Loading…
Market Lens
PIE × Equity
How PIE coverage lines up with sector-relative returns across 20 defense-prime, drone-pure, and chip-supply tickers. Doctrine corpus depth is shown alongside as context. Source of truth: docs/research/market-lens/REPORT.md.
—active tickers
—PIE window (weeks)
—universe
—doctrine docs
Methodology & caveats
Window: PIE coverage spans Apr 5 – May 26, 2026 (8 weeks, 36 trading bars). Equity returns and alpha are computed strictly inside this window.
How to read the charts
- Alpha scatter — x: PIE mentions (preds + briefs), y: alpha vs sector benchmark (ITA for primes/drones/intl, SOXX for chip-supply). Dashed line at y=0 is the benchmark; above = outperforming, below = underperforming.
- Doctrine docs — shown as a context column in the table (not a chart axis): the count of doctrine-corpus documents mentioning the ticker. A level reading on how embedded a name is in defense doctrine, independent of this window's PIE coverage.
What this is not
- Not investment advice, not a backtest, not a recommendation. Coverage is not a signal of upside or downside.
- Not entity-resolved. PIE mentions are surface keyword tagging (e.g. “Lockheed” in a flag body counts as a mention of LMT). False positives and category confusion exist.
- No per-flag sentiment label. A bullish mention and a bearish mention count the same.
- Doctrine corpus has no per-doc timestamps in the current chunker. Doctrine docs is a level reading on corpus depth, not a momentum signal.
- Unadjusted closes from Yahoo Finance public chart endpoint — no dividend / split adjustment beyond what Yahoo provides.
Headline observations (descriptive only)
- Doctrine-heavy defense primes (LMT, NOC, BA, RTX) traded in line with or modestly under ITA over the window — doctrine corpus depth did not translate to PIE-window outperformance.
- The chip-supply cluster (INTC, STM, ON, QCOM) shows the biggest positive
alpha vs SOXX during the window, but this is a hypothesis worth investigating,
not a finding.
nof active days per chip ticker is small (often < 15) and PIE coverage of chip-supply names is itself sparse. - Same-day and 5-day-forward correlations between PIE mentions and returns are near-zero in absolute terms (|r| < 0.3 for nearly all rows). No leading or lagging relationship is supported by this window.
Full caveats and the source-of-truth numbers live in REPORT.md. The packager (pipeline/market_lens_packager.py) imports its window and palette from the research module so the page and report cannot drift apart on the basics.