Reference: Formulas

Complete formula reference for A House Divided. All values are sourced directly from game code. Formulas marked with (*) use diminishing-return normalization curves: see the NPI normalization section below.

Primary score formulas#

State-level primary score#

Used for: US House, US Senate, US Governor, US State Senate, UK Commons, UK Regional Council, DE Bundestag, JP Shūgiin, JP Sangiin.

Maximum: 100 points (raw), then multiplied by an infamy penalty.

• Alignment splits into two: 25 pts on the state's cached economic/social lean, 15 pts on party platform.
• When the state has no cached lean, scoring falls back to a single 40-pt party-only alignment: max(0, 40 − (|econDiff_party| + |socialDiff_party|) × 2.0).
• Final score = raw × (1 − 0.05 × min(100, max(0, infamy))/100), a 5% reduction at infamy=100.

Quick alignment reference (state-vs-state-lean OR state-vs-party):

NPP primary score penalty: When at least one player is in the same party's primary, all NPP candidates in that primary receive a ×0.5 multiplier on their total score. An NPP with a raw score of 70 competes at an effective 35 against you. NPPs aren't subject to the infamy penalty.

Presidential primary score#

Used for: US President only.

Maximum: 100 points

Same infamy penalty applies on the final score. Key differences from the state formula:

• No state-position alignment: presidential primaries are national, only the party platform matters.
• Alignment stays dominant at 40 pts, so parties still favor candidates who agree with them.
• Favorability is worth 25 points, high enough to matter but never enough on its own to outweigh alignment.
• Party influence and national reach split the rest (20 and 15 pts), with party influence still outranking raw reach.
• NPI normalization curve is different: uses a diminishing-returns curve 1 − exp(−NPI/45) (normalizeNationalReachPresidentialPrimary), not the sqrt curve that state PI uses. The diminishing shape keeps favorability and reach meaningful against a moderate NPI gap.

Canvassing formula#

Cost: ₳100 + 1 action. Home state / region only.

baseBoost        = 0.05 (percentage points)
distance         = |charEcon − demoEcon| + |charSocial − demoSocial|
alignmentMult    = max(0.1, 1.0 − distance × 0.15)
seasonMult       = isCampaignSeason ? 2.0 : 1.0
rawBoost         = baseBoost × alignmentMult × seasonMult
adjustedBoost    = rawBoost × (1 − |currentModifier| / 20)   // diminishing returns
newModifier      = clamp(currentModifier + adjustedBoost, −20, +20)

• Campaign season = election active with endTime within 4 hours (4 turns)
• Cap: ±20 percentage points turnout modifier
• Decay: 2% of current value per turn toward zero

Alignment multiplier table:

Campaign ops trees#

Fundraising, Ground Game, Media Spending, and Opposition Research are each a branch tree on the campaign page (/campaign/[id]), not a flat level ladder. Each tree unlocks a starter node, then up to three further branches (for example Media Spending's Broadcast and Television branches, Opposition Research's Dossier, Scandal Leak, and Counter-Intel). Branch magnitudes stack with the starter and with each other. Maintenance is deducted from campaign funds each turn; branches that can't be covered auto-downgrade rather than collapsing the whole tree, with no refund.

Full per-tree cost and effect tables, current to the live constants: Campaign Strategy and Campaign Manager.

Campaign season multiplier: All campaign passive effects (Media Spending, Opposition Research, Travel Presence, Primary Campaign bonus) double during the final 4 turns of an election.

Vote accumulation (general elections)#

Per-turn vote pool weighting#

Early band:
  turnPool = 0.50 × totalPool / earlyTurnCount

Ramp band (up to 8 turns before the final band):
  turnPool = 0.20 × totalPool / rampTurnCount

Final 4 turns:
  turnPool = 0.30 × totalPool / 4

The final 4 turns collectively deliver 30% of all votes cast in an election. Races of 4 turns or fewer split the pool evenly.

Total appeal pipeline (per demographic group, per candidate, per turn)#

reach         = normalizeNPI(influence)         // sqrt curve, capped at 1.0 when PI/NPI reaches 100
positionRaw   = max(0, 50 − |econDiff|×5 − |socialDiff|×5)
positionScore = 25 × (positionRaw / 50)^1.5 + 0.5
direction     = up to 5 per aligned ideological axis
influenceAppeal = includeInfluence ? normalizeNPI(influence) × 12.5 : 0
appeal        = positionScore + direction + influenceAppeal
effectiveFav  = clamp(favorability + groupApproval × 0.5, 0, 100)
approval      = (effectiveFav / 100)^0.8
orgShare      = statePartyOrg / sum(all party Org in the state)
partyOrg      = orgShare^0.2
personalFloor = 0.1 × reach × approval
effectiveOrg  = max(partyOrg, personalFloor)
infamyMult    = 1 − 0.05 × (min(100, max(0, infamy))/100)
weight        = appeal × reach × approval × effectiveOrg × infamyMult × other configured general-election factors

State races use influence for reach but not influenceAppeal; presidential generals include it. Directional bonuses are continuous around the center and may be suppressed when a leaning candidate points against their party on that axis. General-election factors also include registration resistance, seeded registration baseline, candidate Support, campaign presence, NPP and regime multipliers, and race-specific coattails where configured.

Office strength multipliers#

NPI normalization (*)#

Political Influence (PI) and National Political Influence (NPI) pass through normalizing curves before use in formulas. normalizeNPI is a sqrt curve clamped to [0, 100]: meaningful spread at the high end, hard-capped at 1.0:

normalizeNPI is used by state-level primary scoring AND by general-election reach (state and presidential).

Presidential primaries use a separate curve (normalizeNationalReachPresidentialPrimary): diminishing returns via 1 − exp(−NPI/45). NPI=25 → 0.426, NPI=55 → 0.706, NPI=80 → 0.831, NPI=100 → 0.892, NPI=200 → 0.988. Above the high end the curve continues to creep toward 1.0 but never reaches it, so the celebrity-bonus regime is naturally bounded.

Party organization scalars#

Party org (0 to 100) affects:

An empty state Org map gives every party a neutral 1× fallback. In a populated map, a party with 0 Org has 0 Org weight before the personal floor. A 3:1 Org lead produces about a 1.25:1 Org-weight advantage because of the 0.2 exponent.

Turnout modifier decay#

Each turn: newModifier = currentModifier × 0.98

Effect accumulates over turns:

Related#

• Reference: Turn Order: phase execution order.
• Election Mechanics: full election structure and vote accumulation.
• Primaries: declaration rules and NPP dynamics.
• Canvassing: turnout boost formula in detail.
• Campaign Strategy: upgrade tables and action allocation.