MS Financial Engineering, New York University, May 2027
Quantitative research that only uses what was knowable at the time.
I work on rates, equity factors, volatility, and market structure, and I build the point-in-time data engines underneath them. Every result on this site is judged out of sample, and the ones that did not survive are reported next to the ones that did.
Seeking full-time quantitative research roles from summer 2027.
Latest research
The ten largest firms rose from 24% to 41% of S&P 500 market capitalization between 2018 and 2025, and by August 2026 connectedness among the eleven sectors had reached its lowest level in the sample. Does concentration explain the disconnection?
Concentrated sectors co-move less with the rest of the market: a one-point rise in top-three share lowers monthly R² by 0.75 points (permutation p = 0.04), though concentration does not explain why markets disconnected.
Research
Two more SSRN preprints and two working papers, alongside the paper above. Every result is out of sample, and every paper reports what did not work.
Naive inflation-regime conditioning overfits; empirical-Bayes shrinkage fixes it in the US, UK and Germany, and a blend with a market proxy reaches log-loss 1.056 and a 72.7% hit rate.
Built with only past data, capped leverage and trading costs, volatility management lifts momentum's Sharpe ratio from 0.43 to 0.94 and helps none of the other five factors. A closed-form formula for momentum's market beta explains why.
A full-stack platform that extracts risk-neutral moments from live option chains without a pricing model; a GARCH vol-targeted sleeve beat SPY on risk (Sharpe 0.85 vs 0.68) while naive short-variance harvesting lost money.
Four independent stress signals that must agree before de-risking: maximum drawdown -6.2% vs -41.7% for SPY, and the same calibration transfers to factor and international portfolios.
Master's thesis, in progress
MS thesis, New York University. Advisor: Prof. David Shimko. In progress, expected May 2027.
Tests in real time when debt dynamics turn self-reinforcing. The marketable Treasury debt is rebuilt bond by bond from 11,000+ auctions since 1979, and 39 macro series are read from their full revision histories, so no historical judgment sees data that had not yet been released.
Reads fiscal stress through the rates market: the Treasury volatility risk premium (implied versus realized) and a regime-switching stress state, each tested as a leading signal of the fiscal gap under multiple-testing control.
Simulates 30-year debt paths under two independently estimated macro engines, a vector autoregression and a regime-switching hidden Markov model, on common random numbers, pricing five policy responses against four shocks, down to sector employment and recovery times.
Projects
Research is only as honest as its data pipeline. These are the engines and live systems behind the papers.
SEC EDGAR, XBRL and Bloomberg fundamentals dated to the second the SEC accepted each filing. Four independent layers enforce the cutoff, and any violation fails at parse time rather than months later inside an inflated backtest. Survivorship-free, 65 tests.
A walk-forward regime-switching fundamental strategy on ChronoFund data, 2008 to 2026, net of 2/15 fees. Fama-French five-factor attribution reports alpha decay rather than hiding it: 8.37% (t = 2.73) in 2008 to 2017, fading to an insignificant 2.42% over the full period.
Infers the risk-neutral density of SPY same-day options and runs a scheduled live paper-trading loop. Every decision is a pure function of an as-of timestamp, and a test proves past decisions cannot change when future data is appended.
Sizes Polymarket and Kalshi binary contracts with Kelly on an empirical-Bayes win probability whose prior is a zero-edge bet, so no evidence means no position. Deterministic replay audits every strategy change; 64 tests with CI.
How I work
Every data read is stamped with the moment it became knowable. Filings are dated to the second the SEC accepted them, and macro series are read from their full revision histories, so a backtest cannot see a number before the market did.
Out-of-sample splits, heteroskedasticity-robust inference, and multiple-testing control. When a better test kills a result, the result goes, and the correction is published next to the finding.
Tested code behind every paper, full-stack research platforms, and scheduled live systems, built by one person who can explain every line.
Teaching and experience
Writing
A Markov chain over Fed policy moves and CPI regimes, walk-forward validated on first-release inflation data. Naive regime conditioning overfits; shrinkage recovers a small edge that replicates across the US, UK and Germany.
A plain-English primer on forwards, futures, options and swaps, with the everyday deals they come from.
Contact
I am graduating in May 2027 and looking for full-time quantitative research roles. The fastest way to reach me is email.
