Biotech catalyst strategy - paper portfolio - 2026-10-19 update
📌 Snapshot of the Oct 19, 2026 update (orders for the next open). Always-current page with the full portfolio and daily log →
Updated after the US close of 2026-10-19 · paper trading only, not investment advice · research study
Account value
$1,004,450
Since start
+0.4%
BIO-EW since start
+1.0%
XBI since start
+1.4%
Cash
80%
Positions
5 / 25
Closed trades
0
| Since start | Book | BIO-EW | XBI | IBB | IWM |
|---|---|---|---|---|---|
| Return | +0.4% | +1.0% | +1.4% | +0.7% | -0.2% |
Rules (pre-registered strategy M, unchanged from the study): every week, US small and mid-cap biotech stocks (SIC 2834/2836/2835/8731, price ≥ $2, market cap $50M-$10B, ≥ $1M daily dollar volume, under $2B revenue) are scanned for an 8-K press release in which the company guides the timing of a clinical-trial readout ("topline data expected in Q1 2027"). Stocks whose guided window starts 40-100 weekdays ahead are ranked nearest first and the top 5 are bought at the next open, 4% of the account each (max 5% of the stock's daily dollar volume), up to 25 positions; idle money stays in cash. Each stock is sold at the open 3 weekdays before its readout window starts (the date planned at purchase), or at the next open after the company files an 8-K with topline results, an FDA approval, a complete response letter or a clinical hold - the trade is a pre-readout run-up, not a bet on the binary outcome. No stop-loss. Costs 0.75% per side under $2B market cap, 0.25% above. Data: SEC EDGAR 8-K full-text search + XBRL, Massive prices.
Honest caveats. The research study FAILED its own pass bar: from 2012 to 2026 the strategy beat the equal-weight biotech universe (BIO-EW) in every test period but lost to the XBI ETF in the 2023-26 holdout (11.9% vs 16.9% a year) and in most years; the gains came from a few big trades and from the candidate set rather than the ranking. This paper book tests whether the edge over equal-weight biotech persists live. Pre-stated review on 2028-10-19 (2 years): keep it only if it beats BIO-EW by at least 3 points a year after costs and is not worse than XBI by more than 5 points a year. The 8-K classifier misses some readouts, so some positions will still be holding when the binary result comes out.
Orders for the next open
Buy: none (new picks every weekend)
Sell: none
Open positions (5)
| Ticker | Bought | Value | Return | Readout window | Planned sell | Source |
|---|---|---|---|---|---|---|
| PEPG | 2026-10-19 | $42,035 | +5.9% | 2027-01-01 – 2027-06-30 | 2026-12-29 | Results from the 12.5 mg/kg cohort are expected in the first half of 2027, at which time PepGen plans to engage with regulators in an end of Phase 2 meeting to discuss plans for the registrational program. 8-K |
| ZURA | 2026-10-19 | $38,685 | -2.6% | 2027-01-01 – 2027-06-30 | 2026-12-29 | • TibuSURE enrollment complete, exceeding initial enrollment target with 91 participants randomized in SSc; topline data expected H1 2027 8-K |
| OLMA | 2026-10-19 | $42,256 | +6.4% | 2027-01-01 – 2027-03-31 | 2026-12-29 | • Completed enrollment in the pivotal Phase 3 OPERA-01 trial; top-line data are now expected in the first quarter of 2027 8-K |
| ORKA | 2026-10-19 | $41,046 | +2.9% | 2027-01-01 – 2027-03-31 | 2026-12-29 | The Company now expects to report Week 16 data from ORCA-SURGE, a Phase 2 trial in moderate-to-severe PsO, in the first quarter of 2027. 8-K |
| MGNX | 2026-10-19 | $40,428 | +1.8% | 2027-01-01 – 2027-06-30 | 2026-12-29 | The Company anticipates completing enrollment of these 20 patients by year-end 2026 and reporting updated study results in the first half of 2027. 8-K |
Daily log
| US close | Account | Return | BIO-EW | XBI | Cash | Pos. |
|---|---|---|---|---|---|---|
| 2026-10-19 | $1,004,450 | +0.4% | +1.0% | +1.4% | 80% | 5 |
| Bought 5: PEPG, ZURA, OLMA, ORKA, MGNX | ||||||
Claude picks vs random picks
Claude book (test): the same candidates, slot size, 25-position cap, exits and costs as the main book; only the selection differs. Every weekend each of the week's candidates gets a dossier (8-K guidance and press releases, current ClinicalTrials.gov records, recent papers, cash runway, offerings, other catalysts) and one Claude analysis (strongest available model, web search only) that estimates the chance the stock beats the equal-weight biotech basket by the planned sell date (p_runup), the chance of a binary event before it (p_early_shock) and a 0-100 score; the top 5 scores are bought. Scores are frozen before the Monday open. The main book's tie-order picks are effectively random within the same candidate set (the study found M's ranking no better than random), so this is Claude picks vs 5 random picks.
| Book | Account | Return | vs BIO-EW | vs XBI | Closed trades | Hit rate | Positions |
|---|---|---|---|---|---|---|---|
| Main book: 5 tie-order (random) picks | $1,004,450 | +0.4% | -0.6% | -0.9% | 0 | - | 5 |
| Claude book: Claude's top 5 | $1,004,895 | +0.5% | -0.5% | -0.9% | 0 | - | 5 |
Hit rate = share of closed trades with a positive return after costs. vs BIO-EW / XBI = return difference in points since the start (2026-10-19 open).
Claude's top 5 for the signals of 2026-10-16 (35 of 35 candidates scored; frozen 2026-10-16 22:55 New York time). Main-book picks this week: PEPG, ZURA, OLMA, ORKA, MGNX.
| # | Ticker | Score | p_runup | p_early_shock | Planned sell | Key reasons (Claude) |
|---|---|---|---|---|---|---|
| 1 | LBRX | 56 | 55% | 12% | 2026-12-29 | • FACT: NOVA-2 topline guided to 1H27 (reiterated 2026-09-15 and in the 2026-10-06 R&D Day release); CT.gov PCD is 2027-06 and the trial started 2026-03-25, so a readout before the 2026-12-29 sell date is unlikely. Early-shock risk … • FACT: Cash $327.8M at 2026-06-30 plus $150M PIPE at $34.94 closed ~2026-07-30 (runway guided beyond Q2 2029). JUDGEMENT: a dilutive raise inside the window is unlikely, though management has shown it will raise opportunistically. • FACT: Stock pulled back ~26% from $49.48 (2026-09-02) to $36.74 (2026-10-16), below its 50-day MA (~$43), after a Roth downgrade to Neutral/$35 (2026-10-05) and a Zacks 'strong sell' (2026-10-14); offset by Guggenheim ($80, 'best … |
| 2 | OLMA ★ | 55 | 55% | 15% | 2026-12-29 | • Pivotal OPERA-01 top-line guided to Q1 2027 (slipped from 'this fall' in May 2026); the planned sell date sits 3 trading days before the window, giving a classic pre-readout run-up window as the catalyst approaches. CT.gov … • Stock is at a 52-week low ($6.58, -82% from high, -33% in 20d) after the August data delay and the 2026-09-14 AstraZeneca SERENA-4 (camizestrant+palbociclib, 1L) failure; market cap $576M vs $461M cash (EV ~$115M) leaves a deeply … • SERENA-4 read-through is to the 1L combination setting (OPERA-02), not OPERA-01's post-CDK4/6i monotherapy-vs-SOC setting, where oral SERDs have a consistent class record (EMERALD/EMBER-3/VERITAC-2 all positive vs fulvestrant); … |
| 3 | TECX | 55 | 54% | 12% | 2026-12-29 | • Guided TX45 APEX Phase 2 topline 'early Q1 2027' (narrowed from 'late Q4 2026 or early Q1 2027' in May); last patient enrolled June 2026 + 24-week treatment puts last visits ~Dec 2026, so data before the 2026-12-29 sell date is … • Stock has sold off hard into the catalyst (-24% 20d, -35% 60d, -46% from 52-week high, $20.65 close) with UBS initiating at Neutral/$26 PT on 2026-10-11; no company-specific negative news found to explain the slide, which leaves … • Balance sheet strong: $227.1M cash at 2026-06-30 (~56% of $404M market cap), runway guided into Q1 2029; makes a dilutive offering before data unlikely, though the company did use its ATM ($11.7M net in Q2 2026). |
| 4 | DYN | 54 | 55% | 12% | 2026-12-29 | • Stock sits at its 52-week low ($14.25, -47% from high, -40% over 60d) after the 2026-09-08 Novartis/Avidity HARBOR Phase 3 miss on vHOT in DM1 drove a ~30% read-through selloff; this creates an oversold setup into two … • Dual catalyst stack just after the sell date: z-rostudirsen PDUFA 2027-01-21 (Priority Review) and ACHIEVE REC topline in Q1 2027; the holding period ends 2026-12-29, inside the typical pre-PDUFA run-up window. • Key risk to the bull case: ACHIEVE REC primary endpoint is 6-month vHOT, the same endpoint HARBOR failed on at 54 weeks; sentiment on DM1/vHOT remains impaired (Morgan Stanley PT cut to $36 in Sept; HCW cut to $32), so negative … |
| 5 | RNAC | 54 | 54% | 15% | 2026-12-29 | • Setup fits a classic pre-Phase-3 run-up: AURORA (128 pts enrolled, ACTIVE_NOT_RECRUITING, est. primary completion 2026-12-30) topline reaffirmed for 1Q27 in the 2026-09-29 8-K; the sell date (2026-12-29) sits just before the … • Stock is depressed: $6.27, -22% over 20d, -41% from 52-wk high, only +7% above 52-wk low, and fell from $7.28 (2026-09-30) despite positive retreatment data on 2026-09-29. No specific cause for the October slide was found … • Discrete-offering risk is moderate-low: $147.6M cash at 2026-06-30 vs ~$77M TTM burn, plus $50M K2 term loan drawn 2026-05-22 (company stated runway into 2028 per 2026-05-26 release seen in search). But the Leerink ATM is active … |
★ = also a main-book pick this week. Claude's reasons are model output with web search; they can be wrong.
Pre-stated review. 12-month check (2027-10-18): among all candidates whose planned holding window has ended, Claude's top-quintile scores (within each week) vs bottom-quintile scores, realised abnormal return = stock open-to-open from the buy day to the planned sell day minus BIO-EW over the same window; t-statistic with standard errors clustered by signal week. 24-month check (2028-10-19): the Claude book must beat the main book by at least 3 points a year after costs to be worth keeping. Calibration: p_runup vs the realised share of candidates that beat BIO-EW.
Quintile test: no completed holding windows yet (0 weeks so far; the first planned sell dates are at the end of December 2026).
Calibration table (p_runup vs realised): appears once the first holding windows have ended.
Shadow book V1 (test)
Shadow book V1 (tracked, not the real book): the same rules, but any scheduled catalyst counts - 8-K readout timing guidance, a PDUFA date or an FDA advisory committee meeting 40-100 weekdays ahead - and only topline-result 8-Ks trigger an early sale. In the study V1 beat M in 2020-22 and 2023-26, but that was found after the fact, so M stays the real book.
Shadow account
$1,004,340
vs real book
-0.01%
Shadow max drawdown
0.0%
Closed trades
0
Holds (5): ANAB, COGT, MLYS, PRAX, VNDA
Next open - buy: none; sell: none
Research (price-only, 2012-2026, after costs): strategy M 15.1% a year 2012-19 vs BIO-EW 7.6% and XBI 20.1%; 2020-22 0.0% vs BIO-EW -14.6% and XBI -4.3%; 2023-26 11.9% vs BIO-EW 2.7% and XBI 16.9%; max drawdown about -47%. Without its 10 best trades each period was roughly flat or negative. BIO-EW = equal weight of every eligible stock, rebalanced monthly with the same costs (here from the first open of the book). Not investment advice.
4 redactions applied (4 cards).
- cards: 4
Published via wiis. Originals are encrypted at rest; this page reflects the redacted version. report abuse.
Discussion
No comments yet.