Estimate decision reversal probability

What it's for

Shows executives whether a decision is genuinely robust, what new evidence is likely to change it, and whether buying that evidence is worth its cost before commitment.

What you give it

Inputs split into evidence read from your connected systems, calibration your team owns, and numerical controls that affect precision but never the result's meaning.

{
  "alternatives": [
    {
      "future_signal_standard_error": 8,
      "id": "build-platform",
      "prior_value_mean": 100,
      "prior_value_standard_deviation": 30
    },
    {
      "future_signal_standard_error": 8,
      "id": "buy-platform",
      "prior_value_mean": 99,
      "prior_value_standard_deviation": 30
    }
  ],
  "correlations": [
    {
      "alternative_a": "build-platform",
      "alternative_b": "buy-platform",
      "correlation": 0.35,
      "id": "shared-demand"
    }
  ],
  "current_decision_id": "build-platform",
  "future_evidence_cost": 2,
  "maximum_reversal_probability": 0.2,
  "seed": 73,
  "simulation_draws": 5000
}

What you get back

This is the actual output of running the example above — computed by the same function the platform calls, not an illustration.

{
  "alternative_diagnostics": [
    {
      "alternative_id": "build-platform",
      "posterior_value_standard_deviation_after_evidence": 7.69695557,
      "posterior_variance_reduction_fraction": 0.9342,
      "prior_value_mean": 100,
      "prior_value_standard_deviation": 30,
      "probability_selected_after_future_evidence": 0.52,
      "probability_truly_best_under_prior": 0.51
    },
    {
      "alternative_id": "buy-platform",
      "posterior_value_standard_deviation_after_evidence": 7.69695557,
      "posterior_variance_reduction_fraction": 0.9342,
      "prior_value_mean": 99,
      "prior_value_standard_deviation": 30,
      "probability_selected_after_future_evidence": 0.48,
      "probability_truly_best_under_prior": 0.49
    }
  ],
  "assumptions": [
    "Alternative values follow the submitted joint multivariate-normal prior and future signals are unbiased normal measurements with declared standard errors, independent conditional on true values; all values, thresholds, and evidence costs share one horizon and unit.",
    "The current decision is supplied explicitly and audited against the declared prior decision rule; reversal means the governed posterior-mean choice changes after the planned evidence, not that a manager was wrong or that the future signal is causal.",
    "Expected evidence value is model-conditional and subtracts only the declared evidence cost; delay, option loss, implementation, safety, and organizational costs must already be represented or separately governed.",
    "Alternative IDs must represent aggregate actions, projects, products, or strategies, never named employees; decision fragility is not an automatic instruction to hire, fire, fund, stop, or investigate anyone."
  ],
  "configuration": {
    "future_evidence_cost": 2,
    "maximum_reversal_probability": 0.2,
    "minimum_action_value": 0,
    "seed": 73,
    "simulation_confidence_level": 0.95,
    "simulation_draws": 5000
  },
  "decision": "current_decision_is_fragile_to_planned_evidence",
  "method": "multivariate_normal_preposterior_decision_reversal_v1",
  "summary": {
    "current_decision_aligned_with_prior_rule": true,
    "current_decision_id": "build-platform",
    "decision_reversal_probability_after_evidence": 0.48,
    "expected_regret_of_current_decision": 13.1556,
    "gross_expected_value_of_planned_evidence": 12.4406,
    "net_expected_value_of_planned_evidence": 10.4406,

How it works

Sequential Bayesian & bandits — Learn while deciding — update beliefs as evidence arrives and choose where the next unit of effort is worth spending.

1. 1 Freeze aggregate alternatives, the explicit current choice, comparable prior value means and uncertainty, planned-signal precision, cross-alternative correlations, the no-action threshold, and the full cost of obtaining and waiting for evidence.
2. 2 Draw coherent latent values and future signals from the declared joint prior, update all alternatives with the multivariate-normal posterior, and apply the same governed posterior-mean decision rule on every draw.
3. 3 Report reversal probability with Monte Carlo precision, switch destinations, posterior variance reduction, current-choice regret, and gross and net preposterior evidence value; flag a current choice that does not match the declared prior rule before interpreting fragility.

Before you trust it

Every tool in the catalog ships with the conditions under which its answer is meaningful — and the conditions under which it should abstain instead of guessing.

Calibrating it to your org

Calibration workflow

1. 1 Define the management decision, target outcome, aggregate unit, privacy boundary, cadence, and prediction/intervention horizon for this organization.
2. 2 Build a tenant-scoped historical cohort using only information available before each prediction or decision; preserve zero periods, censoring, assignment probabilities, and unresolved outcomes when the method requires them.
3. 3 Estimate statistical parameters on training history, but obtain costs, utilities, risk tolerance, practical-effect thresholds, capacity, and policy constraints from accountable decision owners.
4. 4 Validate on later time windows or held-out aggregate units at the deployment grain, against a simple baseline and the function-specific validation strategy.
5. 5 Deploy only if the returned decision clears evidence, overlap, calibration, robustness, and guardrail checks; warning, unsupported, schema-gap, and fallback decisions are abstentions.
6. 6 Monitor realized outcomes, data drift, coverage, and decision regret; recalibrate at a governed cadence or after a detected regime/definition change, never merely because a stakeholder dislikes the result.

Call it from your AI

You don't wire up 388 tools in your MCP client. The GitRevio MCP server exposes 18 tools, three of which let an agent search the catalog, read a tool's schema, and run it — so the assistant finds this one on its own.

gitrevio_capabilities_search
  { "q": "estimate how often planned evidence would" }
  → finds "estimate_decision_reversal_probability"

gitrevio_capability_describe
  { "capability_id": "estimate_decision_reversal_probability" }
  → returns the input schema and agent guidance shown on this page

gitrevio_capability_run
  { "capability_id": "estimate_decision_reversal_probability", "arguments": { ... } }
  → returns the result shown above

Works in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.

Related tools

See every tool in Risk, tails & resilience →