phase-a-complete

Phase A & A+ Complete Report

Page ID: 322d5568-29bd-8153-92eb-df1b05970a9e
Created: 2026-03-13T05:53:00.000Z
Last Edited: 2026-03-13T05:53:00.000Z
URL: https://www.notion.so/Phase-A-A-Complete-Report-322d556829bd815392ebdf1b05970a9e

Properties

• Phase: A
• Status: Complete
• Key Finding: Filter set works. ATH drawdown + Pipeline Phase most important. IPO age 0.5-5Y boosts 180d mean to 27.4%. Fat tail 1.7x at >500%.
• Date: 2026-03-10
• Name: Phase A & A+ Complete Report

Content

Phase A & A+ Complete Report — Biotech Long Strategy

Date: 2026-03-10

Period: 2015-02-01 to 2025-12-01 (131 months)

Author: Lucky 🍀 (automated analysis)

---

1. Executive Summary

Strategy Thesis

Buy biotech stocks that have valid clinical pipelines, adequate cash runway, and have collapsed ≥90% from ATH — capturing the right-tail optionality of event-driven recovery.

Key Findings

1. The filter set works. Filtered universe (avg 23.8 stocks/month) shows statistically significant outperformance vs benchmark at 60d, 90d, and 180d windows (Mann-Whitney p < 0.001).
2. The edge is in the right tail. Mean returns are positive (1.2% to 9.7%), but medians are negative. The alpha comes from fat-tail events: 1.4x–1.7x higher probability of >50%–>500% returns.
3. ATH drawdown is the most powerful filter — removing it drops 180d mean from 9.7% → 6.5%.
4. IPO age 0.5–5Y dramatically boosts returns — 180d mean jumps from 9.7% to 27.4%.
5. Pipeline thickness ≥6 trials is the strongest enhancer — 180d mean of 28.3% (6-10 bucket) and 54.8% (11+ bucket), though sample sizes are small.
6. Phase 3 stocks outperform Phase 2 — P3+ 180d mean 16.5% vs P2-only 6.4%.
7. Recovery, catalyst proximity, and prior pump filters do NOT improve returns — they reduce sample size without meaningful benefit.

Recommended Phase B Filter Configuration

• Filter | Parameter | Value
• ATH Drawdown | threshold | -90%
• Cash Runway | min quarters | 4
• Pipeline Phase | minimum | Phase 2
• IPO Age | range | 0.5–5 years
• SEC Filing | required | Yes
• New: Pipeline Thickness | minimum trials | ≥3 (optional enrichment)
• New: Phase 3 Flag | flag for weighting | P3+ gets higher weight

---

2. Data Foundation

2.1 Universe Scope

• Source: FMP API → DuckDB (biotech/data/biotech.duckdb)
• Total tickers collected: 1,104
• After Category B exclusions: 1,054
• Exchanges: NASDAQ, NYSE, AMEX
• Sector: Biotechnology / Life Sciences

2.2 Data Quality: Category B Exclusion

Investigation identified 111 symbols with extreme price data:

• Category | Count | Issue | Fix
• A: Extreme but consistent | 61 | Correct split-adjusted prices, just very large numbers | Return cap (done)
• B: Corrupt/inconsistent | 50 | Mixed adjusted/unadjusted prices within series | Excluded
• Clean | 993 | No issues | None Category B symptoms: FMP returns a MIX of split-adjusted and raw prices for the same symbol. Example: PPCB oscillates between $0.20(adjusted)and$6,582 (raw) on consecutive days. Creates false >1000% daily returns.

50 excluded symbols: ABVC, AMAM, ARGS, AVTE, AXLA, BBI, BCTX, BOLD, CBLI, CELZ, COSM, CTOR, CWBR, CYBN, DRMA, DRUG, GLSI, GNCA, HGEN, HPCO, IFRX, IMCC, IMUX, JBIO, KMPH, KTRA, LIAN, LSDI, MCRB, MNMD, MTP, OTIC, PDSB, PKBO, PLRZ, PPCB, PRTG, QLI, QNCX, SILO, SISI, SPRB, SXTC, SYN, SYRS, TCDA, TMBR, TPST, VRPX, ZSAN

Return cap: Forward returns >5000% or < -99.99% are excluded as data artifacts.

2.3 Database Schema

• Table | Key Columns | Distinct Symbols
• daily_prices | symbol, date, open, high, low, close, volume | ~1,104
• balance_sheet | symbol, date, cash_and_equivalents, short_term_investments, total_debt | ~1,100
• income_statement | symbol, date, operating_income | ~1,100
• clinical_trials | symbol, phase, start_date, completion_date, primary_completion_date | ~800
• sec_filings | symbol, filing_date, filing_type | ~900
• insider_trades | symbol, transaction_date, owner_type | —
• stock_splits | symbol, date, numerator, denominator | —
• excluded_symbols | symbol | 50

---

3. Phase A Baseline Results

3.1 Filter Definitions

All 5 filters are applied as intersection (AND logic) on the 1st of each month:

• | Filter | Logic | Avg Passing

• 1 | ATH Drawdown ≤ -90% | current_close / max(all_historical_closes) - 1 ≤ -0.90 | 258.8/mo
• 2 | Cash Runway ≥ 4 quarters | (cash + short_term_investments - total_debt) / abs(operating_income) ≥ 4 | 376.7/mo
• 3 | Pipeline Phase ≥ Phase 2 | Active clinical trial in PHASE2, PHASE2/PHASE3, or PHASE3 at rebalance date | 243.3/mo
• 4 | IPO Age ≥ 2 years | first_trading_date ≤ rebalance_date - 730 days | 548.8/mo
• 5 | SEC Filing History | Has at least one SEC filing before rebalance date | 684.7/mo
• | Combined (intersection) | All 5 filters pass | 23.8/mo

3.2 Return Distribution: Filtered vs Benchmark

• Window | Group | Mean | Median | P75 | P90 | P95 | Max | N
• 30d | filtered | 1.2% | -3.0% | 10.3% | 28.5% | 45.4% | 525.0% | 3,119
• 30d | benchmark | -0.1% | -2.5% | 8.6% | 25.1% | 40.4% | 1228.0% | 84,516
• 60d | filtered | 3.3% | -4.7% | 16.2% | 46.6% | 68.5% | 597.8% | 3,119
• 60d | benchmark | -0.1% | -4.8% | 11.8% | 37.0% | 61.0% | 1396.1% | 84,518
• 90d | filtered | 5.5% | -6.1% | 20.5% | 59.3% | 94.0% | 1340.8% | 3,119
• 90d | benchmark | 0.0% | -6.9% | 14.0% | 46.2% | 77.9% | 2847.1% | 84,517
• 180d | filtered | 9.7% | -11.4% | 30.0% | 90.5% | 152.8% | 1005.9% | 3,119
• 180d | benchmark | 0.1% | -12.5% | 17.5% | 65.7% | 113.1% | 4507.3% | 84,517 Key insight: Medians are always negative for both groups — most biotech stocks lose money. But filtered stocks have substantially higher means, driven by fatter right tails.

3.3 Mann-Whitney U Test (filtered > benchmark)

• Window | p-value | Significant?
• 30d | 0.1207 | ❌ No
• 60d | 0.0007 | ✅ Yes
• 90d | 0.00001 | ✅ Yes
• 180d | 0.000000 | ✅ Yes The 30d window lacks significance — the filter edge needs time to manifest. Statistical significance strengthens dramatically with longer holding periods.

3.4 Fat Tail Analysis

• Window | Threshold | Filtered | Benchmark | Ratio
• 30d | >50% | 4.0% | 3.5% | 1.2x
• 30d | >100% | 1.0% | 0.9% | 1.1x
• 60d | >50% | 9.0% | 6.8% | 1.3x
• 60d | >100% | 2.9% | 2.1% | 1.3x
• 90d | >50% | 12.4% | 9.1% | 1.4x
• 90d | >100% | 4.5% | 3.3% | 1.4x
• 90d | >200% | 1.5% | 0.9% | 1.6x
• 180d | >50% | 17.9% | 13.0% | 1.4x
• 180d | >100% | 9.1% | 6.0% | 1.5x
• 180d | >200% | 3.1% | 2.0% | 1.6x
• 180d | >500% | 0.4% | 0.3% | 1.7x At 180d, filtered stocks are 1.7x more likely to deliver >500% returns.