← Back to scans

Will Alberta vote to secede from Canada before the next general election?

KXALBERTAREFYES-29 · Elections · 2026-04-14
10%
Agent
22%
Market Price
-12.0%
Edge
45%
Confidence
Volume: 56,730
Spread: 3.0c
Days to resolution: 1280
Markets in event: 1
Final Rationale
My independent estimate of 5.2% was built on a ~55% chance of a referendum being held and a 6% conditional probability of YES winning. I PARTIALLY ACCEPT the Devil's Advocate's critique that the 6% conditional YES probability is too low: the base rate sample of 0/3 is tiny (Laplace correction suggests ~20%), Quebec reached 49.4% from a similar starting position, and we lack post-April 2025 polling. I adjust the conditional YES probability upward to ~12-15%, giving P(ref)×P(YES|ref) ≈ 0.55×0.13 ≈ 0.07. I ACCEPT the math error flag — my ensemble pulled the number away from the stated framework. I PARTIALLY ACCEPT the market interpretation concern: the market at 22% likely conflates 'will a vote happen' with 'will YES win,' and its wide historical range (3%-71%) with medium volume suggests low reliability. However, I cannot fully dismiss the market signal, so I adjust modestly upward to 0.10. The remaining 12-point gap from the market is justified by: (1) polling at 30-35% is far below the 50% needed, (2) Alberta's independence movement lacks the decades of institutional infrastructure Quebec had, and (3) Premier Smith herself opposes secession, making a government-backed YES campaign unlikely.
Price vs. prediction
How the market moved after the agent's call. Solid blue line is the actual market price; dashed line is the agent's static forecast at $t{=}0$.
$n = 58$ follow-ups
1. Planner Opus · Fermi decomposition + tool plan
Tools chosen: kalshi_data article_search web_search wikipedia_lookup code_execution
Sub-questions (Fermi decomposition)
  1. Will Alberta hold a formal secession referendum before the next Canadian federal general election? (w=0.50)
  2. If a secession referendum is held, will a majority of Albertans vote YES to secede? (w=0.30)
  3. Will the Alberta UCP government or a successor government pursue a formal independence referendum process within the next 3 years? (w=0.20)
Planner reasoning
This question asks whether Alberta will both HOLD a secession referendum AND vote YES before the next Canadian federal general election (~October 2029, ~42 months away). The key causal chain has multiple sequential steps: (1) Alberta government must decide to pursue a referendum, (2) introduce and pass referendum legislation in the AB legislature, (3) hold a campaign period, and (4) win a majority YES vote. The reference class is subnational independence movements in stable democracies (Quebec 1980/1995, Scotland 2014) — in every case, formal referendums were rare and YES votes rarer still. Critical contextual factors: the Canadian Clarity Act requires the federal government to negotiate only if the question is 'clear' and the majority 'sufficient,' creating major legal hurdles; Premier Danielle Smith has focused on sovereignty assertions short of full separation; Alberta polling shows ~30-35% support for independence, far below a winning threshold; no referendum legislation has been introduced. The temporal analysis matters: even if the Alberta government made a strong commitment TODAY, the minimum realistic timeline to a referendum is 12-18 months, leaving a tight window. The main scenario for YES would require a major political shock (e.g., federal policies deeply antagonizing Alberta, a dramatic UCP mandate shift after a provincial election) to accelerate the process. This is an extremely low-probability event even over a 3.5-year horizon.
2. Research Parallel tool calls · legacy v2
ToolStatusTimeItemsSummary
kalshi_data OK 1.5s -
article_search OK 0.1s -
web_search OK 68.5s -
wikipedia_lookup OK 1.4s -
code_execution OK 0.0s -
3. Evidence Synthesis Sonnet · 15 items · legacy v2
#StrengthCredibilityDirectionSourceClaimPriced In
1 STRONG 78 ↑ UP web_search A referendum on Alberta independence is scheduled for October 19, 2026, and Stay Free Alberta claims to have gathered the required 177,732 signatures (10% of votes in last provincial election) by March 30, 2026, ahead of the May 2 deadline. No
2 MODERATE 72 ↓ DOWN web_search Elections Alberta has not yet confirmed the signature count and cannot do so until after the May 2, 2026 deadline, leaving the official triggering of the referendum unverified. No
3 STRONG 85 ↑ UP web_search Alberta passed Bill 54 (Election Statutes Amendment Act) in 2025, lowering the threshold for citizen-initiated referendums, directly enabling the current petition process for a secession question. Yes
4 MODERATE 80 NEUTRAL web_search Premier Danielle Smith has committed to putting any citizen initiative that meets the signature threshold on the ballot, but personally does not support separatism and frames her position as 'a sovereign Alberta within a united Canada.' Yes
5 STRONG 88 ↑ UP web_search Elections Alberta approved the petition application on December 22, 2025, and authorized signature collection from January 3 to May 2, 2026, establishing a clear procedural pathway for the referendum question. Yes
6 MODERATE 60 ↑ UP kalshi_data Kalshi market 'KXALBERTAREFYES-29' is currently priced at 22%, up 5% over 7 days but down 5% over 30 days, with moderate daily volume of 336 contracts. Yes
7 MODERATE 45 ↓ DOWN code_execution Base rate analysis of subnational secession referendums estimates only a 6% probability that Alberta holds a formal referendum before the next federal election, based on historical reference class. No
8 STRONG 90 ↓ DOWN wikipedia_lookup The Clarity Act requires the federal government to negotiate secession only if the referendum question is 'clear' and the majority is 'sufficient,' creating significant federal-level legal and political hurdles even if Alberta votes YES. Yes
9 STRONG 72 ↓ DOWN web_search Polling consistently shows only approximately 30-35% of Albertans support full independence, well below the majority threshold needed to win a secession referendum. Yes
10 MODERATE 85 ↓ DOWN wikipedia_lookup The 1995 Quebec referendum — the closest historical analog — narrowly failed at 49.42% YES despite decades of sovereignty movement infrastructure, illustrating how difficult it is to achieve a majority YES even under favorable conditions. Yes
11 MODERATE 45 ↓ DOWN code_execution Base rate model estimates only a 30% probability that a YES vote would prevail conditional on a referendum being held, based on historical subnational independence votes and current polling. No
12 STRONG 83 ↑ UP web_search Premier Smith passed Bill 54 and committed to holding a referendum if signature thresholds are met, representing a formal government embrace of the referendum process even without personally advocating independence. Yes
13 STRONG 80 ↑ UP web_search The referendum question proposed is specifically 'Do you agree that the Province of Alberta should cease to be part of Canada to become an independent state?' — a full secession question, not merely a sovereignty or autonomy question. No
14 MODERATE 82 ↓ DOWN wikipedia_lookup The Alberta Sovereignty Act (2022) focused on asserting provincial autonomy against federal overreach, not outright secession — indicating the UCP's prior preference for confrontational federalism over independence. Yes
15 MODERATE 75 ↑ UP web_search The Liberal Party of Canada's re-election in April 2025 is cited as the proximate political catalyst that accelerated the Alberta independence petition effort and legislative changes. Yes
Information Gaps
  • No confirmed independent verification of signature count — Elections Alberta has not audited or confirmed the claimed 177,732 signatures; significant uncertainty remains about whether the threshold is officially met.
  • No current polling data specifically on Alberta independence support post-April 2025 Liberal re-election, which may have shifted sentiment upward from the baseline 30-35%.
  • No information on whether the federal government would challenge the referendum question under the Clarity Act or attempt to block proceedings.
  • No data on what happens procedurally if signatures are verified — timeline between verification, referendum scheduling, and campaign period before October 19, 2026.
  • No information on whether Premier Smith faces internal UCP pressure or a leadership challenge that could alter her commitments.
  • No data on the legal validity of the referendum question under the Clarity Act — whether 'cease to be part of Canada' would be deemed a 'clear' question.
  • No information on whether a YES advisory vote in October 2026 would actually trigger secession negotiations or merely be symbolic, which is critical for the question's resolution criteria.
Key Uncertainties
  • Whether Elections Alberta confirms the signature count by May 2, 2026 — this is the immediate binary that determines whether a referendum is formally scheduled.
  • Whether the October 19, 2026 referendum, even if held, would constitute a legally binding 'secession vote' or merely an advisory referendum — the question resolution criteria require a 'vote to secede,' not merely an advisory question.
  • Whether federal intervention (under the Clarity Act or otherwise) could delay or invalidate the referendum process.
  • Magnitude of polling shift in Alberta following Liberal re-election and any new federal policies antagonizing Alberta — current ~30-35% support is below a winning threshold but could rise.
  • Whether the 22% Kalshi price reflects the probability of the referendum being held AND producing a YES vote, or just the referendum being held — interpretation matters significantly.
  • Timeline: even with a successful October 2026 referendum, the question asks about voting to 'secede,' which requires YES majority; the next step of actual secession negotiation would be far longer.
Market Context
Price: 0.22 · 7d: +0.05 · 30d: -0.05
Volume: MODERATE · Depth: MEDIUM
Reliability: MEDIUM — moderate daily volume (336 contracts), wide historical price range (3%-71%) suggests high uncertainty and thin liquidity at extremes; recent 5% uptick likely reflects signature threshold news but 30-day decline suggests prior market skepticism; the market appears to price the referendum being held rather than YES winning, making it potentially conflated with the sub-question structure.
4. Independent Forecaster Opus · No market price · legacy v2 Did NOT see market price
Independent estimate: 5% (65% confidence)
base rate
4%
62% conf
evidence driven
8%
55% conf
contrarian
4%
78% conf
Spread
4pp
Low disagreement
sq1: Will Alberta hold a formal secession referendum before the next Canadian federal general election?
55%
Base rate ~2% for subnational referendums. Massive updates from: Bill 54 passed (+15%), active petition with approved application (+15%), claimed sufficient signatures (+20%), Smith's commitment (+5%). Downward: unverified signatures (-10%), Clarity Act risks (-3%). Final: 55%.
sq2: If a secession referendum is held, will a majority of Albertans vote YES to secede?
6%
Base rate: 0/3 subnational referendums in stable democracies succeeded. Current polling at 30-35%, well below 50%. Quebec with decades of infrastructure only reached 49.4%. Possible upward shift from Liberal re-election (+3%). Final: 6%.
sq3: Will the Alberta UCP government or a successor government pursue a formal independence referendum process within the next 3 years?
55%
Base rate ~2%. Bill 54 and active petition process push significantly upward. Smith committed to process despite opposing secession. Largely tracks SQ1. Final: 55%.
Show full reasoning trace
base rate: Subnational secession referendums are extremely rare in stable democracies (~1-2% in any 3-year window). Of those held, none succeeded (0/3 in modern Western democracies).
evidence updates: The active petition process with Bill 54, approved application, and claimed signatures dramatically increase the probability a referendum is HELD (~55%). However, polling at 30-35% and the historical failure rate of secession referendums keep the YES probability very low (~6%).
combination method: Conditional: P(secede) = P(referendum) × P(YES|referendum). The two probabilities multiply, making the joint event very unlikely despite a reasonable chance of a referendum occurring.
final: Alberta has a realistic ~55% chance of holding a secession referendum by October 2026, but only about a 6% chance of YES winning conditional on it being held. Combined probability of Alberta voting to secede: ~3.5%.
ensemble: {'ensemble_count': 3, 'perspectives': {'base_rate': {'p_yes': 0.035, 'confidence': 0.62}, 'evidence_driven': {'p_yes': 0.08, 'confidence': 0.55}, 'contrarian': {'p_yes': 0.04, 'confidence': 0.78}}, 'spread': 0.045, 'individual_p_yes': {'base_rate': 0.035, 'evidence_driven': 0.08, 'contrarian': 0.04}}
5. Devil's Advocate Sonnet · legacy v2 Market price revealed
Math Error Never Happened Fallacy Missing Info Reasoning Flaw
Challenges
  1. The 55% probability for a referendum being held seems well-reasoned given the evidence, but the 6% conditional probability of YES winning may be too low. The forecaster cites 30-35% current polling but acknowledges there's NO post-April 2025 polling data. A referendum campaign itself could shift opinion significantly — Quebec went from ~40% to 49.4% during its campaign. If Alberta starts at 30-35% with a well-organized campaign and ongoing federal grievances, a shift to 50%+ is unlikely but not as extreme as 6% implies. A more appropriate conditional probability might be 10-15%.
  2. The question asks whether Alberta will 'vote to secede' — but the resolution criteria may hinge on interpretation. The proposed referendum question asks if Alberta should 'cease to be part of Canada,' and the referendum is advisory. Does 'vote to secede' require a YES majority in any referendum, or does it require actual secession? The forecaster should clarify the resolution criteria more carefully. If it merely requires a YES majority vote (even advisory), the probability should be higher than if actual secession is required.
  3. The market at 22% with moderate volume and the note that 'the market appears to price the referendum being held rather than YES winning' is an important observation. But this suggests the market participants may be interpreting the question differently — possibly as 'will there be a secession vote' rather than 'will the vote succeed.' If so, the market price isn't necessarily wrong about the YES probability; it may be answering a different question. The forecaster should consider whether their question interpretation matches the market's.
  4. The forecaster's ensemble spread is very narrow (3.5% to 8%), which seems artificially tight given the enormous uncertainty flagged in the information gaps. With unverified signatures, no post-election polling, unknown federal government response, and unclear resolution criteria, the confidence interval should be much wider. The 0.65 confidence level seems appropriate but the ensemble doesn't reflect the true range of outcomes.
  5. The '0/3 subnational referendums in stable democracies succeeded' base rate is being applied too rigidly. The sample size of 3 is far too small to derive a reliable base rate. Quebec came within 0.6% of succeeding, Scotland reached 45%, and Brexit (while national) did succeed against expectations. With N=3, the appropriate Bayesian posterior is much more uncertain than '~0%' — a Laplace estimate would give roughly 1/5 or 20% chance of success conditional on a referendum being held.
  6. The combination method (P(referendum) × P(YES|referendum) = 0.55 × 0.06 = 0.033) is mathematically correct but the final reported probability is 0.052, which doesn't match this calculation. The ensemble averaging appears to pull it up to 5.2%, but the stated mathematical framework gives 3.3%. This inconsistency should be resolved.
Suggested adjustment: +4pp
Three main reasoning flaws justify upward adjustment: (1) The conditional probability of YES at 6% is too low given the small base rate sample (N=3), the proximity of Quebec's result, and the absence of current polling data — a more defensible estimate would be 10-15%. (2) The Laplace correction on the base rate of 0/3 referendums succeeding gives ~20%, not ~0%. (3) The mathematical combination (0.55 × 0.06 = 0.033) doesn't match the stated 0.052, suggesting internal inconsistency. Using P(ref)=0.55 × P(YES|ref)=0.12 gives ~0.066, and accounting for uncertainty about resolution criteria and campaign effects, ~0.09 is more defensible. Adjusting upward by ~0.04 from 0.052 to approximately 0.09.
Market comparison: Forecaster: 5.2%, Market: 22%. The large divergence (16.8%) is notable. The market reliability is rated MEDIUM with moderate volume, and the market's wide historical range (3%-71%) suggests thin liquidity and high uncertainty rather than strong consensus. The market may be partially conflating 'will a referendum be held' with 'will YES win,' which would explain the higher price. However, the forecaster's conditional YES probability of 6% is likely too low based on the reasoning flaws identified (tiny base rate sample, no current polling, campaign effects). A portion of the divergence is justified by genuine reasoning errors, but the market at 22% likely overestimates the true probability given that polling at 30-35% makes a YES win genuinely unlikely.
6. Aggregator Opus · final calibrated forecast
10% (45% confidence)
My independent estimate of 5.2% was built on a ~55% chance of a referendum being held and a 6% conditional probability of YES winning. I PARTIALLY ACCEPT the Devil's Advocate's critique that the 6% conditional YES probability is too low: the base rate sample of 0/3 is tiny (Laplace correction suggests ~20%), Quebec reached 49.4% from a similar starting position, and we lack post-April 2025 polling. I adjust the conditional YES probability upward to ~12-15%, giving P(ref)×P(YES|ref) ≈ 0.55×0.13 ≈ 0.07. I ACCEPT the math error flag — my ensemble pulled the number away from the stated framework. I PARTIALLY ACCEPT the market interpretation concern: the market at 22% likely conflates 'will a vote happen' with 'will YES win,' and its wide historical range (3%-71%) with medium volume suggests low reliability. However, I cannot fully dismiss the market signal, so I adjust modestly upward to 0.10. The remaining 12-point gap from the market is justified by: (1) polling at 30-35% is far below the 50% needed, (2) Alberta's independence movement lacks the decades of institutional infrastructure Quebec had, and (3) Premier Smith herself opposes secession, making a government-backed YES campaign unlikely.
7. Entry Decision Opus Trader
Side
NO
Position Size
$800
Contracts
1026
Trader Confidence
45%
Trade Rationale
The forecaster estimates p_yes at 10% versus the market price of 22%, yielding a 12% edge on the NO side. The reasoning is sound: polling shows only 30-35% support for secession (well below the 50% threshold needed), Alberta lacks Quebec-style institutional infrastructure for independence, and Premier Smith herself opposes secession. Even after accepting the Devil's Advocate's upward adjustments, the forecast remains significantly below market price. However, the forecaster confidence is only 0.45, the time horizon is extremely long (1280 days) introducing substantial uncertainty, and the orderbook depth is 0 contracts suggesting potential liquidity issues. The market's 22% likely overweights the probability of a referendum happening and conflates it with YES actually winning.
Allocation Logic
The 12% edge is meaningful but tempered by low forecaster confidence (0.45), very long time horizon, and uncertain liquidity. Sizing below baseline at $800 reflects these offsetting factors — enough to capture the edge but conservative given the risks.
Entry price: $0.78
Current: $0.88
Status: OPEN
P&L: $102.56