Direct technical context
The same person who scopes the question reads the model evidence, traces the data, writes the findings, and handles the technical handoff.
Founder · Direct delivery
Evgenii Azarov
The person who scopes your review is the person who runs it.
StatGazer is my one-person firm by design. I validate models for investment teams and review research statistics for people who publish — one reviewer across econometrics, statistics, machine learning, and financial engineering, with twenty years of private teaching and consulting behind every engagement.
20 years teaching & consulting, since 2006 · 100+ students from five countries · NDA-first, founder-only access · Insured — E&O and cyber (Hiscox)
- Delivery
- Founder-led scoping, technical review, research, engineering, and handoff for family offices, investment teams, and researchers.
- Track record
- Independent model and backtest reviews for family offices and investment teams; statistical consulting for research teams and individual researchers; private teaching in econometrics, statistics, R, Python, and financial engineering since 2006 — 100+ students from five countries.
- Credentials & fields
- PhD in Law, Russia (2012); finance training at NYU. Econometrics, statistics, machine learning, financial engineering, software engineering.
- Public links
- LinkedIn · GitHub · Google Scholar · ORCID.
Why founder-led
No bench handoff between the sale and the work.
For model validation, backtest review, and quant research, the expensive mistakes often live in details: timestamp alignment, undocumented assumptions, fragile validation, unreproducible notebooks, and ambiguous model purpose. Those details are easy to lose when the person selling the engagement is not the person doing the review.
Senior attention by default
StatGazer is intentionally small. The model is not a large team selling junior leverage; it is focused review, research, and engineering for problems where quality matters more than headcount.
Clear limits
When a request is outside StatGazer’s role, the firm says so. The service is technical consulting, not legal advice, tax advice, investment advice, or regulatory assurance.
A note from the founder
I started StatGazer after twenty years of teaching and consulting, because the expensive mistakes I kept finding were never exotic. A feature that quietly used tomorrow's data. A split that flattered the model. An interpretation the outputs didn't actually support. Catching those before money or credibility depends on them is careful, unglamorous work — and it fails the moment it is delegated down a bench. So I kept the firm small on purpose: the person you talk to is the person who reads the code, traces the data, and signs the findings.
Outside client work I still teach — a hundred-plus students across five countries — and build small software end to end, because reviewing other people's pipelines honestly requires shipping your own. If a model or a paper of yours is about to be judged, that is the moment this firm was built for.
— Evgenii Azarov
Public work
Public work.
Everything below is public and runnable. Client work never is.
Common risk factors in cryptocurrency
Reproducible R pipeline: cross-sectional predictors, factor models, attention, macro–sentiment–hashrate and blockchain network factors.
Valuation of BIM effects in PPP projects
DCF, scenario analysis, Monte Carlo, meta-regression and real options in R (MIT licence).
Working standard
Evidence your team can inspect.
Prospective clients should not have to trust adjectives. The standard is practical: a model review should say what was checked, what failed, what held up, what could not be verified, and what should be fixed first.
What the work emphasizes
- Point-in-time correctness, leakage checks, survivorship, and data lineage.
- Validation design, out-of-sample behavior, calibration, and model limitations.
- Reproducible notebooks, code paths, and documentation that travel with the result.
- Written findings that distinguish facts, assumptions, unresolved risks, and recommendations.
Public proof artifacts
The site includes a synthetic sample findings memo and public synthetic notebook reference so buyers can inspect the format without sharing client data. Synthetic artifacts are not a track record or client case study; they show the standard of documentation, reproducibility, and technical framing.
The discipline borrows its vocabulary from model-risk practice — the Federal Reserve's current SR 26-2 guidance names the same things that matter outside banking: independent challenge, conceptual soundness, validation evidence, limitations, and documentation. StatGazer uses that as a practical parallel, not as a claim of regulatory assurance or certification.
Professional boundary. Technical model review, validation, research, and engineering consulting — not financial-statement audit, regulatory assurance, investment, legal, or tax advice.
Company facts
Built for vendor review before confidential work starts.
Institutional buyers often need basic facts before a scope is approved: entity, jurisdiction, contact route, data handling, and the boundary of the service. StatGazer keeps those facts visible rather than hiding them behind a sales call.
Entity and contact
- Legal entity: StatGazer LLC, a New York limited liability company.
- Operating base: New York, United States; operating globally.
- Direct contact and vendor-onboarding route: hello@statgazer.com.
- Professional liability (errors & omissions) and cyber liability coverage in place, underwritten by Hiscox. Certificate of insurance — with additional-insured status where your vendor policy requires it — available on request.
- Vendor documents on request: W-9, certificate of insurance, NDA.
Data handling
Engagements are NDA-first. Where possible, StatGazer works inside the client environment or on the smallest useful extract. Access is founder-only unless otherwise agreed in writing. Client data is used only for the engagement and is not used for third-party AI training.
How buyers should evaluate the firm
Ask for evidence before sending sensitive data.
Verify fit from public artifacts first — the sample memo, the review standard, the notebook, the profiles — and only then share confidential material.
What to verify
Confirm the entity, contact route, public profiles, scope boundary, data handling terms, and whether the engagement will be founder-delivered. Those are the facts that matter before NDA and access.
What to include in your enquiry
Ask how the review would be scoped, which artifacts are required, what would be considered out of scope, and how findings would be written for both technical and decision audiences.
What not to infer
Do not infer audit assurance, investment advice, legal advice, tax advice, or guaranteed model performance. The role is independent technical review, research, engineering, and documentation.
Next step
Start with the technical question.
Use the consultation form for high-level context only. We reply by email and agree a written scope before any paid work begins. Sign an NDA before sharing confidential materials. You can also write directly: hello@statgazer.com.