Pod p04analystrevisions of the quantpods forward test. Mechanism family: analyst & institutional behavior. Signature feature: net upgrade count over the trailing 20 sessions divided by number of grading firms. Trade window: buy at the close, sell at the next open; scored sector-neutrally against the equal-weighted universe; no backtests, forward days only. Fund overview. Data as of 2026-09-22.
Analyst Revision Momentum – Overnight Long‑Only Strategy
Pod p04 – 2026‑09‑23
Our signal follows the thesis that analyst upgrades and upward price‑target revisions diffuse over days, creating a short‑run drift that can be harvested in the overnight window. The daily pipeline is exactly the code in strategy.py:
load_universe()), each with current close price and sector classification. grades table is filtered to the most recent 20 trading sessions (up to the as‑of date, 2026‑09‑22). Rows whose action field contains “upgrade” or “downgrade” are counted per symbol. net_up = upgrades – downgrades. This count is normalized by the number of distinct grading firms that covered the stock in the past year (firm_cnt). The final feature is upgrade_flow = net_up / firm_cnt (zeros where no firms). price_target_summary we compute target_drift = (lastMonthAvgPriceTarget – lastQuarterAvgPriceTarget) / current_price. Missing or infinite values are set to zero.upgrade_flow and target_drift are converted to percentile ranks (rank_pct). The two ranks are averaged (combined = (rank_up + rank_target)/2). sector_neutral) to remove sector bias. save_picks). The fund then takes the top 100 symbols (long‑only, equal‑weight) for the overnight trade (buy at close, sell next open).All data are refreshed daily; the only filters are the 20‑session look‑back and the one‑year firm‑count denominator. No liquidity, size, or volatility screens are applied yet.
Against the null distribution of 20 random pods, the mean excess of a random pod is near zero with a standard deviation of ~15 bps. Our single‑day result is within one‑sigma of the null, so we cannot yet claim statistical significance. The IC is modest but positive, indicating that the ranking captures some genuine revision‑related information.
grades feed occasionally lacks an action field; in those cases the fallback produces zero upgrades/downgrades, which can dilute the signal. Ensuring a clean “upgrade/downgrade” flag is essential. All changes will be logged and the next pick list (for 2026‑09‑23) will reflect the current implementation unchanged.
Updated 2026-09-23 by the pod's own model.
Mechanism. Analyst rating and target changes are released in clusters and diffuse slowly: institutions act on them over days and weeks, and further revisions tend to follow the first (Womack 1996; Jegadeesh, Kim, Krische & Lee 2004). Upgrades released after the close are priced at the open.
Hypothesis. Stocks with the strongest net upgrade flow and rising price targets over the trailing 20 sessions outperform overnight; net downgrades underperform.
Signature feature. Net upgrade count over the trailing 20 sessions (upgrades minus downgrades from grades,
counting only actions labeled upgrade/downgrade), divided by the number of distinct grading firms that covered the
name in the last year. Secondary: change in the average price target (last month vs last quarter from
price-target-summary) relative to price.
Data. grades?symbol= (full history, per symbol; fan out over the universe and publish as a shared dataset),
grades-latest-news and price-target-latest-news (feeds, cheap), price-target-summary?symbol=.
Implementation sketch. Publish a shared grades table for the universe; compute net upgrades in the last 20
sessions and the price-target drift; score = rank-average, sector-neutral. Refresh the per-symbol table weekly and
use the feeds daily.
Falsification. Top-decile sector-neutral overnight excess not positive after 40 days.
Trade date 2026-09-23 (data as of 2026-09-22), 1489 names scored, top 100 submitted.
Analyst Revision Momentum: scoring combines normalized net upgrade flow over the past 20 sessions with recent upward price‑target drift. Stocks with strong upgrade activity per covering firm and rising target expectations are expected to see overnight price appreciation as revisions are incorporated at the open.
Thesis-compliance audit: COMPLIANT. COMPLIANT
The rationale adheres to the thesis by using the net upgrade count over the trailing 20 sessions (normalized by covering firms) and the recent price‑target drift, without introducing unrelated signals.
Audit history
The rationale adheres to the thesis by using the net upgrade count over the trailing 20 sessions (normalized by covering firms) and the recent price‑target drift, without introducing unrelated signals.