The Biotech Insider Playbook
Everyone tracks Nancy Pelosi's trades. Copying her 2026 buys made +1.8%. Copying biotech insiders buying their own stock made +57.3% in six months. We copied all 440 of their buy filings into a simulated $100,000 account, measured every fill down to the second, and wrote down every trade.
No cherry-picking, and no fills you couldn't actually get: a basic backtest claims +83%, and this article breaks down, step by step, why the number you'd really bank is the one above.
Why biotech insiders are a different animal
Insider buying is a decent signal everywhere. In biotech it's a different sport, because the companies themselves are all-or-nothing bets, and the people buying know which.
Binary catalysts
A biotech's value hinges on discrete events: trial readouts, FDA decisions, partnership deals. When they land, the stock doesn't drift 2%; it gaps 50–300% overnight. Insiders live next to those events.
You hear about it in 2 days, not 45
Corporate insiders must file SEC Form 4 within two business days of a trade (Congress gets 45 days under the STOCK Act). The information is fresh enough to act on. Our feed ingests filings from EDGAR within minutes.
Skin in the game is legible
A CEO wiring $99k of personal cash into their own sub-$3 stock, or a Vertex founder putting $3M into a small neuro company, is a costly, public signal. Form 4 tells you exactly who, how much, and at what price.
The Catalyst Surfer: a hair-trigger trailing stop
Three rules. No discretion, no chart-reading, no thesis. This is the strategy behind the headline number.
Buy the disclosure
A biotech insider files a Form 4 purchase → buy at that day's closing price. Every filing, no cherry-picking. Position size: 5% of the account.
Sell any 0.5% dip
A trailing stop follows the highest price since entry. The moment the stock slips 0.5% off that high, which for most stocks is immediately, you're out.
Four days, max
Whatever happens, sell after 4 trading days. If the filing preceded a catalyst, it has shown itself by then. If not, there's nothing to wait for.
What actually happens, trade by trade
- Most trades lose, on purpose. The median trade is −0.5%: the stock does nothing, dips, the stop fires. Win rate is just 35%.
- The winners are enormous. When a filing precedes a catalyst, the stock gaps up so hard the 0.5% trail rides it for +30%, +65%, even +296%, and the tail pays for every small loss many times over.
- Risk stays capped. Because no position is ever held through a slow bleed, the account's worst drawdown all year was −3.1%. This is a lottery-ticket harvester with a seatbelt.
2026 year-to-date, $100k account
Gap-aware fills, no leverage. On 2x margin the same year ends at $325,420 (+225.4%) with a −6.1% max drawdown. Spoiler before you fall in love: these are daily-bar numbers with filing-day-close entries. The 6pm section shows what happens to this strategy when the entries become real. (It isn't pretty. The 15-day hold is the one that survives.)
Anatomy of one real trade
June 2026: Joshua Boger, the founder of Vertex Pharmaceuticals, filed a $3M purchase of CervoMed (CRVO), a micro-cap neuro biotech where he sits on the board. Here's the whole trade, day by day.
CervoMed · CRVOFounder of Vertex Pharmaceuticals, on CervoMed's board. He wired $3,000,000 of personal cash into a stock trading near $2.48, and the filing hit EDGAR at 5:37pm, after the close, when nobody was watching.
- 1You copy the filing, buying at $2.48
Jun 15: Joshua Boger, founder of Vertex Pharmaceuticals and now a CervoMed director, files a Form 4 for a $3.0M purchase. The rule buys at that day's closing price. Your trailing stop arms 0.5% below.
- 2The catalyst hits and your stop rides up with the price
Jun 16: trial results land. The stock gaps up, touches $7.45 intraday, and closes at $4.52. The trailing stop follows the high: it now sits at $7.41, locked far above your entry.
- 3The give-back day: you're out at the open, +65%
Jun 17: the stock opens at $4.09, well below the trail. You can't sell at a level the market gapped past overnight. The fill you'd actually get is the opening price. Two days, +65% banked.
- 4What a naive backtest claims instead
A simulator that fills at the stop level books this exit at $7.41, claiming +199%, at a price that traded for moments the prior afternoon. That single assumption is most of the gap between the unreachable 196% account and the realistic 83% one.
What is buying within seconds actually worth?
Here's the thing about biotech Form 4s: 75% of them hit EDGAR at or after 4pm, while the market is closed and nobody's watching. Our feed picks them up about a minute later. So we measured the whole year at one-second resolution: same filing, same finish line, seven different moments to press buy.
The decay curve: 426 filings, 2026-01-01 → 2026-07-12, measured in seconds
| You buy at… | Fillable | Median return* | Mean return* | Median edge vs next open |
|---|---|---|---|---|
| filing +5s | 219/426 | +2.1% | +4.3% | +2.1% |
| filing +30s | 181/426 | +1.2% | +2.4% | +1.4% |
| filing +1m | 116/426 | +0.2% | +1.6% | +0.4% |
| filing +5m | 209/426 | +1.0% | +2.5% | +0.5% |
| filing +15m | 196/426 | +0.6% | +1.9% | +0.3% |
| filing +60m | 131/426 | +1.4% | +2.8% | +0.0% |
| next morning's open | 426/426 | -0.4% | +0.4% | – |
*Return from that entry to the close of the next full trading session, a common finish line, so only the entry moment differs. Prices are real prints from Polygon 1-second data, pre/after-market included. 90% of these filings landed outside market hours. "Fillable" = someone actually traded at that moment; only 219 of 426 filings printed within 5 seconds, and 42% traded zero dollars in the entire first minute. The median first-minute tape is about $3,323, so even a $5k order is often the whole market. Fast is an edge for small accounts only.
One strategy, four numbers — your fills decide which
The exact same signal list returns anything from +196% to +32% over these six months. The only thing that changes is how the orders fill: the top number needs a bot hitting the ideal price the instant a filing lands, and each step down swaps in a more realistic assumption for trading thin biotech names. Here are all four, labeled by what it takes to hit them.
Every stop-out fills at exactly the stop price, even when the stock gapped far below it overnight. This is how most published backtests (including our own early research) are computed.
If a stock opens below your stop, you get the open price, not the stop price. Losses become real; the fantasy exits at untouchable intraday levels disappear.
Every buy fills a little higher and every sell a little lower than the printed price. That's the cost of actually trading thin biotech names.
Drop every stock trading under $1, where bid-ask spreads alone can exceed 2% and fills at the printed price are wishful thinking.
What you could actually have banked
We reran the whole year with entries a real account could get: a bot filling within the hour of each filing going live (often in the after-hours session), a slower same-day buy, and a plain next-morning-open buy, using the exit that survives realistic execution, the 15-day hold.
| Signal filter | Bot: ≤ 1 hour | +3 hours | Next morning | Filing-day close(unbuyable, for reference) |
|---|---|---|---|---|
All biotech insider buys 440 signals | +40.7%★ +3.6%/trade | +39.9% +3.6%/trade | +24.6% +2.8%/trade | +48.4% +5.4%/trade |
Big money (value ≥ $100k) 288 signals | +28.9% +4.0%/trade | +36.8%★ +4.1%/trade | +29.1% +3.4%/trade | +54.8% +6.4%/trade |
Skin in the game (ownership +10–50%) 117 signals | +22.2% +3.8%/trade | +24.9%★ +4.9%/trade | +21.6% +4.5%/trade | +35.8% +6.6%/trade |
CEO buys 72 signals | +25.5%★ +6.9%/trade | +17.6% +4.9%/trade | +16.9% +4.8%/trade | +37.1% +9.5%/trade |
$100k / 20-slot account, entry per column, exit at the close of the 15th trading day, no stop. Big number = account return over the six months; small number = average return per trade (unconstrained by slot capacity; the cleaner apples-to-apples read, since 15-day holds keep slots busy and force the account to skip signals). ★ = best executable entry for that filter.
The trail didn't survive contact with reality
Run the Catalyst Surfer's 0.5% trail on hourly bars with executable entries and it collapses: +3.1% for the bot, -3.5% for the next-morning buyer. A stop that tight trips on every hourly wobble the daily-bar backtest never saw. The two-week hold is the exit that keeps the edge. That's why it headlines this section.
Pickier filters mostly don't help
We swept the obvious upgrades. CEO-only, CEO buys over $100k, and cluster buys all earn more per trade but fire so rarely the account sits idle; rarity is the tax on selectivity. The one filter that beat the field: skip anything under $1 (+58.1%, and stricter about fills, not looser). The risk-adjusted sleeper is C-suite buys on 2x margin: +72.9% with a smaller drawdown than the unlevered firehose.
The Two-Week Rider: the exit that survived
Same signals, opposite temperament: buy the filing, hold 15 trading days, sell. No stop at all. Less glamorous than the hair-trigger trail, but as the 6pm section showed, it's the one that keeps paying when the entries get realistic.
2026 year-to-date, $100k account
Which temperament are you?
The Catalyst Surfer loses 65% of the time in tiny paper cuts and pays you in rare windfalls. Psychologically brutal, financially calm: the account barely ever draws down.
The Two-Week Rider wins 58% of the time and feels much better trade to trade. But with no stop, you ride every failed catalyst down. Its worst drawdown was −17.7%, nearly six times deeper, for a smaller final balance.
Both beat the market handily in 2026. The difference isn't the signal. It's which failure mode you can live with.
The trades that made the year
Not hypotheticals, not survivor-picked screenshots. These are the ledger's five biggest winners plus one perfectly average stop-out, each linked to the actual filing sitting on our live feed.
ADTX
Unlock the full Biotech Insider Playbook
The trades that made the year, with links to every filing, plus the second strategy and the full execution-reality breakdown.
Free account · No credit card · Every number from real 2026 SEC Form 4 filings
Methodology & caveats
How the numbers were produced, and where they'd bend in the real world.
- Universe
- Every SEC Form 4 open-market purchase in biotech, pharma, diagnostics, and medical devices filed 2026-01-01 → 2026-07-12. Same-ticker same-day filings collapse into one signal: 440 signals total.
- Account model
- $100,000 starting capital, up to 20 concurrent positions of ~5% of equity each, profits compound into later trades. Signals arriving with all slots full are skipped, exactly as a real account would have to.
- Entry
- Filing-date closing price: the first price a copier could realistically get after the disclosure became public. No hindsight, no pre-announcement entries.
- Exits
- Daily OHLC bars, split- and dividend-adjusted. Stops and trails fill at the opening price when a stock gaps through the level overnight (the "gap-aware" rung). Fills at untouched intraday levels are counted only in the "optimistic" rung, for comparison with typical published backtests.
- Benchmark
- SPY total price return over the identical window: +10.5%.
- Timing studies
- Entry-timing ladder uses EDGAR acceptance timestamps (98% coverage) against 60-minute bars including pre/post sessions; the minute-resolution study uses 1-minute bars, which exist only for the trailing ~30 days. Extended-hours fills at bar opens are optimistic on thin names; treat fast-entry rungs as upper bounds.
- Strategy provenance
- Both exit strategies were discovered in our earlier research on 2023–2026 data and were not re-tuned on this window, but that history overlaps 2026, so treat results as partially in-sample, not out-of-sample proof.
These filings hit our feed about a minute after EDGAR
Every trade in this article appeared on the live feed the day it was filed, most of them while the market was closed and nobody was looking. Filter the feed to Healthcare, sort by filing date, set an alert, and catch the next one yourself.