AGENFinished
- Final return
- −53.9%
- vs similar-stock shorts
- −56.5 pp
Dates & source
- Fired
- Jul 14, 2026
- Exit date
- Oct 12, 2026
Sell short after an FDA clinical hold
Sell short Sell When the FDA Puts a Trial on Hold · Company filings
Selling short: it profits when the stock falls — shares are borrowed to sell and bought back later. It needs a brokerage account approved for short selling (a margin account), and you can lose more than the money you put in.
This sells the stock short, aiming to profit if it falls. A clinical hold is the FDA stopping a drug trial, or refusing to let it start, over a safety or quality concern. This sells the company's stock short for three months the first time it reports the hold, and measures the result against a basket of other biotech stocks (the XBI fund) rather than the big-pharma-heavy health-care sector, because that is the company's real peer group. A report that says the hold was lifted is the opposite news and never counts.
Technical details
Source: EDGAR text flags (Wells notice, subpoena, civil investigative demand, reverse split, FDA hold / CRL / warning letter / breakthrough, strategic alternatives, buyback, dividend raise / cut, convertible notes, ransomware, short-seller reply, activist settlement, special dividend, forbearance, covenant waiver, guidance withdrawal, Paragraph IV; 8-K, 10-Q, 10-K and 6-K)
Newest idea 51 days ago
Updated Oct 16
On this page
How the test works
The planned holding period in these tests is 3 months (90 days), from opening to closing an idea.
Sell short when it fires
Sell borrowed shares at the next trading day’s close after the record posts.
Hold 3 months
The same hold for every idea, with no target. One exception: in these tests a short is bought back early if the price doubles — a −100% result.
Close it 3 months later
Buy the shares back 3 months after the sale, on the next available trading close.
Every number on this page uses this sell rule — the other sell rules we tested are under “The full test results and the math” below.
How it did in past tests
Research only — not yet confirmed by new trades
Publication check: Research only
Not passed yet — 4 of 16 checks:
- Some trades have no price we could check, often at companies that later stopped trading. The range shown counts only the priced ones; counting the rest at a poor result pulls its low end below zero.
- It needs at least 20 finished trades on 10 separate days after the rule was frozen.
- Some of those new trades are still open; each has to finish its full hold.
- Those new trades then have to beat similar stocks after costs, by more than luck across every rule we tried.
Passed — the other 12: Current calculation · Public disclosure dates · Complete finished trades · Every trade compared · Enough separate trading days · Positive edge after costs · Survives without the best year · Survives without the best three stocks · Survives higher costs · Rule frozen before the test · Rules tried alongside it counted · First new signals set aside.
Tested exit: 3-month hold. Historical results alone do not approve a rule for publication. New event tests include an estimated 0.20 percentage point round-trip trading cost, with a 0.50-point stress check.
New trades so far: 0 finished, across — separate days after the rule was frozen.
0 signals set aside across 0 disclosure days; 0 still need complete results. Missing results stay in the count.
Left: did worse · right: did better. Zero matches the comparison. pp = percentage points; the scale adjusts to include the result.
In past tests, selling each stock short made
−0.3%
on average over 3 months · 222 tries
Selling similar stocks short instead returned
−4.7%
the short position did 4.4 percentage points better
Made money in
107 of 222
tries (48%)
222 tries finished the 3-month test · 8 still open · 1,675 signals ever
Past results don't predict future ones.
This strategy is measured against XBI.
| How it did against similar stocks | 4.4 percentage points better than selling similar stocks short — over 3 months |
|---|---|
| Paid for borrowing the shares? | Yes, about 1.1 percentage points off the average. Every figure above already takes it out. It is our estimate, from how easy each stock was to borrow. Without the fee: 5.5 percentage points better than selling similar stocks short. |
| Range for the average (95%) | +0.5 pp to +8.2 pp vs selling similar stocks short — the range the average probably falls in, not a best-to-worst case |
| Tested on enough tries? | 222 tries on 190 separate days ✓ enough |
| Works in different markets? | 50% from a single year ✓ spread across years |
| Spread across many stocks? | 14% from a single stock ✓ broad |
| Still works without its biggest winner? | +5.1 pp vs selling similar stocks short without ARVN ✓ holds up |
The 5 newest of 222 finished ideas: 3 of 5 beat selling similar stocks short — one block each, newest first
- AGEN · Jul 14, 2026 · Return −53.9% · −56.5 pp vs similar-stock shorts
- MGTX · Jul 8, 2026 · Return +23.6% · +16.2 pp vs similar-stock shorts
- CRNX · Jul 7, 2026 · Return −2.7% · −3.0 pp vs similar-stock shorts
- BBIO · Jul 6, 2026 · Return +14.6% · +11.7 pp vs similar-stock shorts
- EDSA · Jun 12, 2026 · Return +12.1% · +29.4 pp vs similar-stock shorts
did betterdid worseno comparison
AGEN · Jul 14, 2026 · Return −53.9% · −56.5 pp vs similar-stock shorts
Inspect all 5 displayed outcomes
pp = percentage points above or below the comparable short-position return.
- AGEN · Jul 14, 2026 · Return −53.9% · −56.5 pp vs similar-stock shorts
- MGTX · Jul 8, 2026 · Return +23.6% · +16.2 pp vs similar-stock shorts
- CRNX · Jul 7, 2026 · Return −2.7% · −3.0 pp vs similar-stock shorts
- BBIO · Jul 6, 2026 · Return +14.6% · +11.7 pp vs similar-stock shorts
- EDSA · Jun 12, 2026 · Return +12.1% · +29.4 pp vs similar-stock shorts
Ideas for the week of Oct 12
This week's list isn't set yet — it's set on the first fresh data of the week of Oct 19.
This rule published no ideas for the week of Oct 12.
Every idea
Results under the 3-month hold · 8 still open · 222 finished · each idea's buy-back day is 3 months after it fired · an open idea shows its result so far · a short position’s result is what selling the stock short made.
MGTXFinished
- Final return
- +23.6%
- vs similar-stock shorts
- +16.2 pp
Dates & source
- Fired
- Jul 8, 2026
- Exit date
- Oct 6, 2026
CRNXFinished
- Final return
- −2.7%
- vs similar-stock shorts
- −3.0 pp
Dates & source
- Fired
- Jul 7, 2026
- Exit date
- Sep 1, 2026
BBIOFinished
- Final return
- +14.6%
- vs similar-stock shorts
- +11.7 pp
Dates & source
- Fired
- Jul 6, 2026
- Exit date
- Oct 5, 2026
EDSAFinished
- Final return
- +12.1%
- vs similar-stock shorts
- +29.4 pp
Dates & source
- Fired
- Jun 12, 2026
- Exit date
- Sep 10, 2026
| Stock | Fired | Status | Buy-back day | Result | vs similar-stock shorts (pp) |
|---|---|---|---|---|---|
| AGEN | Jul 14, 2026 | Finished | Oct 12, 2026 | −53.9% | −56.5 pp |
| MGTX | Jul 8, 2026 | Finished | Oct 6, 2026 | +23.6% | +16.2 pp |
| CRNX | Jul 7, 2026 | Finished | Sep 1, 2026 | −2.7% | −3.0 pp |
| BBIO | Jul 6, 2026 | Finished | Oct 5, 2026 | +14.6% | +11.7 pp |
| EDSA | Jun 12, 2026 | Finished | Sep 10, 2026 | +12.1% | +29.4 pp |
pp = percentage points: the short-position return minus the comparable short-position return. A dash means the comparison is unavailable.
Showing 5 of 230 ideas
225 more ideas, each with its result
Members see every idea this rule fired, open and finished, under any of the sell rules we tested — and can download them.
Already a member? Sign in
The full test results and the math
pp means percentage points: the difference between two returns. Actual returns are shown in %.
Average result (3 months)
−0.3%
range for the average (95%): −4.6% to +4.1%
Edge over selling similar stocks short
+4.4 pp
range for the average (95%): +0.5 pp to +8.2 pp
Held a year instead (selling short): −3.9% on average (170 tries).
Steadiness (Sharpe ratio) −0.02 — choppy: how even the results were from one try to the next.
Every buy-back rule we tested
Every sell rule we tested, side by side, comes with membership.
Risk and style checks
- After style factors (market, size, value, momentum…) the edge is +3.9 pp instead of +4.4 pp — the edge holds up after them. Tilts: Size −1.46 · Market −1.22 · Value +0.46 · R² 0.14.
- The stocks it picked trade a median $17.8M a day; 0% of its ideas were in thin stocks (under $2M a day).
- At one percent of that volume a trade is about $178K; with 8 ideas open now, an equal-weight book is about $1.4M.
- The edge by market fear on the day it fired (the VIX): calm −0.8 pp (43 tries) · normal +8.4 pp (119 tries) · stressed +4.6 pp (22 tries).