About

Original, empirical work on quantitative methods, AI, and financial market dynamics.

A personal initiative. Applied Markets publishes original, empirical work on quantitative methods, AI, and financial market dynamics. Every artifact begins with a question that has a tractable experimental form.

Each piece of work starts with a specific, scoped question — one that has a tractable experimental form. The output is a traceable record of method, data, and result. Partial findings are published. Failed experiments are published. The record is the product.

Artifacts are released as experiments, articles, and Python notebooks. Every release comes from direct implementation and first-hand measurement — not summaries of papers, not passive study notes.

Research Pillars

Pillar I

Quantitative Finance

The mathematical foundations that markets run on. Statistics, probability, information theory, and stochastic processes — studied through original derivations and empirical analysis. The goal is not to reproduce textbook results but to test where the standard assumptions hold and where they break.

probability information theory stochastic processes risk modeling market microstructure scoring rules
Pillar II

AI for Finance

Modern AI applied to financial reasoning, forecasting, and market simulation. LLMs, reinforcement learning, and agent-based models — tested on real data with honest results. The question driving each experiment is whether the AI method is genuinely useful or just appears to be.

LLM reasoning forecasting prediction markets RL for execution agent simulation causal inference

Why These Questions

Genuine unknowns

Still open

These are not settled questions with known answers. The experiments exist because the results are not already obvious.

Narrow enough to test

Scoped

Each question is specific enough that a single focused experiment can produce real evidence, not just suggestive patterns.

Empirical first

Data-driven

Theory informs the setup, but the data decides. Hypotheses are stated before experiments run, not after results arrive.

Compounding record

Accumulates

Each experiment builds on prior ones. A running record of what has been tested and what it showed is more valuable than isolated results.

What Gets Published

Traceable results

Reproducible

Every result comes with enough method detail to reconstruct. Data sources, parameters, and decision points are explicit.

Reproducible method

Open code

Notebooks and code are published alongside results. If it can't be reproduced, it's a note — not a release.

Honest failure

Including negatives

Experiments that don't confirm the hypothesis are published. A null result is still a result, and often the more informative one.

No repackaging

Original only

No summaries of other people's work. No passive learning notes dressed as research. Original implementation and first-hand observation only.

Editorial Boundary

Operating Loop

01 — Choose

Pick

Pick one market phenomenon, instrument, or mechanism to study.

02 — Define

Set

Set one hypothesis and the key questions around it.

03 — Run

Execute

Execute the smallest real experiment that produces evidence.

04 — Publish

Record

Record the result, even if it is partial or failed.

Behind the Work

Applied Markets is a personal initiative, maintained alongside Prachalabs.com and Pracha.me. A public record of ongoing research — not a consulting practice, not a publication venue, and not a trading operation.

Every release comes from original experiments, direct implementation, and first-hand observation. Not summaries of papers, not passive learning notes.