About Electorate HQ

What this is

Electorate HQ is an independent project that combines two things: a presidential election simulator in which 51 AI state delegations deliberate over any matchup you can write, and a page for every real race on the 2026 midterm ballot, built from public sources and refreshed continuously.

The premise is that elections are arguments between what candidates offer and what places need, and that simulating the argument, openly and with the reasoning shown, is a better way to explore politics than another hot take. The simulator makes counterfactuals cheap: swap a running mate, move the year, change one plank, and see which states move.

Why simulate an election at all

Forecasting an election and simulating one are different activities. A forecast tries to tell you what will happen, and is judged right or wrong on election night. A simulation asks a conditional question: given these candidates, these platforms and these conditions, how might the argument play out state by state? Electorate HQ does the second thing. That is why nothing here is called a prediction, and why the site keeps no scoreboard of its own accuracy: it would be measuring the wrong thing.

The value of a simulation is that counterfactuals become cheap. Swap a running mate, move the election to 1996, rewrite one plank of a platform, and watch which states move and which do not. Doing that by hand takes an afternoon of argument. Doing it here takes a minute, and every state shows its reasoning, so you can disagree with a specific step rather than with a number.

What the site covers

The 2026 midterms are covered race by race: every Senate contest, every governorship, and all 435 House districts, each with its own page carrying the district's real geography, what the office decides, current reporting drawn from live sources, and a clearly labeled simulation. State profiles collect what a state votes on and the issues that structurally matter there. Issue guides explain how a subject behaves electorally without arguing a position on it. A ballot lookup resolves any address to its actual districts through the Census Bureau's public geocoder.

The presidential simulator is the part that has no equivalent elsewhere: 51 state delegations, one per state and the District of Columbia, reading current headlines about their own state and reasoning through a matchup you invent.

Who runs Electorate HQ

Electorate HQ is an independent project built and edited by Sahil Manwani, a biomedical engineering student at the University of Texas at Austin who is also pursuing a minor in entrepreneurship.

The honest description of the expertise here is engineering, not political science. What that means in practice is that the site does not ask you to trust anyone's political judgement. Verified facts come from public records and are checkable: election dates and offices from official sources, district geography computed from Census boundary files, headlines linked to the publishers that wrote them. Everything the AI produces is labeled as a simulation wherever it appears, and the rules those simulations follow are published in full on the methodology page rather than described in the abstract.

The project began as an attempt to answer a question that is hard to answer any other way: not who will win, but how an argument between two platforms would actually play out across fifty-one very different places. Editorial decisions, the standards on this site, and every correction are the responsibility of one person rather than a committee, which is why the contact page reaches someone who reads it.

Sahil Manwani · Founder and editor · The University of Texas at Austin
Biomedical engineering, with a minor in entrepreneurship

Principles

• Nonpartisan by construction: the same rules apply to every ticket, and the site endorses no one.
• Simulations are always labeled as simulations, and never dressed up as polling or fact.
• Facts come from public sources; where nothing is known, the site says so instead of inventing.
• Your political choices in the simulator are never turned into advertising profiles.

How it is made

The site runs on multiple independent AI providers with automatic failover, reads live headlines and encyclopedia background from public sources, and publishes its rules openly on the methodology page. Standards for separating fact from simulation, and for corrections, live on the editorial standards page.

Found an error, or a simulation behaving badly? Tell us via the contact page. Reports like that have directly improved the product before.