How It Works

The model starts from the historical tribunal success rate for the selected claim type, then adjusts up or down based on three case-specific factors you assess: jurisdiction, legal merits, and evidence quality.

Weights control how much each factor influences the overall prediction. They must total 100%. Adjusting one weight automatically redistributes the others.

The result is only as good as the assessments you put in. Two experienced practitioners may reasonably differ on factor scores — the tool makes that disagreement visible and structured rather than hidden in a gut feel.

📚 Mathematical Foundation: Weighted Linear Scoring Model

This tool uses a weighted linear scoring model — additive in structure, sharing the form of the General Linear Model (GLM), but with an important distinction: the factor weights ($w_i$) are set by the practitioner, not estimated from a regression dataset. The historical baseline $H$ is an empirically-derived prior drawn from HMCTS ET hearing success data disaggregated by claim type; the remaining terms adjust it based on case-specific professional assessment.

This model: $H$ is the empirically-grounded intercept — the historical hearing success rate for the selected claim type, derived from HMCTS ET_3 tables. The three factor terms adjust upward or downward from that baseline. Each factor score is centred at 50, so a neutral assessment (50) contributes zero adjustment; scores above 50 improve the estimate, scores below 50 reduce it.

📚 Mathematical Foundation: Monte Carlo & PERT

This tool models litigation uncertainty using Monte Carlo simulation. It performs thousands of trials, sampling from probability distributions for each head of loss to generate a range of potential outcomes.

By aggregating these distributions (e.g., Unfair Dismissal + Vento + Pension), the model reveals the probability density of the total award, helping to identify the most likely settlement range.

For each factor, set your baseline assessment (your best estimate), then define the low (pessimistic) and high (optimistic) plausible values. The tornado diagram shows the swing each factor produces.

How It Works

The tornado diagram tests each factor one at a time. It holds all other factors at their baseline values, then swings one factor from its low to its high estimate. The resulting change in the overall prediction is plotted as a horizontal bar.

Red bars show the downside swing (baseline to low). Green bars show the upside swing (baseline to high). Longer bars mean that factor has more leverage over the final result.

Use this to prioritise case preparation: if evidence quality produces the widest swing, that's where to focus resources. If limitation is narrow, it's already secure.

📚 Mathematical Foundation: One-Way Sensitivity Analysis

Sensitivity analysis is a core technique in decision analysis and operations research. The tornado diagram implements one-way sensitivity analysis: each input variable is varied independently while all others are held constant (ceteris paribus).

The percentages are empirical priors — historical claimant success rates at full merits hearings, derived from HMCTS ET data (4-quarter average, 2019/20). Jurisdiction-specific data has not been published since ECM Reform migration in late 2022. Treat these as a starting point for analysis, not as current outcome predictions. Items marked ⚠ have small sample sizes or strong selection effects.

Your current best estimates for each factor (these are your baseline — the tool will calculate what they'd need to be).

The amount offered to settle. The tool calculates the minimum success probability needed to justify rejecting this offer.

Your best estimate of the total tribunal award (including all heads of loss) if the claim succeeds. Use the Settlement Range Modeller median if available.

Total estimated costs of proceeding to hearing. Enter 0 if publicly funded or no representation costs.

How It Works

The break-even probability is the point where the expected value of trial equals the settlement offer. Below this probability, accepting the offer is rational. Above it, trial has a higher expected value.

The factor thresholds reverse-engineer this: for each factor, what score would it need to reach (holding the others at your current assessments) to push the overall probability above break-even?

If a required threshold is above 100% or below 0%, it means that factor alone cannot push the result past break-even — it depends on the other factors too.

📚 Mathematical Foundation: Expected Value & Threshold Inversion

The tool finds the indifference probability $p^*$ where the expected value of litigation equals the certain settlement value, then inverts the prediction model to find factor-level thresholds.