Congress goes to the Democrats, stocks rise — who wins, who loses and what it means for Romania

Wall Street did not celebrate a party's victory on 4 November, but the fact that no party can now make laws alone. For Romania, the important event was not the vote but the talks at Cotroceni Palace with the Pentagon's number two.

The analysis answers three questions: what came out of the ballot boxes and how firm the figures are, how the market reacted by party and who funds whom, and what changes for Romania. We measured the reaction on our own data and checked the predictions we published in August about these elections.

The testable thesis: firms react to elections according to the party they fund, but the broad index has no preferred party. And the new Congress can make the president's decisions harder without being able to decide for him on tariffs or troops.

What voters decided and how firm the figures are

The results are not certified. NBC News projections from the evening of 4 November, US time, give Democrats 54 Senate seats and Republicans 44, and a majority takes 51. Democrats took seats from Republicans in Kansas, Iowa, Texas, Ohio, Maine, North Carolina and South Carolina.

Two races are open. In Mississippi, the gap on Wednesday morning was 43 votes, and in Alaska the count may take weeks. In the House of Representatives, at 18:28 Washington time, Democrats had 242 seats called, Republicans 151, and 42 remained uncalled. The NBC model estimates a final result of about 259 to 176, but that is an estimate. Only the direction is certain: both chambers go to the Democrats.

Our scorecard from 24 August shows where we went wrong. Of the eight predictions the projections can already judge, six had the right direction. We gave Josh Turek only 48% in Iowa and believed, at 52%, that Republicans would hold the Senate. We underestimated the wave, not the trend.

The prediction of 24 August	Probability	Provisional result
Democrats win the House	70%	Yes
Republicans hold the Senate	52%	No
Democrats gain at least two net Senate seats	65%	Yes
James Talarico wins in Texas	52%	Yes
Josh Turek wins in Iowa	48%	Yes
A Democrat wins a Senate seat in a state Trump carried by at least 10 points in 2024	70%	Yes
Democrats reach at least 230 House seats	60%	Yes
Divided Congress: Democratic House, Republican Senate	37%	No

The provisional Brier score on these eight rows is 0.17, where 0 is perfect and a constant 50% prediction gives 0.25. We did better than a coin, but underestimated the scale. The official scoring comes after certification, on the August deadlines.

How the market reacted on 4 November

The night began badly for shares, with futures falling as the first results came in. At the close, according to CNBC data, the Dow Jones gained 3.03%, its best day since April 2025, the S&P 500 rose 1.58% and the Nasdaq Composite 1.44%.

The explanation from strategists quoted by CNBC has two parts. The first is gridlock: markets prefer that no party has too much power. The second is tone: the president said he accepts the results and will work with Democrats, so the fear of a prolonged dispute disappeared. Another force of the day was oil. Its price fell more than 8% after Vice President JD Vance blamed the defeat on the Iran war, and the market concluded that pressure to end it will grow.

Rose	4 November	Fell	4 November
Microsoft	+6.68%	Trump Media (DJT)	−24.54%
Amazon	+6.12%	Dell	−8.82%
Apple	+5.31%	Marathon Petroleum	−8.11%
First Solar	+4.63%	Micron	−7.00%
General Motors	+4.63%	Lockheed Martin	−6.88%
JPMorgan	+4.18%	Exxon Mobil	−4.62%
Enphase	+3.71%	Northrop Grumman	−4.53%

By sector, against the average of all firms, energy lost 5.78 percentage points, while consumer and staples gained 2.49 and 2.24. The simple average of American firms in our data rose 1.40%.

Democrats or Republicans: what the market prefers

The short answer: we cannot say it prefers either party. We counted the S&P 500's reaction after 49 elections between 1928 and 2024. For the presidency we recorded the winner's party, and for midterms the party that gained House seats. None of the differences is statistically firm.

The measure	Republicans win	Democrats win	Difference	t statistic
Presidency, next day	+0.20%	−1.26%	+1.46 pp	1.87
Presidency, after 60 sessions	+3.44%	+0.59%	+2.85 pp	0.89
Congress, next day	+0.06%	+0.57%	−0.51 pp	−0.69
Congress, after 60 sessions	+4.98%	+6.32%	−1.34 pp	−0.50

For the presidency, the sign goes in the direction found by Snowberg, Wolfers and Zitzewitz, who used betting markets: shares rise when a Republican's chances grow. Over the long run, Santa-Clara and Valkanov found higher returns under Democratic presidents. At midterms, a Democratic gain was not bad news: the index rose the next day in 9 of 12 cases, against 8 of 12 for Republican gains.

Who funds whom

We read the Federal Election Commission data on corporate political action committees, known as PACs, up to 1 September of the election year. We counted only money given directly to candidates and kept firms with at least 50,000 dollars. For the 2026 cycle there are 291 listed firms, totalling 86.6 million dollars, 57% of corporate PAC money.

Sector	Firms	Republican share
Energy	16	88%
Consumer staples	14	70%
Materials	16	69%
Utilities	25	65%
Technology	60	63%
Industrials	47	62%
Health	26	57%
Consumer	15	57%
Financials	61	56%
Telecom	9	54%

All the large sectors give more to Republicans, who ran Congress when this money was raised. Intensity makes the difference. Among firms with at least 200,000 dollars, the most tilted to Republicans are HF Sinclair, Marathon Petroleum, SpaceX, Chevron and Williams. Those tilted to Democrats are few and less decisive: TransDigm and Truist. Large military contractors such as Lockheed Martin and General Dynamics give to both parties, with a Republican surplus. This money is raised from employees, not from the company's treasury, and does not include lobbying, super PACs or individual donations.

Firms follow the party they fund

Here we found a measurable effect. For each election since 2016 we compared the next-day return of firms in the same sector with their political tilt. The figure in the table shows how many percentage points a firm giving everything to Republicans gains against a balanced one. The rules, including the threshold, were sealed before any calculation.

Election	Winner	Effect	Firms	In the expected direction
2016	Republicans	+2.93 pp	201	Yes
2018	Democrats	−1.46 pp	211	Yes
2020	Democrats	−2.47 pp	225	Yes
2022	Republicans	−0.23 pp	221	No
2024	Republicans	+2.71 pp	264	Yes
2026, separate test	Democrats	−3.86 pp	283	Yes

Across the five earlier elections, the average effect in the winner's direction is 1.87 percentage points, with a t statistic of 3.21 and the right sign in four of five cases. It passed the threshold set in advance. The result fits Akey's paper, which found about 3% more for firms that donated to the winners of close congressional races. The year 2026, held apart, confirmed the sign: −3.86 points, t −3.61.

Three caveats matter. First: over two days the average effect stays positive but is no longer firm (t 0.92), and if we replace the winner with the surprise against expectations it falls to t 0.67. The sign of the surprise is our own judgement.

Second: 2026 has an extra factor, oil. We checked it after seeing the result, so the test is not sealed: without energy firms the 2026 effect stays at −3.87 points, and without energy and military contractors at −4.28. On sets of five ordinary days, the same average has a standard deviation of only 0.27 points, against the 1.86 at elections.

Third: political tilt may hide exposure to tariffs or regulation, not the party itself.

The effect cannot be traded after the result. Almost all the move is in the overnight jump, so it is already in the opening price. The data describe how the market values political ties; they do not offer a buying rule.

What the new Congress can and cannot do

The Democratic majority cannot override a presidential veto, which takes two thirds in both chambers. Experts at the Council on Foreign Relations draw three conclusions from this.

• Tariffs stay with the president, on powers already delegated. Congress can ask for selective cuts but cannot impose them alone.
• Money and inquiries are the new power: appropriations start in the House, and committees can call hearings, demand documents and subpoena officials.
• The Iran war will enter hearings, and Democrats will try to end it by law, again needing Republican votes.

The new Congress starts on 3 January 2027. Until then the old one remains, with its Republican majority.

What changes for Romania

The money: a good day, but a small one

The reaction in Bucharest was modest. According to ZF, the BET index closed at 34,984 points, 0.5% above the previous day, and Banca Transilvania and BRD each gained 1.6%. Nuclearelectrica, at +2.7%, rose after the announcement on restarting reactor 1, so not because of the vote. Nor does the 15% fall in Rompetrol Rafinare, its fifth in a row, have to do with the elections.

The dollar fell at the NBR rate from 4.7761 to 4.7518 lei, or 0.5%, while the euro rose slightly, from 5.3416 to 5.3443 lei. The dollar's move comes from its weakening worldwide, not from a change linked to Romania: against the euro, at the ECB rate, it lost about 0.8%. The reference yield published by the NBR for 10-year government bonds, on the offer side, fell from 6.97% to 6.89%. A single day does not show a trend.

The channel with a real effect for households is oil. If the fall holds, it reaches fuel and, with a delay, inflation. If the Iran war continues, it can return just as fast.

The troops: the decision lies elsewhere

On the same day, President Nicușor Dan received Elbridge Colby, the Pentagon's number two, at Cotroceni. The president publicly confirmed the talks on Black Sea security and Romania's support for American efforts in Iran. Sources quoted by G4Media, officially unconfirmed, say Romania offered to pay for American troops and equipment to stay, that the American official gave no clear answer, and that Donald Trump is due to decide in early December.

The vote changes less than it seems. Opposition to the withdrawal from Romania already existed and was Republican: in October 2025, Roger Wicker and Mike Rogers, the Republican chairs of the armed services committees, criticised the decision not to keep the rotational brigade. The law already sets a minimum of 76,000 troops in Europe, but does not say how many must be in Romania.

The counter-hypothesis deserves weight. Politico, quoting five NATO diplomats, writes that allies fear the opposite effect: blocked at home, the president could press harder on the alliance. The force posture review is expected by December, before the new Congress takes office. So the decision that matters for Romania may be taken before the new majority has any instrument at all.

Nuclear energy: a long road, little risk from the vote

The Doicești project depends on American money. The final investment decision was approved in February, and the American agency DFC sent a letter of intent for up to 2 billion dollars, non-binding. We found no position of the new Congress on the project. The main risk remains cost and technology, not the party running the committees.

Politics at home

The vote has already entered the Romanian debate. Traian Băsescu wrote that American voters rejected the MAGA movement. Such readings say more about the camps at home than about Romania. The American elections turned on American issues, and their effect on Romanian voting cannot be measured now.

The causal chain

Links 1, 2 and 7 are documented from sources. Link 4 is a connection in time, explained by analysts, not a measured cause. Links 5 and 6 are measured, but do not separate the vote from the day's other news. Link 3 is the markets' explanation, not a certainty. Link 8 is our conclusion.

Scenarios for the market: what history says

We started from 7,568.92 points, the election-day close, and counted what the index did 60 sessions after each of the 24 midterms. It was up 19 times. The scenarios below are these frequencies, so they take no account of the Iran war, today's interest rates or the jump of 4 November.

Scenario	S&P 500 level on 1 February 2027	Cases of 24	Probability
A: fall from election day	below 7,569	5	20%
B: rise under 10%	between 7,569 and 8,326	11	47%
C: rise of at least 10%	above 8,326	8	33%

Prediction table

Prediction	Horizon	Probability	How it is checked	Verification source
Scenario A: the S&P 500 closes below 7,569 points	1.02.2027	15–25%	The index's official close	spglobal.com
Scenario B: the S&P 500 closes between 7,569 and 8,326 points	1.02.2027	42–52%	The same close	spglobal.com
Scenario C: the S&P 500 closes above 8,326 points	1.02.2027	28–38%	The same close	spglobal.com
The US officially announces a further cut in troops stationed in Romania	by 31.01.2027	30–40%	Statement from the Pentagon or the Romanian Defence Ministry	war.gov, mapn.ro
Romania and the US officially announce a Romanian contribution to the cost of the American presence	by 30.06.2027	10–20%	Official agreement or statement	presidency.ro, war.gov
A limit on the president's power to impose tariffs becomes law	by 30.06.2027	0–10%	Law published on congress.gov	congress.gov
A congressional committee holds a public hearing on the Iran war	by 28.02.2027	85–95%	The armed services and foreign relations committee calendars	congress.gov
The NBR euro rate stays below 5.40 lei	31.12.2026	85–95%	The reference rate of that day	bnr.ro
The NBR dollar rate stays between 4.51 and 4.99 lei	31.12.2026	80–90%	The reference rate of that day	bnr.ro
The BET index closes above 34,984 points	31.12.2026	55–65%	The close of the year's last trading day	bvb.ro
DFC or EXIM announces binding financing for construction at Doicești	by 30.06.2027	5–15%	Statement from the institution or RoPower	dfc.gov, exim.gov

When we consider the analysis wrong

• "Firms follow the party" falls if, at the next federal elections, the effect measured the same way has the opposite sign and at least two t units.
• "The market has no preferred party" falls if an extension with new elections takes any of the differences in the table above two t units.
• "The vote changes little for troops" falls if the new Congress passes, by 30 September 2027, a law explicitly naming the presence in Romania or on the eastern flank and the administration complies.
• "The day is a small move for Romania" falls if the 10-year reference yield stays more than 0.5 points below its 3 November level until 31 January 2027 with no domestic explanation.

Method, definitions and figures

The data. The S&P 500 comes from a daily series since 1927, without dividends. Shares are daily closes from the olivLaw database, available since October 2016. The "next day" reaction runs from the last close up to election day to the first close after it.

Political tilt is (R − D) / (R + D), where R and D are the firm PAC's money to Republican and Democratic candidates. We used only transactions before 1 September of the election year, so that nothing from after the vote enters. The effect is the slope of a regression across firms, within the sector, with a price of at least 5 dollars and moves of at most 30%. We left out sectors with fewer than five firms.

The preregistration. The questions, the threshold (t of at least 2.776 and at least four of five elections), the election table and the separate reading for 2026 were sealed on 4 November, before we calculated any return. The zone is not clean: we knew the literature, the reactions of 2016–2024 and the press reports of the morning of 4 November. So the maximum verdict is "correlated". The checks without energy and military contractors and the comparison with ordinary days came afterwards and are exploratory.

The probabilities. Scenarios A, B and C are counted frequencies. The exchange rates are also frequencies, on 40-session moves over the last five years: the euro stayed below an equivalent threshold in 91% of cases and the dollar within the ±5% band in 88%. The others are editorial judgements, in intervals of at least ten percentage points, and will be publicly scored at their deadlines.

Limits of the analysis

The election results are uncertified projections, and the House figures change as races are called. We found no official Brent close for 4 November, so we give only the reported fall. Romania's payment offer and the date of the American decision come from anonymous sources.

Name matching between firms and PACs catches 47–57% of the money, by cycle, and a few firms with several share classes are read on a symbol other than the main one. The sector classification has errors: UnitedHealth appears under financials and Northrop Grumman under technology. The six elections in the firm test are few, and a single day does not separate the vote from other news, especially the Iran war. This text is not investment advice.