2026 Primary (August 18, 2026) · 50 races, 139 candidates filed
Competitiveness = 65% vote (2024 + last leg.) · 20% R/D registration · 15% money (top-2 cash gap).
Alaska’s electorate is ~63% undeclared/nonpartisan (U+N) — averaged across the 40 current House districts, ranging 57–73%. So the registered R/D ratio describes only about a third of voters; races are decided by how that undeclared plurality breaks, which varies by region (urban undeclareds lean D, Mat-Su/rural lean R).
Empirically, across the 40 current House districts, the 2024 presidential vote crosses even at R/D ≈ 1.1 — not 1.0 (correlation R² = 0.92). A district needs more registered Republicans than Democrats just to be a 50/50 seat, and the toss-up zone is wide:
| R/D registration | Reality (2024 pres.) | Rating |
|---|---|---|
| < 0.6 | D+20 or more | Safe D |
| 0.6 – 0.9 | D+6 to D+20 | Lean / Likely D |
| 0.9 – 1.35 | D+6 … R+6 | Toss-up |
| 1.35 – 1.6 | R+6 to R+12 | Lean R |
| 1.6 – 2.0 | R+12 to R+20 | Likely R |
| > 2.0 | R+20+ | Safe R |
A district at R/D 1.3 — more registered Republicans than Democrats — is still a toss-up, not a safe R seat. That’s why the R/D chip on each card is colored by this competitive band (neutral amber in the toss-up zone), not by the naive “R ≥ D” line.
The competitiveness rating is a 0–100 index blending three signals, each first normalized to a 0–100 “competitiveness contribution” (so dollars can’t swamp vote margin on raw units):
| Signal | Weight | Contribution |
|---|---|---|
| Vote (2024 pres. + last leg. result) | 65% | 100 − 2·|margin|, blended |
| R/D registration | 20% | 100 − 2·|implied margin| (even ≈ 1.1) |
| Money (top-2 cash gap) | 15% | 100·(1 − gap ÷ combined raised) |
The vote term blends the 2024 presidential baseline with the seat’s most recent legislative result (House 2024 / Senate 2022) — because districts split tickets (HD-18 went R+17 president but R+0.6 for House in the same election). The blend is adaptive: when the incumbent is running we trust their demonstrated margin (65% legislative / 35% presidential); for an open seat a close prior race keeps it competitive while a blowout defers to the presidential baseline. Uncontested prior races (no real runner-up) are ignored. Vote is weighted highest because it integrates how the undeclared bloc actually votes.
Money is dropped (weights renormalized) when fewer than two candidates have filed, so empty-finance races aren’t flagged uncompetitive — it participates more as pre-primary reports land. Open seats get a small volatility bump. Tiers: Toss-up ≥ 84, Lean ≥ 72, Likely ≥ 54, else Safe; 1-candidate races are Uncontested.
For the House seats that went to ranked-choice rounds in 2024, we parse DOE’s cast-vote-record export (every ranked ballot) to show the real round-by-round result: first-choice shares, the final winner and margin, ballot exhaustion (Alaskans often don’t rank past #1), and where each eliminated candidate’s voters sent their second choice. This is the only source of true coalition behavior — but it reflects 2024 candidates, so read it as a guide to how a seat behaves under RCV, not a 2026 prediction.
Defined as |2024 presidential margin| − |last legislative margin| — how much closer a seat votes down-ballot than at the top of the ticket. A high positive value means lots of voters split their ballot (HD-18: R+17 president, R+0.6 House → +16), so the seat is winnable by persuasion; a value near zero is straight-ticket (a base-turnout fight); negative means the seat is held more firmly than its presidential lean suggests, usually incumbency. Available as a map layer (Crossover), a list sort, and a per-district stat. Shown only where the prior legislative race was contested.
When every candidate in a race is from one party (e.g., four Republicans, no Democrat), the R-vs-D competitiveness rating can’t see the real contest — the seat is effectively decided within the party through the top-4 primary and ranked-choice general. These races carry an ⚑ Intra-party flag so a “Safe/Lean R” (or D) rating isn’t mistaken for a sleepy race. The partisan rating still shows which party holds the seat; the flag says the fight is inside that party.
The current district lines took effect in 2022, so only 2022 and 2024 are apples-to-apples (2018/2020 were old lines). The thresholds above are derived from 2024 presidential results — the cleanest two-way R-vs-D contest under the current lines. 2022 is not folded in because it was Alaska’s first top-4/ranked-choice cycle with cross-partisan statewide races (Peltola (D) won the U.S. House; the U.S. Senate was Murkowski (R) vs. Tshibaka (R)), so it offers no clean two-way partisan margin to average in. A dedicated 2022 results pull (e.g. the Governor race) could refine these bands.