Can You Beat the Market by Following Pelosi's Trades? I Backtested 34 Public Signals
Turning the Pelosi family's disclosed purchases from 2020 through 2025 into a tradeable rule produced higher historical average returns than both the S&P 500 and Nasdaq-100. But weak years, clustered observations, and dependence on Nvidia keep it far from a proven trading system.
Data as of September 14, 2026.
The Pelosi trading record creates an obvious temptation. Nancy Pelosi and her husband, Paul Pelosi, are known for several well-timed technology trades. Why not wait for the disclosures to become public and simply follow them?
I wanted to test a more practical question than “How much did the Pelosis make?”: Can an ordinary investor still earn excess returns after the disclosure becomes public?
I translated that question into a simple rule. I collected Nancy Pelosi's periodic transaction reports from the U.S. House of Representatives. When a report disclosed a new stock purchase or call-option purchase, I bought the underlying stock at the next trading day's open, held it for 3, 6, or 12 months, and compared the result with SPY, QQQ, and ONEQ over the same dates.
The result is tempting. It is also a good reminder not to declare victory too early.
The backtest: higher averages, limited consistency
I reviewed 39 Pelosi periodic transaction reports in the House disclosure indexes for 2020 through 2025 and checked 130 transaction rows. After excluding sales, option exercises, gifts, exchanges, private investments, and duplicate amendments, 44 new purchases collapsed into 34 unique “filing date + ticker” signals across 16 filing dates.
I counted the same ticker only once when it appeared more than once on the same filing date. Because the disclosures provide value ranges rather than exact position sizes, I used an equal-weighted average instead of amount weighting. Each hypothetical entry and exit includes a 0.10% transaction cost. Returns are pre-tax and include dividends.
The first chart shows the headline result. The bars are the average net returns of the follow signals; the two lines are SPY and QQQ, in that order. Returns rose for all three as the holding period increased, but the gap between the disclosed trades and the benchmarks widened as well.
| Holding period | Follow signals: average net return | SPY | QQQ | ONEQ | Excess vs. QQQ | Signals that beat QQQ |
|---|---|---|---|---|---|---|
| 3 months | 4.27% | 2.20% | 1.14% | 0.74% | +3.13 pp | 52.9% |
| 6 months | 10.66% | 6.65% | 6.07% | 5.55% | +4.59 pp | 50.0% |
| 12 months | 20.76% | 12.50% | 10.39% | 9.65% | +10.37 pp | 58.8% |
At first glance, the conclusion looks straightforward: the average signal beat every benchmark at all three horizons, with the largest advantage after 12 months.
But these are arithmetic averages across 34 historical signals, not the annualized return of a portfolio funded with a fixed amount of capital. The signals include repeated tickers and overlapping holding periods. More importantly, 34 signals came from only 16 filing dates. They should not be treated as 34 independent experiments.
The gap between average excess return and win rate is another warning. At six months, exactly half of the signals beat QQQ. A small number of large winners therefore contributed a meaningful share of the average return.
The year-by-year record is much less smooth
If the method were persistent, I would want it to work with reasonable consistency across different market environments. The yearly breakdown shows the opposite: large swings and very uneven sample sizes.
The number of signals in those years was 6, 15, 3, 1, 4, and 5, respectively. That context belongs next to the chart: the enormous 2023 bar represents one signal, while the negative 2021 result represents 15.
The 15 signals filed in 2021 accounted for 44% of the full sample. Their average one-year net return was -19.53%, trailing QQQ by 6.39 percentage points. One particularly painful example was the Roblox bullish signal filed in December 2021. Buying the stock after disclosure and holding it for a year would have produced a net loss of roughly 70.54%. Copying the option itself could have been even more damaging.
At the other extreme, 2023 contained a single Nvidia signal. The underlying stock returned about 183.42% over the following year, making a huge contribution to the overall average. One observation cannot establish that the strategy worked broadly in 2023, much less that it will repeat.
This is also why I am skeptical of annual “best congressional trader” rankings. Those rankings often measure returns from the lawmaker's original transaction date. A follower can act only after the disclosure. The two calculations answer different questions. A recent NBER working paper, covering trades by members of Congress and their immediate families from 2012 through 2023, also found no persistent market outperformance on average. The reported trading behavior looked more consistent with reacting to public signals and sentiment.
Disclosure delay is only one source of uncertainty
House rules generally require covered securities transactions above $1,000 to be reported within 30 days of learning about the transaction and no later than 45 days after it occurred. The House financial disclosure guidance and annual indexes provide a FilingDate, but they do not give me a complete, verifiable historical timestamp for when every filing first became visible to the public.
Using the next trading day after FilingDate can therefore introduce an information-availability bias. To test how much that assumption mattered, I delayed the entry by an additional 3, 7, and 30 calendar days, then bought at the next market open.
The average edge did not disappear in this sample. With those extra 3-, 7-, and 30-day delays, the 12-month excess returns over QQQ were 10.76, 10.09, and 10.13 percentage points, close to 10.37 points in the baseline rule. This is a sensitivity check, however. It does not prove that I reconstructed the earliest public availability of every historical filing.
Dependence on a single stock mattered more. Six of the 34 signals involved Nvidia. After removing all six, average excess return versus QQQ fell to -0.77 percentage points at three months and -1.25 points at six months. The 12-month result remained positive, but dropped to 5.12 points.
The chart below compares excess returns over QQQ. The first line uses all 34 signals; the second uses the 28 non-Nvidia signals. The short- and medium-term edge turns negative without Nvidia, while the 12-month advantage is roughly halved.
The evidence therefore supports a narrower statement than “buy whenever Pelosi buys”: some disclosed long-term bullish signals may deserve further research, but the short- and medium-term advantage in this sample is highly sensitive to a few powerful technology winners.
Why I did not copy the options
Many of the Pelosi family's best-known trades used long-dated call options. To compare securities and years on a common basis, I treated a call purchase as a bullish signal and backtested the underlying stock.
That choice does not reproduce the family's actual option returns. It avoids larger problems, though. The disclosures provide value bands rather than exact position sizes. Complete historical option quotes, bid-ask spreads, and execution prices are difficult to reconstruct. By the time a filing becomes public, an option with the same strike and expiry can also have a very different implied volatility and risk-reward profile.
It is equally important to distinguish a new option purchase from the exercise of an old one. For example, a July 2025 Broadcom filing lists the transaction as a stock purchase, but the description says it resulted from exercising calls bought in 2024. That transaction realizes an earlier view rather than creating a new bullish signal, so I excluded exercises like this one.
I calculated returns with prices adjusted for splits and dividends. Yahoo's explanation of adjusted prices confirms that Adjusted Close incorporates applicable splits and dividend distributions. SPY serves as the investable proxy for the S&P 500; QQQ tracks the Nasdaq-100; and ONEQ represents the Nasdaq Composite. Each ETF has fees and tracking differences, so the comparison is against tradeable products rather than frictionless theoretical indexes.
How I would use congressional disclosures
This test did not give me a rule to place an order whenever a filing alert arrives. It made the disclosures more useful as the top of a research funnel.
If I continue tracking them, I will use four constraints:
- Count only new stock purchases and new call-option purchases; exclude exercises, gifts, exchanges, and filing amendments.
- Do not buy because the name says Pelosi. Reassess the company's fundamentals, valuation, and remaining catalysts at the time of disclosure.
- Focus on the 6- to 12-month thesis and avoid treating delayed disclosures as short-term signals.
- Measure exposure to Nvidia and other popular technology stocks separately, and compare results with QQQ so that sector beta does not masquerade as stock-picking skill.
I would gain confidence only if new, out-of-sample signals continued to beat QQQ over 12 months and the advantage did not come from one or two stocks. I would abandon the “copyable excess return” hypothesis if the edge disappeared, could be explained by technology-factor exposure, or was overturned by better first-publication timestamps.
My conclusion
The backtest produced a more interesting and restrained answer than the social-media version of the story. Buying the stocks behind the Pelosi family's disclosed bullish trades after filing produced higher average returns than SPY and QQQ across these 34 signals from 2020 through 2025. The 12-month horizon was especially strong.
But a higher historical average is a long way from a persistent, tradeable edge. The 2021 failure, the enormous contribution from one 2023 Nvidia signal, incomplete publication timestamps, and overlapping positions all prevent me from treating this as a validated stand-alone system.
I will continue to use congressional disclosures as candidate-generation signals, not as reasons to buy. Any follow strategy still has to answer the same question: after I can actually see the information, pay the trading costs, and accept the risk, is the remaining opportunity worth taking?
This article records a personal research framework and conditional judgments. It does not constitute investment advice, a return guarantee, or a basis for buying or selling securities. The historical backtest, return table, and trend charts are the author's research tools; they are neither objective ratings nor forecasts of future results. Markets and company circumstances may change, and investors should independently verify the information and assume their own risk.