What we mean when we say your ballot is worth "3.2× the average voter"
Most election maps tell you who will win. This map answers a different question: how much power does your ballot actually carry? Not all votes are equal — a voter in a tight Maine district with a live Senate race holds many times the influence of a voter in a safe seat with no Senate election. We measure that, end to end, in three layers.
A single vote matters when an election would otherwise be exactly tied. That probability is driven by two things:
fi(50%) is the forecast probability density at an exact tie; Ni is expected turnout. This is the approach of Gelman, Silver & Edlin (2012), applied to every congressional race.
A flipped seat matters mostly through what it does to the chambers. We simulate 20,000 joint election nights, where every race shares correlated errors — one national swing, one state-level swing per state, plus race-specific noise (the correlation sizes are estimated from how states and districts actually swung between 2020 and 2024). In each simulated world we know exactly how many seats each party holds in both chambers, so we know precisely what one more seat would have changed.
This is the heart of the model. Holding a chamber isn't an on/off switch, and the two chambers don't add — power flows through specific institutional channels, each requiring its own combination of institutions:
| Channel | What it is | Requires |
|---|---|---|
| Nominations | Judges, justices, cabinet | Senate majority + presidency |
| Reconciliation | Taxes & spending by simple majority | House + Senate majorities + presidency |
| Legislation | Ordinary lawmaking | House majority + 60 Senate votes (or a rules change) + presidency |
| Oversight | Investigations, subpoenas, agenda control | Either chamber alone |
| Leverage | Must-pass extraction (budgets, debt limit) | House majority, or just 41 Senate seats |
A 3-seat House majority cannot pass what a 30-seat majority can — every marginal member is a potential defector. So each chamber's capability is a steep step at control (the Speaker's gavel, committee chairs, the confirmation calendar arrive all at once) plus an S-curve in the working margin. The Senate has four thresholds, not one: 41 (deny the other side cloture — real power for a minority), 50 (control), 60 (break filibusters), 67 (veto overrides, treaties).
A majority with 52 seats might weaken the filibuster; one with 50 can't credibly try. The model treats the 60-vote threshold as softened by a rules-change probability that rises with margin.
There are two ways a seat is powerful: helping your side do things, and stopping the other side. These are not separate scores — blocking power is exactly the affirmative power you deny your opponent. So we score every outcome by the balance of power: each party's capacity across all channels, netted. One seat shifts that balance by the same amount no matter which side it lands on, which is why the map shows a single number that's true for every voter.
Δbalance depends on the whole simulated world — the same Senate seat is worth more when the House has also flipped (channels need both chambers), and a seat that gets a minority to 41 is worth real blocking power even with control out of reach. The presidency enters as a hard requirement for most channels; it's not on the ballot until 2028, so it sits at its current value in every world.
In November 2026 your ballot carries a House vote everywhere, plus a Senate vote in 35 states. Each vote's power is computed against the same simulated worlds, then added — legitimately, because each one's Δbalance already includes everything the other races did in that world. The map's headline number is this combined ballot power, expressed as a multiple of the average voter — a randomly drawn 2026 voter, so high-turnout districts count for more in the baseline. "3.2×" means your ballot carries 3.2 times the power of a typical American's.
This is the actual value model running live. Set the seat counts and the presidency, and watch where the power balance lands — and what the next seat in each chamber would be worth. Positive = Democratic advantage, negative = Republican.
The same function over every seat combination — one map per presidency. Blue = Democratic power advantage, red = Republican. The white dot is wherever you set the sliders; dashed lines mark the House majority (218) and the Senate's 41 / 51 / 60 thresholds. Notice how much of the map a hostile presidency flattens — and how the minority-blocking band at 41 never flattens at all.
Vertical axis: Democratic Senate seats, 34 (bottom) to 66 (top). Color: power balance from R+ (red) through even (dark) to D+ (blue).
Everything on this site reduces to four equations. Parameters are listed below with their justifications — the structure is the thesis, several parameters are still working assumptions awaiting estimation.
f_i = forecast density of the Democratic two-party share at exactly 50% (normal, with national + state + race error components); N_i = expected votes cast (actual 2022 totals, FEC).
A near-step at control (the gavel arrives all at once) plus an S-curve in working margin (every marginal member is a potential defector). The cloture line prices the filibuster's survival as a function of majority margin (ν = max rules-change probability).
A CES power mean. ρ → −∞ is a strict veto-player AND; ρ = 1 is linear. ρ = −2 says institutions are strong — but not perfect — complements: executive action, reconciliation, and the courts leak around missing pieces.
One seat moves the balance identically whichever side wins it, so this is one consistent number for every voter. The expectation runs over 20,000 correlated simulated elections; ΔV is evaluated against each world's other outcomes in both chambers.
| Parameter | Value | Justification |
|---|---|---|
| Channel weights | nominations .25, reconciliation .20, legislation .25, oversight .15, leverage .15 | Working assumption reflecting the modern mix: judiciary and fiscal policy dominate; ordinary legislation is rarer; oversight and must-pass leverage are real but secondary. Estimation target: directional shares of landmark enactments (Mayhew's dataset). |
| α (organizational share) | 0.5 | Half of a chamber's value arrives with bare control (speaker, chairs, calendar), half scales with margin. Assumption; bounded by the observation that bare majorities still confirm judges but struggle to legislate. |
| ρ (CES exponent) | −2 | Institutions are strong complements with leakage (executive orders, reconciliation carve-outs, courts). Will be sensitivity-tested rather than estimated. |
| ν (max filibuster-change prob) | 0.5 | Even a comfortable majority is no better than a coin flip to change the rules — both parties have flinched repeatedly (2013/2017 carve-outs came only under extreme pressure). |
| Sigmoid steepness k | 0.5 control / 3.0 House margin / 1.5 Senate margin | Control is nearly a step; working-margin curves are wider in the House (more members, more defectors). Estimation target: majority-party win rates on contested roll calls vs margin (Voteview). |
| Error components | national 4.5 / state 2.5 / district 3.6 pts | State and district sds estimated from observed 2020→2024 swings; national sd is a prior (one cycle can't identify it); candidate effects: 5.5 pts House, 7 pts Senate (priors). |
Presidential results by district and state: The Downballot. Generic-ballot polling: VoteHub. Turnout: FEC official 2022 results. Boundaries: U.S. Census Bureau (119th Congress). Decisive-vote framework: Gelman, Silver & Edlin, "What is the probability your vote will make a difference?" (Economic Inquiry, 2012).