Northeastern University · M.S. Quantitative Finance candidate
Zhe Chen
Preferred name · Juniper
Financial analysis, investment research & economic consulting
I am a quantitative finance master's candidate with investment research internship experience and hands-on analysis in Python and Excel. Based in Boston and considering entry-level opportunities across the United States.
How do allocation objectives change the balance of historical return, volatility and position concentration within one investment universe?
Data & method
I compared an equal-weight baseline with minimum variance, maximum return, maximum Sharpe and equal risk contribution using 501 aligned daily returns from six anonymized equities and ETFs. Historical prices came from Yahoo Finance via yfinance. A separate Black–Litterman-style exercise updated historical mean returns with two illustrative views before optimizing weights.
My work
For my individual course submission, I prepared the return series, ran the portfolio comparisons in Python, checked weights and risk measures, and interpreted the tradeoffs.
Main finding
The maximum-Sharpe portfolio reached an in-sample Sharpe of 1.37, with 73.6% in its largest holding. Minimum variance lowered annualized volatility to 14.7%, but still placed 52.0% in one holding. Lower estimated risk did not ensure broad diversification.
Limits
Weights were estimated and evaluated on the same sample. The comparison assumes fully invested long-only portfolios, fixed target weights applied to daily returns and no trading costs. It is not a forecast or live trading record.
Historical mean vs. volatility, with concentration
EW Equal weight · 16.7% max weightMV Minimum variance · 52.0%MS Maximum Sharpe · 73.6%MR Maximum return · 100.0%
Bubble area represents the largest holding. Mean is the 252-day annualized arithmetic average, not CAGR; Sharpe uses a 4.5% annual risk-free assumption. All figures are historical and in-sample.
How differently do two anonymized industry portfolios load on common equity factors, and does a broad market series show an asymmetric volatility response after negative returns?
Data & method
I used Kenneth R. French industry and factor data to fit three-, four- and five-factor regressions. Separately, I used Yahoo Finance market prices to compare GARCH and GJR-GARCH volatility models in Python.
My work
For my individual course submission, I aligned returns with factors, estimated the regressions and volatility models, and interpreted the coefficients and model diagnostics.
Main finding
Adjusted R² was about 0.69–0.70 for Portfolio A and about 0.88 for Portfolio B; adding factors changed in-sample fit modestly. For one broad market series, the GJR asymmetry estimate was 0.156 (p = 0.045), consistent with a stronger volatility response after negative returns in that sample.
Limits
These are in-sample descriptions under specific model assumptions. The p-value is close to a conventional threshold; no out-of-sample forecast comparison or causal effect was established.
How much variation did the factors explain?
Portfolio A anonymized industry returnsPortfolio B anonymized industry returns
Adjusted R² measures fit to the same historical observations used for estimation. It does not measure forecasting accuracy.
02 / Selected experience
Investment & research
Experience spanning company diligence, industry research, fund formation and ESG education.
June – August 2026
Wuhan, China
Wuhan Gaochen Investment Management Co., Ltd.
Investment Analyst Intern
Participated in on-site diligence for a technology company and supported review of its business model, financial profile, management and growth plans.
Prepared preliminary company research; participated in interviews and document review; summarized questions for follow-up diligence.
