About

Who is behind this site, why it exists, and how to get in touch.

Highlights
0.7376M drawdown model OOS AUCMarket risk →
+0.245mean 12M sector R²(OOS), recursiveSector outlook →
0.89 → 1.12Sharpe, benchmark → all three layersPortfolio lab →
Monthlylive model refresh since 2026Track record →
25 · 491pre-registered signals · stocks ranked monthlyStock screener →
4 marketsUS framework reproduced: UK, EU, JapanGeo allocation →

All figures are out-of-sample results from the three papers; each link opens the tab where the number lives, with its uncertainty and its misses.

Who runs this

I'm Alexander Uyar, an independent investment researcher. I'm completing an MSc in Finance & Investment at the University of Leeds (distinction average, CFA Institute Student Scholarship recipient), sitting CFA Level 1 in November 2026, with a BSc in Economics before that.

The site began as three papers — drawdown probabilities, sector forecastability, and within-sector stock selection — built with pre-registered designs and strict walk-forward evaluation. Rather than leave them as PDFs, I rebuilt the whole framework as a living system: every number here is a genuine out-of-sample model output, refreshed monthly, timestamped on the track record, and labelled honestly when it's a published result versus a live one.

My work combines quantitative methods — regularised regression, forecast combination, gradient boosting, portfolio optimisation — with fundamental analysis, and the equity research reports published here apply both.

No credentials are claimed beyond the work itself: every method, number and monthly call on this site is published and checkable.

The research

The site is the interactive companion to a three-paper research trilogy, estimated and evaluated out of sample on US data, then reproduced across the UK, Europe and Japan:

  • Market risk — leading indicators of S&P 500 drawdowns, combined into monthly drawdown probabilities. Paper (PDF) →
  • Sector outlook — out-of-sample forecastability of GICS sector excess returns and the stances it supports. Paper (PDF) →
  • Stock selection — within-sector stock ranking where the evidence is strong enough to trust. Paper (PDF) →
  • International reproduction — the same three layers re-estimated in the UK, Europe and Japan, with a geography-allocation layer above them: what transfers across markets and what does not. Geo allocation →

The papers are also collected on the SSRN author page.

What this site demonstrates
SkillEvidence on this site
Walk-forward econometrics & probabilistic modellingPapers 1–2: ridge-logit drawdown probabilities and sector return forecasts, every number strictly out of sample. Papers →
Multi-model forecasting & regularisationNine model families (penalised ARDL, forecast combinations, TVP, XGBoost) raced per sector and horizon. Sector outlook →
Cross-sectional signal research25 pre-registered characteristics on a 491-name point-in-time S&P 500 panel, FDR-controlled. Stock screener →
Validation disciplineThe live engine was reconciled against the papers before launch — it reproduces the published out-of-sample AUCs to ±0.003. Market risk →
Production data engineeringA monthly live pipeline on free data, publishing versioned data packs with an append-only vintage archive. Track record →
Portfolio constructionLayer stacking, cost and turnover analysis, mean-variance tools on the screened universe. Portfolio lab →
Open to research-driven investment roles.Download CV (PDF)
Every claim in this table is checkable on the linked tab — the site is the evidence.
Get in touch

Questions about the data or methodology are welcome — the feedback form works too.

What this site is not

Not an investment adviser, not a newsletter selling trades, and not a track record of personal performance. It is published research: model outputs, shown with their uncertainty and their misses, updated monthly so the record accumulates in public.