NYC 2025 Mayor Race Primary Polling

Ranked Choice Simulation

The Ranked Choice Simulation starts with each candidate holding their share of support in the polling average, adjusted for momentum. In each round of ranked choice voting, the candidate with the least support is eliminated, and their support is redistributed to the remaining candidates. This continues until only two candidates remain, at which point the leading candidate is crowned the winner.

Momentum Factor – Calculating Candidate’s Support in Round 1

Unlike in a general election, a candidate soaring in the polls entering election day is more likely than not to continue gaining support. When this happens, the polls are unable to fully capture their support simply because most of them finish before the last day or two of the campaign.

Compared to general election voters, primary voters are far more open to changing their minds, and there’s always a share of undecided voters making their decision in the last few days. Whether it’s a presidential, Senate or mayoral race, we’ve found our primary forecasts becomes more accurate when we adjust for momentum.

To calculate momentum, we compare each candidate’s current polling to their polling seven days earlier, after excluding for undecided voters. We track how much their support has changed and multiply that change by 25%. This is the Momentum Factor. We then add the momentum factor to their current polling average to project each candidate’s support in the first round.

Calculating How the Vote Will Be Distributed

In ranked choice voting, each voter ranks up to five candidates. When a voter’s first-choice candidate is eliminated, their vote is redistributed to their next preferred candidate. The crux of building a ranked choice simulation is determining how votes are reallocated based on which candidate is eliminated and which remain.

In a race as frequently polled as the NYC mayoral race, we have a wealth of data to help solve this problem. Many pollsters provide data showing how votes are redistributed in each round. We’ve built a specialized polling average to track this, weighing each poll based on its historical accuracy, sample size, and recency—giving greater weight to non-internal campaign polls.

We use four tools to assess how votes are redistributed:

1.      Scenario Specific Ranked Choice Polling: Often, our ranked choice simulation contains a scenario we’ve frequently seen in the NYC Mayor polling. For example, let’s say the final three candidates Brad Lander, Andrew Cuomo and Zohran Mamdani. We have half a dozen polls that show how Brad Lander’s support is redistributed when he is eliminated, and Andrew Cuomo and Mandani are the only candidates left. When available, this tool receives 90% of the weight in the simulation.

2.      Redistribution Across all Scenarios: This tool assesses how a candidate’s support is typically redistributed, regardless of which candidates remain. It is especially useful during the early eliminations, when more than seven candidates are still in the race. At that stage, it matters less if a lower-polling candidate like Jessica Ramos is still active, as she is ranked lower than the leading contenders.

3.      Second-Choice Polling Average by Candidate: We maintain an internal polling average that tracks second-choice preferences based on the voter’s first-choice candidate. This receives only a small portion of the weight, as relatively few pollsters publish this information.

4.      Second Choice Polling Average – All Voters: As a last resort, we use an average that reflects which candidate NYC voters - regardless of first choice - identify as their second choice. This is a blunt tool and is rarely influential in our simulation, likely contributing no more than 1–2% in any round.

Using this information, the Ranked Choice Simulation proceeds round by round, eliminating the lowest-performing candidate and redistributing their support. The process continues until only two candidates remain, allowing us to project a winner.

Interactive Features