notebooks mirror the production workflow: rubric design → data curation → GRPO training → evaluation harness →
safety & alignment probes → error-analysis dashboards → rapid retuning → deployment manifests → master index & documentation.
Across the series, more than seventy Colab notebooks form the backbone of a new learning paradigm.
They do not merely support the text — they are the text in motion.
Each notebook is a conversation between theory and code, between abstraction and observation.
Students adjust parameters, rewire architectures, and test hypotheses — turning lessons into living experiments.
This is not passive reading. This is active reasoning.
Spanning five volumes and dozens of chapters, the series covers the full terrain of modern AI in finance —
from neural learning to agentic negotiation, from structural reasoning to quantum cognition, and the governance-native craft of minimalist, auditable training.
It speaks equally to graduate students, researchers, and financial professionals operating in real, regulated environments.
The volumes form a continuum:
- Volume I: The Toolbox — how intelligence is built.
- Volume II: The Architecture — how intelligence organizes itself.
- Volume III: The Laboratory — how intelligence evolves and reflects.
- Volume IV: The Method — how intelligence is trained minimally, audited rigorously, and deployed responsibly.
- Volume V: The Practice — how reasoning systems are aligned with gRPO, supervised by rubrics, and shipped with governance artifacts.
Together, they redefine the textbook: not just to read about intelligence, but to create it;
to design reasoning systems, observe emergence, and govern what we build.
-
Reynoso, A. (2025). AI for Financial Practitioners: Volume I — Foundations of Financial Machine Learning.
DEFI Capital Research · Cambridge Judge Business School.
Release: https://github.com/alexdibol/courses_education/releases/tag/course-volume1-v01 -
Reynoso, A. (2025). AI for Financial Practitioners: Volume II — Agentic Systems in Finance.
Release: https://github.com/alexdibol/courses_education/releases/tag/course-volume2-v01 -
Reynoso, A. (2025). AI for Financial Practitioners: Volume III — Structural Intelligence & Quantum Reasoning.
Release: https://github.com/alexdibol/courses_education/releases/tag/course-volume3-v01 -
Reynoso, A. (2025). AI for Financial Practitioners: Volume IV — Minimalist Training of Financial Models.
PDF: https://github.com/alexdibol/courses_education/releases/download/course-volume4-v01/VOLUME_4.AI.REVISED.pdf -
Reynoso, A. (2025). AI for Financial Practitioners: Volume V — GRPO-First Reasoning Systems for Finance.
PDF: https://github.com/alexdibol/courses_education/releases/download/course-volume5-v01/VOLUME_5_REVISED_VERSION.pdf -
Reynoso, A. (2025). Ten Experiments in AI Finance: Minimalist Training in Practice.
(Companion Colab suite; governance-native, dependency-light implementations.)
- Read the PDF for the target volume.
- Open the matching Colab notebooks (linked in-text) and run locally in the browser.
- Tweak parameters and instrumentation (loss logging, curvature, drift, saliency).
- Use the governance prompts to document experiments, rationale, and outcomes.
Reynoso, A. (2025). AI for Financial Practitioners: A Pentalogy on the Future of Reasoning, Markets, and Intelligence.
DEFI Capital Research · Cambridge Judge Business School.
Available at https://github.com/alexdibol/courses_education