ADJUSTED PRICE INDEX / NOVEMBER 11 = 100 · NOT DOLLAR PRICES
Loading verified Webull observations…
BACKFILL / FDA SHORTAGE RESEARCH
Our hypothesis
When a drug enters shortage, manufacturers that can still supply it may capture demand from disrupted competitors. We test buying confirmed available suppliers after a 20-session wait, holding five sessions, and shorting SPY in proportion to each stock's prior beta to reduce broad market exposure.
EXPLORATORY REFINEMENT · NET OF COSTS · FULL RESULTS BELOW
01 · MARKET / ACTUAL WEBULL DAILY OBSERVATIONS
HELENE BENEFICIARY CASE / OCTOBER–NOVEMBER 2024
Actual stock movement
Webull · historical cache
Adjusted daily prices indexed to 100 at the November 11 entry close, so stocks with different dollar prices can be compared. Every point is an observed trading session; no simulated intraday ticks or smoothed prices. This is the previously inspected case, not a new blind test.
02 · RESEARCH / SAVED IN-SAMPLE RESULTS
MEASURE THE FULL STRATEGY
Strategy results
Loading saved research
CALENDAR-TIME PORTFOLIO
Net cumulative return
Selected strategySame-date controlsJUN 2014 — SEP 2024
Loading saved daily NAV…
Exact trading rules & source record +
OUT-OF-SAMPLE STATUS
The 2024–2026 period was previously inspected. Refined candidates require genuinely unseen events for confirmation.
03 · LATER PERIOD / EXPECTATION & OBSERVED MOVEMENT
BAXTER DISRUPTION → OTHER SUPPLIERS
The trade and SPY hedge
Exploratory · previously inspectedOCT 2024 — OCT 2026 / LATER-PERIOD AUDIT
Demand finds another supplier.
Available competitors may capture displaced demand and earn positive aggregate returns after a market hedge and costs. This is a qualitative hypothesis; no numerical price forecast was recorded.
Oct 10 · supplier statementOct 11 · announcement date usedNov 11 · illustrative entryNov 18 · illustrative exit
No qualifying trade.
Loading the evidence audit…
Zero eligible trades means performance is unavailable. This case illustrates the economic idea; it does not validate the strict strategy out of sample.
EXPECTATION → OBSERVATION
Open the saved outcome.
Standard-cost saved net hedged returns on invested stock capital. FMS uses a recorded Yahoo adjustment fallback; ICUI and SPY use Webull. Opening this chart does not perform a new holdout evaluation.
Reveal the daily saved beneficiary results. The endpoints were already inspected during research.
NOVEMBER 11–18, 2024 / FIVE TRADING SESSIONS
The beneficiary week versus SPY.
NET HEDGED / SAVED ACCOUNTING—Return on invested stock capital
STOCK-ONLY BASKET / SELECTED SOURCE—Equal-weight FMS + ICUI
AHEAD OF SPY / BEFORE COSTS—Percentage points, not hedged P&L
SAVED DAILY RECORDS
Hypothetical positionsNet beta-hedged movement
Reduce market exposure.
The hedge separates some broad US market movement from supplier-specific performance. Each stock's hedge uses its estimated beta from 250 returns ending before entry.
SPY fell 1.77% during this window, so the short hedge contributed positively. In a rising market it can reduce returns. Sector and company risks remain.
Contributions reconcile with saved trade cashflows. Costs include stock and hedge transactions plus financing carry.
Stock + hedge − costs.
| Beneficiary | Stock | SPY hedge | Costs / carry | Net |
|---|
FMS beta 0.536 · ICUI beta 1.648. The basket combines equal stock capital, with each stock's own hedge.
5% total long allocation, 2.5% per beneficiary. Linear illustration of the saved result; additional impact and capacity constraints are not modeled here.
Other previously inspected timings · all three horizons +
The short-window case is one of three inspected horizons, not an independent blind confirmation. These are the saved beneficiary-only results.
| Entry / hold | Entry → exit | Costs | Stock basket | Net hedged |
|---|
Case source vendors, adjustment policy & SHA-256 records +
04 · EVIDENCE / THE OBSERVATION LEDGER
A SOURCE BEHIND EVERY ROW
Trace the shortage.
| Observed | Drug / product | Listed owner | Supplier evidence | Timing | Inspect record |
|---|---|---|---|---|---|
| Loading the frozen ledger… | |||||
Executed trades for the selected research scenario +
| Ticker | Entry → exit | Beta | Entry capital | Net lot return | Net P&L |
|---|
05 · LIVE / TIGER DATA
OBSERVE THE NEXT CHAPTER
Fresh information.
Visible provenance.
Checking connection
Open live quotes, supply notices & market history +
MARKET OBSERVATIONS
SOURCE-TIMESTAMPEDSupplier watchlist
SUPPLY OBSERVATIONS
Latest notices
Point-in-time records from archived FDA shortage pages, not live trading signals.
WEBULL · RECENT DAILY CLOSES
WebullThe four injectable specialists
TIGER DATA / ACTUAL STORED QUOTES
Live market history
No sourced observations received.
Awaiting a configured market feed.
Latest seven days from one recorded vendor. The chart updates only when actual sourced quotes are received; the historical research stays frozen.
Tiger Data stores incoming observations. Quote vendors and FDA notices remain the original sources. A database connection alone does not establish a live market feed.
06 · RISK / CONCENTRATION & CAPACITY
UNDERSTAND THE EXPOSURE
An edge needs boundaries.
Capital discipline
Reduce gross after a 10% drawdown; restore after 20 sessions. Caps constrain entry allocations; closing prices can subsequently drift.
Company contribution
Capacity estimate
—Minimum order-based capacity at 1% of lagged ADV.
Participation capacity only. Spreads are estimates; market impact is not calibrated.
07 · VERIFY / OPEN THE RECEIPTS
MAKE THE RESULT REPRODUCIBLE
Inspect. Download.
Reproduce.
The recorded vendor mix
Download price manifest ↓Run the same accounting
uv run python -m research.final_candidate.runRequires the exact frozen licensed price cache. A new vendor download is a different data vintage. This view never launches a backtest.
PROOF ON SOLANA · FREEZE-V1
Loading receiptThe frozen rules, timestamped on chain
Proves the frozen rules existed at the anchor time. It does not make the 2014–2024 backtest out-of-sample.
Not checked yet
RECOMPUTED FROM GIT TAG
—ON SOLANA DEVNET
—Recomputes the fingerprint from the freeze-v1 git tag and reads the memo live from Solana devnet. Results are cached for 60 seconds.
uv run python scripts/verify_proof.py --tag freeze-v1
TWO INTEGRITY CHECKS, TWO JOBS
The fingerprinted files +
RESEARCH AUDIT · EVERY VARIANT TRIED
Loading auditEverything we tested, not just the winner
Nine specifications were registered before testing. Rows marked “chosen after seeing results” were picked after the registered results were known.
| Specification | Status | Trades | Sharpe 1× | Sharpe 2× | Total return | Max DD | Winner p | Winner − placebo p |
|---|---|---|---|---|---|---|---|---|
| Loading… | ||||||||
source: —
Frozen inputs check · freeze.json · 8 files +
FROZEN SOURCE INTEGRITY
—MATCHING SOURCE HASHES
Full research comparisons & declared limitations +
| Rule | Price policy | Costs | Sharpe | Positions | HAC p |
|---|