How to read the lab — an 84-second introduction

← Back to the lab How to read the lab · 1:24

Quantitative Markets & Institutions Lab

Independent research program

Research on financial models, market regimes, institutions, and investor decisions.

Executed investigations, interactive models, and a versioned research record — published with the evidence needed to challenge them.

Examine the researchExplore the models

06Executed investigations

04Empirical studies

02Numerical validations

05Interactive models

Derivatives · Stochastic assumptions

Constant-volatility dynamics under empirical tests

≈3% → 93% realized vol

Portfolio construction · Estimation risk

Portfolio optimization under estimation error

1.12 → 0.44 Sharpe

Market risk · Forecast calibration

Historical VaR under coverage and independence tests

135 vs 107 breaches

Factor models · Product inference

ETF factor exposure and alpha after multiple testing

0 alphas at 5% FDR

Quantitative Markets & Institutions Lab

Working instruments

Interactive models

Option valuation

Interactive

Use it to: Price European and American contracts; watch Greeks respond as assumptions move.
Do not use it to: Quote executable market prices. These are model values under stated assumptions.

Open the pricing model →

Portfolio geometry

Interactive

Historical VaR backtest

Empirical

Factor exposure

Empirical

Stochastic processes

Interactive

Regime workbench

Interactive

Quantitative Markets & Institutions Lab

Contract & market

Spot100.00

Strike100.00

Volatility σ← live

Maturity1.00y

Rate r4.00%

Model values

6.91Black–Scholes · σ 12.0%

6.90CRR binomial

0.01Difference

9.62Black–Scholes · σ 19.2%

9.61CRR binomial

0.01Difference

12.99Black–Scholes · σ 28.0%

12.97CRR binomial

0.01Difference

Quantitative Markets & Institutions Lab

Research program

Seven published results from six executed investigations

Derivatives · Stochastic assumptions

Constant-volatility dynamics under empirical tests

≈3% → 93% realized vol

Market risk · Forecast calibration

Historical VaR under coverage and independence tests

135 vs 107 breaches

Portfolio construction · Estimation risk

Portfolio optimization under estimation error

1.12 → 0.44 Sharpe

Numerical methods · Simulation error

Monte Carlo convergence, bias, and variance reduction

Error ∝ N⁻¹ᐟ²

Quantitative Markets & Institutions Lab

Empirical finding · Market-risk validation

Historical VaR under coverage and independence tests

01 Conclusion02 Evidence03 Decision consequences04 Identification and limits05 Method, mathematics, and code

135Breaches observed

107Expected at 95%

6.31%Breach rate

2,140Forecast days

Exception record · 2,140 forecast days

Mark placement is schematic — the record stores transition counts, not exception dates.

Quantitative Markets & Institutions Lab

Validation protocol

Five stages from research design to publication

01
Frame a falsifiable question
The test comes before the conclusion.
02
Declare data, uncertainty, and failure criteria
Stated first, in writing, before estimation.
03
Execute the pre-registered design
A refreshed dataset is treated as a new experiment.
04
Classify the evidence
Six classes prevent context from masquerading as proof.
05
Publish the boundary
Every result ships with its identification limits.

Quantitative Markets & Institutions Lab

Versioned research ledger

Research record and reproducibility ledger

A versioned record of every question, method, result, and revision.

Investigation 01 · Cross-method option-pricing validation

Falsification criterion: Reject if any two methods disagree beyond the declared tolerance.
Result: Black–Scholes, CRR, PDE, and Monte Carlo agreed within tolerance.
Status: Validated

Investigation 06 · Historical VaR coverage and independence backtest

Falsification criterion: Reject if Kupiec or Christoffersen rejects at the 5% level.
Result: 135 breaches against 107 expected; the exceptions clustered.
Status: Rejected

Shared reproduction protocol

$ git clone https://github.com/kwisniewski-ops/quant-finance-research-lab.git

$ git checkout ad24c49995832f84ebeceb2f8243011463591a4a

$ pip install -e ".[all]"

$ pytest

Quantitative Markets & Institutions Lab · 2026 Educational research; not investment advice.

Where to begin

Start at Overview.

Overview · Explore · Research · Methods · Record

Educational research; not investment advice.

0:00 / 1:24

Transcript · 12 captions · 1:24

0:00	The problem	Financial models pass validation. Passing validation does not make the model the market.
0:07	The problem, continued	Bonds do not protect a portfolio in every kind of downturn. There is more where that came from.
0:14	The product	Introducing the Quantitative Markets & Institutions Lab. Five tabs. No login. No forecasts.
0:21	The features	Five interactive models. Each one lists what not to use it for.
0:28	The demonstration	Change an assumption; watch the answer move. That is the entire demonstration.
0:36	But wait	There is more. Seven published results from six executed investigations.
0:43	The catch, disclosed	This model failed. 135 breaches against 107 expected — and the exceptions clustered. It was published anyway.
0:51	The guarantee	The test comes before the conclusion. Question, data, uncertainty, and failure criteria — declared first.
0:58	The track record	Two of six investigations are stamped Rejected. This is the product working as designed.
1:06	The support line	Every entry is pinned to one commit. Operators are not standing by. The notebooks are.
1:13	The price	Nothing. Educational research; not investment advice. Refunds do not apply.
1:18	Where to begin	Start at Overview. Read one result. Every result ships with the evidence needed to challenge it.

How to read the lab

Five tabs, in reading order.

01OverviewThe front door: what the program studies, and seven results at a glance.
02ExploreFive working models. Change an assumption; watch the answer move.
03ResearchThe results index — failures filed beside the wins.
04MethodsThe validation protocol: the test comes before the conclusion.
05RecordThe versioned ledger. Every entry pinned to one commit.

Start at Overview. Educational research; not investment advice.

Start at Overview → Go straight to the research →