Rival stocks move like twins, until one runs ahead.
Boostio finds rival companies across every industry, shows which stock is ahead of its usual relationship with its rival, and runs a shared paper strategy that buys the laggard and shorts the leader, only when the gap has a measured habit of closing.
Three kinds of leadership, one honest gate
Rivals
Every week Boostio screens every US-listed stock (6,986 at the last screen) and pairs the largest companies in each industry, confirmed by their SEC industry code. Rivalry is a hypothesis, not proof.
Who leads
Business leadership is revenue reported to the SEC. Stock leadership is which share is ahead of the pair's own last year, in standard deviations. They are often different companies.
Does the gap close?
A measured twin needs daily returns that move together (correlation 0.60+), a gap that passes a mean-reversion test at 5% and a half-life of 63 sessions or less.
A paired paper idea
Only then does a 2 standard deviation gap become an idea: long the laggard, short the leader, both legs together. Exit when the gap closes, stop at 4. No time limit.
Why there is no paper trade today
No paper trade qualifies on the Oct 22, 2026 close
- 17 measured twins today, none with a 2 standard deviation gap that is clear of a one-day company-specific move.
- 39 rival pairs have one stock 2+ standard deviations ahead without a measured relationship. The reel's rule would trade them; Stock Trade Twins does not, because a gap with no habit of closing is not evidence that it will close.
| Wide gap, no measured twin | Ahead now | Why there is no idea |
|---|---|---|
| JPM / BAC | JPM by 3.98 sd | Mean reversion: ADF t-statistic of the spread: -0.42 (needs <= -2.873 (5% critical value)) |
| MRSH / AON | MRSH by 3.01 sd | Mean reversion: ADF t-statistic of the spread: 0.57 (needs <= -2.873 (5% critical value)) |
| ECHO / ASTS | ECHO by 2.97 sd | Co-movement: daily return correlation, 252 sessions: 0.42 (needs >= 0.60) |
| DOW / ALB | DOW by 2.88 sd | Co-movement: daily return correlation, 252 sessions: 0.25 (needs >= 0.60) |
| TMUS / CHT | CHT by 2.74 sd | Co-movement: daily return correlation, 252 sessions: 0.16 (needs >= 0.60) |
Every rival pair, and who is ahead
380 pairs
Paired positions, entered and closed together
| Pair | Legs | Entered | Gap at entry | Gap now | Net if closed now | Status |
|---|---|---|---|---|---|---|
| DUK / EXC | Long EXC · Short DUK | Oct 20, 2026 | 2.01 sd | 1.69 sd | $27.70 | open |
| PKG / SW | Long SW · Short PKG | Oct 20, 2026 | 2.04 sd | 2.34 sd | -$271.38 | open |
| Closed pair | Legs | Entered | Exited | Why | Net (after costs) |
|---|---|---|---|---|---|
| BAC / JPM | Long BAC · Short JPM | Oct 14, 2026 | Oct 15, 2026 | divergence stop | -$0.58 |
Three kinds of evidence, never added together
What the reel says
- Buying Pepsi whenever it fell behind Coke and shorting Coke whenever it ran ahead, since 1972: "up over 46,000% on paper".
- The opposite (buy the one that ran ahead, short the one that fell behind): "down over 99%".
- "A research paper" found the effect "across thousands of stock pairs across the world".
From an Instagram reel by Brando Le (@eliteoptionstrader2), Oct 12, 2026. Not verified: the reel names no paper and gives no rules, costs or borrow assumptions.
+1.11% mean per trade, after costs
- 260 simulated trades over 372 pairs, to Oct 20, 2026; 52% won.
- 95% interval -0.07% to 2.29%: it includes zero. No edge is shown.
- Costs and borrow took $15,937 of $44,644 gross at $10,000 per leg.
- Chasing the leader instead (the reel's "opposite"): -2.34% per trade.
Same rules as the forward book, each pair replayed on its own, fills at the next open. The pairs are today's large companies (survivorship bias) and the data covers about ten years. It can rule an idea out; it cannot prove one.
-$0.58 net, 1 closed
- First filled Oct 14, 2026; 2 open, 0 waiting for the open.
- Read rule, fixed in advance: nothing is concluded before 30 closed trades and 90 days, and an edge counts only if the whole 95% interval is above zero.
This is the only tier that tests the rules on data they have not seen. Simulated, never real money.
Rivals that do not trade like stocks (yet)
Some rivals are private or newly listed. Their listing status is checked every week against SEC's ticker list, Nasdaq's symbol lookup and EDGAR. Tickers are never invented, and a fund or ETF that carries a company's name is never used as its price.
Anthropic
No listed sharesFrontier AI lab (Claude models), sold through an API and subscriptions. Checked Oct 21, 2026.
- SEC ticker list: no ticker for the company.
- Nasdaq lookup: no common stock.
- EDGAR: 79 names match; the top results are third-party funds and SPVs, not the company.
- Carries the name but is not its stock: ANTW (ETFs).
No public stock, so no price, no relationship and no paper idea. Business size is not measurable from public filings.
OpenAI
No listed sharesFrontier AI lab (GPT models, ChatGPT), sold through an API and subscriptions. Checked Oct 21, 2026.
- SEC ticker list: no ticker for the company.
- Nasdaq lookup: no common stock.
- EDGAR: 43 names match; the top results are third-party funds and SPVs, not the company.
- Carries the name but is not its stock: OAIW (ETFs).
No public stock, so no price, no relationship and no paper idea. Business size is not measurable from public filings.
SpaceX
Listed: SPCXSpace Exploration Technologies: launch vehicles and satellite internet (Starlink). Checked Oct 21, 2026.
- SEC ticker list: SPCX (Nasdaq).
- Nasdaq lookup: SPCX.
- EDGAR: 153 names match; the top results are third-party funds and SPVs, not the company.
- Carries the name but is not its stock: XAIW, LOFD, LOFF, SNK and 5 more (ETFs).
Trading on Nasdaq since Jun 12, 2026: 92 sessions, so a one-year relationship cannot be measured yet. Market value $2.01T.
No twin assigned for SpaceX: classification places it beside Alphabet and Meta (SEC SIC 7370 / Nasdaq "Computer Software: Programming Data Processing"), which is weak business evidence, and no listed rival is of comparable scale.
Since its listing: SPCX -10.1%; for context only, not twins: RKLB -50.0%, ASTS -53.0%.
Anthropic and OpenAI: research only. Anthropic has no listed shares (verified 2026-10-21); OpenAI has no listed shares (verified 2026-10-21). No price relationship can be measured, so no paper signal.
Rules fixed before the first run
The gate
Spread = the log price of one stock minus the other, compared with the 252 trading days before each day. Measured twin: return correlation 0.60+, a mean-reversion test at 5% (augmented Dickey-Fuller) and a half-life up to 63 sessions. Idea at 2 sd unless a one-day move of 4+ sd (news, earnings, a corporate action) happened in the last 10 sessions.
Costs in every number
Per leg and side: 5 to 20 basis points of slippage by liquidity plus 1 basis point of fees. Short legs pay 0.50% a year to borrow (a general-collateral assumption; hard-to-borrow fees and recalls are not modeled) and owe the dividends. Exact ex-dividend dates are approximated from SEC-reported dividends.
Data, attributed
Daily prices: Nasdaq.com (split-adjusted). Filings, revenue and industry codes: SEC EDGAR. Earnings dates: Nasdaq earnings calendar, shown as a risk note. A day counts only after the close is final; bad rows are rejected (15 so far); stale prices block ideas.
The reel does not cite a paper; the rules follow the documented distance approach of the academic pairs literature (Gatev, Goetzmann and Rouwenhorst, 2006) plus a stationarity test, and later studies found simple pairs profits fading after the early 2000s (Do and Faff, 2010). Rules are versioned (STT-v1) and never tuned on the results.
Stock Trade Twins, answered
What are stock trade twins?
Two rival companies whose stocks usually move together, such as Coca-Cola and PepsiCo or Home Depot and Lowe's. Boostio treats rivalry only as a starting hypothesis: a pair counts as a measured twin when the price gap between the two has actually tended to close.
Does Boostio trade these pairs with real money?
No. The strategy runs a shared paper book: simulated positions with modeled trading costs, short borrow fees and dividends. There is no broker, no order and no real money, and nothing on this page is investment advice.
Is the 46,000% Pepsi result from the reel real?
It is the creator's claim and Boostio has not verified it. The reel names no paper and gives no rules, costs or borrow assumptions, so Boostio shows it as a source claim and never as a result.
Where does the data come from?
Daily stock prices from Nasdaq.com, company filings and industry codes from SEC EDGAR, and earnings dates from the Nasdaq earnings calendar. Missing or stale prices are shown as a data problem and never create a signal.
Does the twin strategy work?
Not proven. The historical simulation is a diagnostic whose 95% interval includes zero, and only the forward paper record can show whether the rules work. Boostio does not guarantee any result.
Related research: Stock Besties maps the business ties behind AI stocks, who funds, buys from and supplies whom, with a source for every link. A rival pair and a business link are different questions.
Follow the research behind the paper book
Boostio is a research-first platform in private beta: crypto candidates, daily market stories, public paper trading and now rival stocks. Exploring this page stays free and open.
Explore the private beta