Mary Peltola leads Alaska’s reported first-choice Senate count by just 637 votes. Yet Votendi’s central scenario points to a larger final advantage: approximately 51.3% for Peltola and 48.7% for Republican Senator Dan Sullivan, a margin of about 7,400 votes after ranked-choice transfers.
- What the reported count actually says
- Why the remaining ballots could help Peltola
- Yes, the 51.3–48.7 estimate already includes ranked-choice voting
- Five scenarios, from a clearer Peltola win to a Sullivan comeback
- Could stronger transfers alone save Sullivan?
- What Polymarket’s 97% does — and does not — mean
- What to watch next
Polymarket has displayed an approximately 97% implied chance of a Peltola victory. That market snapshot and our projected vote share measure different things. Neither is an official result, and the market price can change.
Votendi scenario analysis · November 5, 2026
Peltola 51.3% · Sullivan 48.7%
Conditional estimate, not a race call. This assumes 65,000 outstanding ballots and a particular mix of voting methods. That remaining-ballot inventory has not been independently verified.
The central question is not simply who benefits from voters’ second choices. It is who the remaining voters chose first — and whether they resemble the early and absentee voters already counted.
What the reported count actually says
The Decision Desk HQ snapshot used in this analysis has Peltola on 102,962 first-choice votes and Sullivan on 102,325. The two other named candidates, Gerald Heikes and Daniel J. Sullivan Jr., have 4,846 and 3,600 respectively.
Those figures are not the final head-to-head result. Outstanding ballots still need to be counted, and lower-ranked choices may then change the totals.
Why the remaining ballots could help Peltola
The reference breakdown used in our model shows Peltola winning roughly 58% of the early and absentee first-choice votes cast for the two leading candidates, compared with about 47% on Election Day. The earlier report on the state’s voting-method breakdown describes that same divide: Sullivan ahead among Election Day voters, Peltola ahead among early and absentee voters.
Our central case assumes another 65,000 ballots: 20,000 early ballots not yet counted, 35,000 absentee ballots and 10,000 questioned or other outstanding ballots. We apply the earlier voting-method patterns to the first two groups and use the Election Day split as a cautious proxy for the last group.
This is a modelling assumption, not a confirmed list of ballots still waiting to be counted. Later batches could come from different communities or different kinds of voters. That is the main reason the outcome can diverge from the central estimate.
Yes, the 51.3–48.7 estimate already includes ranked-choice voting
Under Alaska’s counting system, if nobody wins a majority, the lowest-placed candidate is eliminated and their ballots move to the next eligible candidate each voter ranked. This continues until a winner emerges. A ballot with no remaining eligible choice becomes inactive and is excluded from the final-round percentage.
Our estimate includes transfers on both already reported and assumed outstanding ballots. It is a projected final vote share, not Peltola’s first-choice share or her probability of winning.
Some preliminary simulations show Sullivan winning 51–49. FairVote explains why that is not a statewide forecast: the available digital ballot file covers only part of the reported count and starts with Sullivan ahead. Our model accounts separately for already reported votes outside that file; it does not add them again as uncounted ballots. Our Alaska ranked-choice explainer covers this distinction in more detail.
Five scenarios, from a clearer Peltola win to a Sullivan comeback
The table changes Peltola’s first-choice support among the remaining voters who choose either her or Senator Sullivan. It keeps the assumed 65,000 outstanding ballots and the baseline transfer patterns unchanged.
| Peltola’s share of remaining head-to-head first choices* | Modelled final result Peltola – Sullivan |
Approximate winning margin |
|---|---|---|
| 56.4% · central case | 51.3% – 48.7% | Peltola +7,400 |
| 54% · weaker late support | 50.8% – 49.2% | Peltola +4,400 |
| 52% · a much closer finish | 50.3% – 49.7% | Peltola +1,900 |
| 51% · near the tipping point | 50.1% – 49.9% | Peltola +600 |
| 50% · an even remaining split | 49.9% – 50.1% | Sullivan +600 |
*The first column excludes first-choice votes for other candidates. Those ballots are handled separately through the transfer model. Final percentages exclude inactive ballots. All figures are rounded and represent scenarios, not measured outcomes.
The model’s tipping point is approximately 50.5% of the remaining first-choice votes for Peltola and Senator Sullivan combined. Below that, Sullivan moves ahead under the baseline assumptions. Above it, Peltola wins.
In other words, Sullivan’s route is to bring the remaining head-to-head first choices close to an even split. The central case expects Peltola to win 56.4% of them, so her late support would have to fall about six percentage points short of that expectation.
Could stronger transfers alone save Sullivan?
The observed backup choices favour Senator Sullivan overall, but they do not all go to him. FairVote’s analysis finds that some voters supporting the lower-placed candidates ultimately transfer to Peltola, while others leave no valid choice between the finalists.
We therefore tested more favourable transfer patterns for Sullivan. Keeping the central first-choice assumptions unchanged, a substantially stronger transfer performance still leaves Peltola at approximately 51.2%.
Even an extreme test sending every still-unknown minor-candidate and write-in transfer to Sullivan leaves Peltola at around 50.8%. Already observed transfers are not changed in that test.
That does not rule out a Sullivan victory. It shows that under this assumed remaining-ballot mix, changes in first-choice support matter more than changes in transfers alone. Both can change in the real count.
What Polymarket’s 97% does — and does not — mean
Polymarket’s Alaska market has quoted approximately 97% for Peltola. We use that as a market snapshot, not a continuously updated price.
A 97% market quote is an implied chance of winning. Our 51.3% scenario is a share of the final active vote. A candidate can be strongly favoured to win by a relatively narrow margin, so the two figures are not contradictory.
However, Votendi is not assigning a calibrated 97% win probability. That would require firmer evidence about the number and composition of outstanding ballots, alongside a tested estimate of how often the model’s assumptions fail. The analysis supports Peltola as the favourite within the stated assumptions, not as a guaranteed winner.
What to watch next
The Alaska Division of Elections schedules its ranked-choice tabulation for November 18, 2026, after eligible ballots have been reviewed and counted, for races requiring additional rounds.
Before then, the most useful signal is Peltola’s share in newly reported batches. A sustained advantage in the mid-to-high 50s among first choices for the two leaders would support the central scenario. Batches nearer an even split would make the race substantially more competitive.
The bottom line: the 637-vote lead understates Peltola’s advantage if the outstanding electorate resembles the early and absentee voters already counted. But that resemblance is an assumption to test against new returns — not a fact to take for granted.
Follow the reported count and race calls in Votendi’s live US midterm results.
Methodology note — November 5, 2026: This is a deterministic scenario exercise, not a new poll, an official ranked-choice count or a historically calibrated election forecast. It combines the reported first-choice snapshot with the preliminary ballot-file tabulation retained in Votendi’s working model and rounded transfer patterns described by FairVote. Reported ballots outside the preliminary file are estimated separately and counted only once. The central case assumes 65,000 outstanding ballots, retains the modelled minor-candidate shares and accounts for inactive ballots. The questioned/other-ballot split and future write-in transfers are assumptions, not observed results. No verified district-by-district inventory of outstanding ballots is available in this model. The Polymarket figure is a separate, changeable market snapshot and does not determine any of the model’s vote totals.
