Formula Deep-Dive

This reference documents the key formulas and calculations that drive game mechanics. Understanding these helps you make optimal decisions.

Election Vote Calculation#

Vote Turn Pool#

Each turn, votes are allocated to candidates based on:

effectiveTurnPool = turnPool × (1 + (stateApproval - 0.5) × 0.2) × officeStrength

Approval is a small modifier, not a direct multiplier: 50% approval gives 1.0x, and each 10 points of approval above or below 50 shifts the pool by about 2% (tallyManagement.ts). officeStrength comes from partyStrengthWeight in the country's office-type config (src/lib/constants/countries.ts5d ago), defaulting to 0.9 when unset. US values:

• Governor: 1.0
• House: 0.9
• Senate: 0.8
• State Senate: 0.85

Candidate Vote Share#

Votes are allocated per demographic group, not with a single state-wide formula. For each group, every candidate gets a weight:

weight = appeal × reach × approval × org × regResistance × regBaseline × supportMood
         × nppPenalty × infamyMult × pgfMult × regimeMult × partyFit
         × stateOrgMult × homeStateMult × partyInfluenceMult

(distributeVotesByGroupLevelAllocation in src/lib/electionEngine/voteDistribution.ts2d ago). The candidate's share of that group's vote pool is weight / totalWeight across candidates in the group. There is no separate "state lean" multiplier: lean is baked into appeal, computed per group from the group's own economic/social lean versus the candidate's position (calcAppeal in src/lib/utils/demographicAppeal.ts2d ago). Several of the other factors (org, regResistance, regBaseline, supportMood, state-org bonus, home-state bonus, party-influence bonus) only apply when relevant (general election, presidential primary, etc.); they default to 1.0x (no-op) otherwise.

• Reach = normalized political influence (or national influence, presidential path), floored at VOTE_REACH_FLOOR = Math.sqrt(1/100) (electionFormulaFactors.ts) so influence 0 never produces a literal 0x weight.
• Appeal = a position score (power curve, exponent APPEAL_POSITION_EXPONENT = 1.5) plus a per-axis tribal "direction bonus" (DIRECTION_BONUS_PER_AXIS = 5) when the candidate's lean matches the group's lean direction, plus an influence score if enabled. Floored at APPEAL_POSITION_FLOOR = 0.5 per axis so no candidate is fully locked out of a group. Max appeal is 50 (demographicAppeal.ts).
• Party org (general elections only) = orgVoteWeight, a diminishing-returns curve on normalized state org share, not a flat 0.3-1.0 range.
• Approval = approvalScalar(effectiveFavorability), a balanced curve centered on 50% favorability.

State Org and Home State Bonuses#

Not part of the vote-share formula above but stack into it as stateOrgMult / homeStateMult (src/lib/electionEngine/constants.ts2d ago):

• MAX_STATE_ORG_BONUS_PRIMARY = 0.25 (primary path cap; the fraction of it delivered follows an uncapped exponential-approach curve, 1 − (1 − 0.75)^(level / 10), not a linear ramp, level 10 buys 75% of the max bonus, level 20 buys 93.8%, approaching but never reaching 100%)
• MAX_STATE_ORG_BONUS_GENERAL = 0.15 (general path cap, smaller so it doesn't dominate lean × party position)
• HOME_STATE_BONUS_PRIMARY = 0.1 (flat bonus in a candidate's home state during a primary)
• HOME_STATE_BONUS_GENERAL (smaller general-election equivalent)

Minimum Vote Floor#

There is no 50%-of-base-vote floor. The actual floors are on the individual factors that feed the weight: VOTE_REACH_FLOOR clamps reach to at least ~0.1x, and APPEAL_POSITION_FLOOR (0.5, about 2% of max appeal) keeps position score above zero per axis. Together these keep even a badly-misaligned, low-influence candidate from hitting a literal zero weight in a group, but they do not guarantee any fixed share of the base vote pool.

Political Influence#

Decay#

decay = influence × 0.0075  (0.75% per turn)

NPP Influence Floor#

NPPs cannot decay below 10% Political Influence.

Presidential Election Scoring#

Primary Score#

score = alignmentScore(0-40) + partyInfluenceScore(0-20) + nationalReachScore(0-15) + favorabilityScore(0-25)

Weights are PRESIDENT_PRIMARY_ALIGNMENT_WEIGHT = 40, PRESIDENT_PRIMARY_PARTY_INFLUENCE_WEIGHT = 20, PRESIDENT_PRIMARY_NATIONAL_REACH_WEIGHT = 15, PRESIDENT_PRIMARY_FAVORABILITY_WEIGHT = 25 (src/lib/primaryScore.ts2d ago, rebalanced 2026-08-19). Party influence, not "party org," is the second term; it scales linearly and is uncapped above the reference scale (party influence 150). Favorability was raised from 10 to 25 because it is the single most decisive variable in the general election (approvalScalar measured at ~0.45 vote-share points per favorability point) while party influence never enters the general-vote formula at all, the old weighting let parties nominate candidates the electorate had already rejected.

Electoral College#

• 538 total electoral votes
• 270 needed to win
• ME and NE split by congressional district with lean proxies (UNIT_LEAN)
• 269-269 tie: deterministic coin flip (SHA-256 of election + candidate IDs)

Independent Penalty#

Independent candidates receive 0.3x vote multiplier.

Demographic Turnout#

Turnout Modifier Range#

Each demographic group: -20% to +20% modifier

Decay#

modifier = modifier × 0.98  (2% decay per turn toward 0)

Canvassing Effectiveness#

boost = baseBoost × alignmentMultiplier × seasonMultiplier × diminishingFactor

Where:

• alignmentMultiplier = 0.1 to 1.0 based on ideological distance
• seasonMultiplier = 2.0 during campaign season (4 turns before election), 1.0 otherwise
• diminishingFactor = smaller boost when existing modifier is already large

Party GOTV#

Party budgets have a GOTV allocation (gotvBudgetPerTurn, gotvBudgetPercent) targeted at a specific category or group (gotvTargetCategory / gotvTargetGroup), which feeds turnout for the tally. There is no verified "auto-boost demographics within 2 points of party position" rule in code, that claim is unconfirmed and removed rather than guessed at.

Fund Generation#

Passive Income#

income = statePopulationTier × donorLevelMultiplier + officeBonus

Office Fund Bonuses#

Action Economy#

Base Actions#

4 actions per turn (base), plus office bonuses:

• Governor: +2 actions
• Other offices: varies

Action Costs#

All standard actions cost 1 action point.

State Adjacency Costs#

• Home state: 1.0x
• Neighboring state: 1.25x
• Non-neighboring: 1.5x

State Lean Calculation#

stateLean = weightedAverage(groupLeans, groupPopulations × groupTurnout)

Lean is stored on an integer -5 to +5 scale per axis (economic, social), not a -1..1 float. Display labels (getEconomicPositionName / getSocialPositionName, src/lib/utils/politics.ts13h ago) use 11 discrete buckets, one per integer value:

Economic axis: Far Left (-5), Strong Left (-4), Left (-3), Lean Left (-2), Center-Left (-1), Centrist (0), Center-Right (1), Lean Right (2), Right (3), Strong Right (4), Far Right (5).

Social axis: Far Liberal (-5), Strong Liberal (-4), Liberal (-3), Lean Liberal (-2), Center-Liberal (-1), Moderate (0), Center-Trad (1), Lean Trad (2), Traditional (3), Strong Trad (4), Far Traditional (5).

Related Pages#

• Election Mechanics, Election system overview
• Demographics & Targeting, Demographic details
• Stats & Actions, Action costs and stats
• Government Approval, Approval system