No two disputes are alike, but the questions worth answering before continuing are remarkably consistent: what are the odds, what will it cost, and what is a rational outcome? Our method exists to answer those questions with discipline — never to manufacture certainty where none exists.
Judgment first, evidence underneath
A Tochelet assessment is led by experienced lawyers and mediators. They frame the matter, weigh each issue, and own the conclusion. Beneath that judgment sits a proprietary analytical model that organises evidence and surfaces patterns from decades of resolved matters. People lead; the model informs. Neither replaces the other.
This order matters. A model can show how comparable disputes have moved; it cannot read a contract for tone, sense the credibility of a witness, or judge how a particular forum is likely to treat a particular argument. Equally, seasoned counsel benefit from a structured, dispassionate reference point that is not swayed by the optimism of the party who is paying. The two together produce something neither delivers alone.
Counsel frame the matter
Senior lawyers and mediators read the claims, the documents and the procedural posture, and decide which questions actually move the outcome.
The model organises evidence
Inputs are structured and compared against patterns drawn from decades of resolved matters — likely cost, likely time, and realistic settlement ranges.
Counsel interpret and decide
The model proposes; experienced lawyers weigh its signals against professional judgment and reach the conclusion. No assessment is issued on output alone.
Transparent, within limits
We explain how we got there and what would change it. We do not predict court outcomes with certainty, and we do not expose proprietary scoring logic.
The proprietary model
Behind every assessment is a proprietary analytical model, built and refined over decades of private, confidential resolved-matter data. Its purpose is narrow and concrete: to surface patterns in how comparable disputes have actually moved — the costs they accumulated, the time they took, and the ranges within which they settled. It is an instrument for rigor, governed at every step by counsel.
What it does
The model gives counsel a structured reference point rather than a verdict. In practice, that means it helps answer questions such as:
- How have disputes with a comparable shape — issue, stage, evidence profile and amount — tended to resolve?
- What ranges of cost and time have similar matters typically required to continue?
- Where, historically, has the realistic settlement range fallen relative to the amount claimed?
These signals are inputs to a lawyer's judgment, not substitutes for it. They make an assessment more grounded and less prone to wishful thinking — nothing more, and nothing less.
What we will not do
We are transparent about what the model does and deliberately reticent about how it scores. The internal weighting, features and scoring logic are proprietary, and we do not publish or expose them — disclosing them would neither help a party decide nor serve the confidentiality the work depends on. We would rather be clear about that boundary than imply a false openness.
The model does not predict court outcomes with certainty. Litigation is inherently uncertain; the model describes patterns and probabilities, not destiny. Any number it informs is a considered estimate, weighed by counsel — never a guarantee.
How the data is protected
The data that informs the model is confidential and stays that way. Confidential matter data is never exposed, never sold, and never used to train anything outside Tochelet. Your matter is reviewed in confidence and does not become product. Our security and data-handling practices are set out in full on the Security & confidentiality page.
What we weigh — input factors
Every assessment is evidence-aware: it works from the specifics of your matter, not from generic assumptions. Counsel and the model consider the same families of input. The table below sets out what each factor covers — the inputs themselves, not the weighting we apply to them.
| Factor | What it covers |
|---|---|
| Dispute amount | The claimed amount, counterclaim exposure that runs the other way, and the possibility of non-monetary outcomes such as injunctions, declarations or specific performance. |
| Stage of process | Where the matter stands today — pre-litigation, demand letter, mediation, statement of claim, discovery, trial preparation or appeal — and how much remains to be spent. |
| Evidence strength | Contracts, correspondence, invoices, admissions, expert reports and witnesses — and, just as telling, the evidence that is missing or contested. |
| Legal & procedural risk | Limitation and timeliness, burden of proof, the forum, uncertainty in precedent and the procedural complexity that can shift an outcome regardless of the merits. |
| Cost to continue | Legal fees, experts, filing and court fees, internal and management time, the cost of delay, and the risk and expense of enforcement after judgment. |
| Settlement leverage | Urgency on each side, reputational risk, cash constraints, the value of a continuing relationship, and the negotiating posture the parties bring to the table. |
From inputs to expected value
These factors do not stay as a list. Counsel, informed by the model, resolve them into four connected views of the matter:
- A risk profile — a candid, issue-by-issue read of strength and uncertainty, rather than a single optimistic headline.
- A cost-to-continue range — what continuing is realistically likely to cost in fees, time and delay before any resolution.
- An expected-value range — the likely outcome of continuing, weighed by probability and net of the cost and time to get there.
- A rational settlement range — the band within which agreement is the economically sound choice, given all of the above.
When the rational settlement range sits above the expected value of continuing, settlement is the disciplined choice — even for a party who believes they are legally right. The visual below illustrates the relationship; it is a depiction of the method, not a forecast of any particular matter.