Trading Platform Research

Home / Research examples / Politicians who skipped NVDA and beat the market

Which politicians who didn't buy NVDA outperformed the market last year?

NexusTrade answered from its own congressional trade data in under three minutes and named 10 members. QuantConnect's congressional dataset needs a separate licence on the Free tier, so its assistant answered from public web trackers and said it could not confirm the NVDA filter. Composer's assistant said it cannot answer questions about politicians' trades.

Documented

The company says so in its own docs. The source is linked and dated.

Observed

I ran it and this is what happened. Inputs and steps are shown.

Opinion

My judgment. You may weigh the same facts differently.

The test

Question
Which politicians who didn't buy NVDA outperformed the market last year?
Inputs
The sentence above and nothing else, typed on November 16, 2026. NexusTrade read last year as calendar 2025. QuantConnect read it as the trailing 12 months.
Method
  1. NexusTrade's stock screener took every purchase disclosed in 2025, set aside any member with an NVDA purchase (stock or option), and kept members with at least five priced purchases.
  2. It bought each stock at the first close after the disclosure became public, held to the last 2025 close, weighted each trade by the midpoint of its disclosed dollar range, and compared it with SPY over the same dates.
  3. QuantConnect's assistant tried its US Congress Trading dataset in a research notebook, was told the dataset is not licensed on the account, and then answered from public disclosure trackers on the web.
  4. Composer's assistant declined the question.
Result
Observed
NexusTradeQuantConnectComposer
Answered without writing codeYesYes, the assistant wrote the PythonYes
Time from question to answer2 min 25 sMessage sent 03:13:43 UTC; the answer finished at 03:16:08 UTC.3 min 40 sMessage sent 23:38:42 UTC; QuantConnect's status changed from Running to Completed between 23:42:14 and 23:42:24 UTC.16 sMessage sent 02:14:52 UTC; the answer stopped growing at 02:15:08 UTC.
Where the trades came fromIts own congressional trade dataPublic web trackers; its own dataset was not licensedNone
NVDA filter appliedYes, on every disclosed NVDA purchaseNot confirmed, in its own wordsNot answered
Members named1090
Limits
Disclosures give dollar ranges and arrive up to 45 days after a trade, so every return here is an estimate built from purchases alone. The two platforms used different windows (calendar 2025 and the trailing 12 months), so the lists are not directly comparable. QuantConnect's list came from the web, and with its congressional dataset licensed it would run the trade-by-trade version it had staged.

The results

Observed

NexusTrade: what it returned

Members with no 2025 NVDA purchase who beat SPY

#MemberPurchases countedReturn on disclosed buysSPY over the same datesAhead of SPY
1Jerry Moran6+60.21%+11.88%+48.33 pts
2Shelley Moore Capito15+36.08%+12.23%+23.85 pts
3Kelly Morrison10+31.17%+21.14%+10.03 pts
4Mitch McConnell5+19.06%+10.42%+8.64 pts
5Dan Newhouse15+10.06%+5.42%+4.64 pts
6Julie Johnson79+17.52%+14.59%+2.93 pts
7Markwayne Mullin74+13.41%+10.58%+2.83 pts
8David J. Taylor45+11.10%+9.96%+1.14 pts
9Greg Landsman5+18.63%+17.87%+0.75 pts
10Thomas H. Kean, Jr.34+12.52%+12.19%+0.33 pts

Returns are modelled from disclosed purchases only. They are estimates of how those purchases did, and they are separate from a member's full portfolio.

NexusTrade's chat showing the question, its written answer and a bar chart of the ten members
NexusTrade's answer as it appeared in the chat.

QuantConnect: what it returned

QuantConnect said its congressional dataset is not licensed on this account

"The US Congress Trading dataset (Quiver) exists on QuantConnect and would answer this exactly, but it isn't licensed on this account."

"For the nine names listed, I cannot confirm their NVDA purchase status with certainty, because the on-platform dataset that would verify trade-by-trade detail isn't licensed here."

Quoted from QuantConnect's assistant.

The list QuantConnect's assistant gave from public trackers

PoliticianChamber and party1-year return on disclosed buysAgainst the S&P 500
Tim MooreHouse R-NC+41.4%+29 pts
John FettermanSenate D-PA+38.5%+26 pts
Robert LattaHouse R-OH+27.6%+15 pts
Josh GottheimerHouse D-NJ+22.0%+10 pts
Ed CaseHouse D-HI+20.9%+9 pts
Richard McCormickHouse R-GA+18.0%+6 pts
Alan ArmstrongSenate R-OK+17.5%+5 pts
Jerry MoranSenate R-KS+13.5%+1 pt
Gary PetersSenate D-MI+13.4%+1 pt

Trailing 12 months to mid-November 2026, as reported by the assistant from web sources.

QuantConnect's assistant answering the politicians question
QuantConnect's answer to the politicians question.

Composer: what it returned

Composer said it cannot answer questions about politicians' trades

"I'm unable to provide information on specific politicians or their investment performance, especially regarding individual stock transactions like NVDA. My capabilities are focused on generating Composer symphonies for automated trading strategies."

Quoted from Composer's assistant.

Composer's assistant declining the politicians question
Composer's answer to the politicians question.

Comparisons that use this test