DAPHNE G.S
I turn messy real-world data into clear strategic insights that founders can actually act on.
Curious by nature, I approach problems by looking at patterns hidden in unconventional markets. I believe the best insights come from places nobody notices.
Pattern recognition applied to markets analysts typically overlook.
The best signal usually sits in the data nobody thought to model.
Two independent investigations, both built from public data.
I turn messy real-world data into clear strategic insights that founders can actually act on — building frameworks that reveal how markets actually behave versus how we think they behave.
Interested in roles at startups solving unconventional problems — especially where rigorous analysis can uncover competitive advantages.
CASE FILES
METHOD: GRANGER CAUSALITY
STATUS: RESOLVED
Do darknet vendors systematically respond to Bitcoin price movements, or is pricing random?
109,098 marketplace transactions analyzed across volatile Bitcoin price cycles.
Python (Pandas, NumPy, Statsmodels) — Granger causality testing across a 1–4 day lag window.
Vendors respond within 24–48 hours with statistical precision — 95%+ confidence, clean one-way causality. Vendors are price-takers.
Economic rationality persists even in unconventional markets — participants actively monitor external signals and act strategically. The pattern generalizes to any marketplace with volatile input costs.
- Bitcoin prices predict darknet prices within 1–4 days at 95%+ confidence
- Vendors are price-takers — prices don't predict Bitcoin, clean one-way causality
- Price stability varies 6-fold by product category, revealing market structure differences
METHOD: SHA-256 DIFF
STATUS: RESOLVED
How often do SaaS companies change their pricing pages, and what does that frequency tell us about their monetization strategy?
10,000+ archived pricing pages from 13 leading SaaS companies, sourced via the Internet Archive CDX API.
CDX metadata harvest → deduplicated HTML download → SHA-256 content-change detection → per-company metrics aggregation → multi-dimensional visualization.
Repricing frequency varies 6× across companies — GitHub every 5 days, ConvertKit every 31. Developer tools reprice every 5–9 days; marketing tools every 31.
Pricing-page revision frequency is a behavioral signal — a proxy for how actively a company experiments with packaging, positioning and features. Useful for competitive intelligence, PMs benchmarking velocity, and investors reading operational maturity.
- Developer tools reprice every 5–9 days; marketing tools every 31 — a 6× variance
- Repricing frequency correlates with strategic maturity and monetization approach
- Historical patterns shift over time, indicating companies mature their pricing strategy
Question First
I start with a question nobody is asking, not the obvious one everyone already knows the answer to.
Find the Data
I locate public datasets, clean them, and validate that they actually answer the question.
Rigorous Analysis
I test hypotheses statistically. No assumptions — Granger causality, not correlation.
Extract the Insight
I translate statistical findings into business implications. Why does this matter? Who cares? What changes?
Present Clearly
I build visualizations that tell the story. No chart junk. Just signal.
AI TOOLS
Kept deliberately separate from each case file's technical stack — this is workflow tooling, not the analysis itself.
I look for the questions
hidden behind the numbers.
The interesting problems are rarely the obvious ones.
WHAT ARE YOU
TRYING TO UNDERSTAND?
Signal identified. Let's see what we can uncover next.