Danny Hwang

EDGAR XBRL API for raw 10-Q/10-K data, Python (pandas, numpy, scipy) for balance-sheet analysis, Datawrapper and Chart.js for visualization, WordPress for publication. All quantitative work is reproducible from public SEC filings.

For me, the answer is narrow: to produce analysis that would still be correct if no one was watching. Retail investing content is flooded with claims sized for engagement rather than evidence — "X% of firms will go bankrupt," "Y is the new safe haven" — most of which don't survive a 10-Q pull. My work at TheFinSense is built around the opposite discipline: every claim ties back to SEC EDGAR XBRL, FRED, or peer-reviewed methodology (Beaver, Altman, Ohlson, Cathcart). If a framework fails under real data — as classical debt-to-equity does on 27 S&P 500 firms with negative book equity — I report the failure, not the workaround. That's the job.

The Q1 2026 Balance Sheet Stress Report. I pulled the most recent 10-Q or 10-K filings for 396 S&P 500 non-financial constituents — 97% of the universe — and found 27 firms reporting negative stockholders' equity. What made it worth writing wasn't the headline count; it was the failure mode it exposed. The classical debt-to-equity ratio is mathematically undefined when equity goes negative, which means the most widely cited leverage metric in retail finance content silently breaks on 6.8% of the S&P 500. Screening tools keep reporting these firms anyway, usually as "zero" or "N/A," and readers have no way to tell the difference between a healthy buyback program and a distressed balance sheet. The study proposes a three-regime framework — Normal, Thin, Broken — that makes the distinction explicit and applies Cathcart et al. (2020)'s large-firm default-probability gap as the correct benchmark for S&P 500 application, correcting a widely-misattributed figure in the process. The entire dataset and methodology are on SSRN and Zenodo (DOI: 10.5281/zenodo.19674350), so anyone can reproduce it.