Methodology

How a chain-fit score is produced

Routefold’s scores are computed by a deterministic engine before any language model is involved. This page documents every category, every sub-factor, every hard constraint, and the exact limits placed on model influence.

Scoring engine v1.0.0Knowledge base v1.0.0Reviewed 2026-07-01

Three rules

The model does not choose the number

A deterministic function maps the Digital Twin and the chain knowledge base to a 0–100 score. It is pure: the same inputs always produce the same result, with no randomness and no model call. Every point is attributable to a named sub-factor with a written reason.

Model influence is bounded and visible

The model explains the score, identifies advantages, trade-offs and unknowns, and may propose an adjustment. That adjustment is clamped to ±5 points regardless of what it returns, requires a written justification to apply at all, is stored in a separate column, and is displayed next to the base score everywhere a score appears.

Missing data lowers confidence rather than being guessed

When a factor has no input, it scores on a documented neutral default and is flagged. Each flag lowers the confidence value shown beside the score, and the specific gap is listed in the report. Routefold does not fill a gap with a plausible-looking number.

The 100 points

Five categories with a fixed base allocation. Your selected objectives tilt these weights, and the result is renormalised back to exactly 100 so the scale never inflates.

Product–ecosystem fit

30base points

Whether this ecosystem is a natural home for this kind of product: does its existing activity, audience orientation and regional strength match what the product does?

Category suitability14 pts
How well established this product's category already is on the chain, using the knowledge base's per-category suitability band.
Audience orientation8 pts
Alignment between the product's consumer / institutional / developer orientation and the chain's centre of gravity.
Geographic alignment4 pts
Overlap between the product's target regions and the chain's regional strength. Neutral when no target region is set.
Stage & ecosystem support4 pts
Earlier-stage products gain more from ecosystems with active grant and business-development support; mature products gain less.

Users and liquidity

25base points

Whether the users and on-chain capital the product needs are actually present, at the depth the product requires.

Liquidity depth10 pts
DeFi liquidity depth measured against whether the product actually needs it. Products that do not need deep liquidity are not penalised for shallow markets.
Stablecoin availability7 pts
Stablecoin depth measured against the product's stated stablecoin dependency.
User base reach8 pts
Size of the chain's user base for the specific audience the product targets — retail, institutional, or developer.

Technical compatibility

20base points

How much of the existing codebase, tooling, audit surface and security posture carries over without a rewrite.

Virtual-machine compatibility9 pts
Whether existing contracts can deploy as-is, need adaptation, or need a full rewrite in another language.
Language & tooling reuse3 pts
Overlap between the languages the team already writes and the languages the chain accepts.
Finality match3 pts
Whether the chain's finality profile satisfies the product's stated settlement requirement.
Security-model fit5 pts
Whether the chain's trust assumptions are acceptable given how much value the product puts at risk.

Cost and operational fit

15base points

Whether per-transaction economics work for this product and whether the team can realistically operate here.

Transaction cost fit7 pts
Per-transaction cost band weighed against the product's transaction profile and cost sensitivity.
Operational capacity5 pts
Whether the operational burden of running here is realistic for the stated team size and budget.
Tooling maturity3 pts
Developer-tooling maturity, weighted more heavily when the delivery time horizon is short.

Strategic optionality

10base points

What this deployment opens up beyond itself: ecosystem support, portfolio diversification, and onward reach to other chains.

Ecosystem support3 pts
Availability of grants, business development and go-to-market support.
Portfolio diversification3 pts
How much this deployment reduces concentration in the family the product is already exposed to.
Interoperability reach4 pts
Maturity of messaging and bridging infrastructure, which determines how easily the next expansion follows this one.

Objective weighting

Each objective you select applies a tilt to one or more categories. Your primary objective counts double. The combined tilt is divided by the total objective weight, so selecting many objectives produces a balanced profile rather than a compounded distortion. The tilted allocations are then renormalised to sum to 100.

Worked example

A product selecting lower transaction costs as its primary objective and user growth as a secondary one moves points toward cost and operational fit and toward users and liquidity, and away from the categories neither objective touches. Technical compatibility keeps its base allocation in absolute terms but represents a smaller share of a fixed 100 points.

The exact weights used are recorded in every report’s factor table, under the deterministic score row.

Hard constraints and penalties

Some conditions are not a matter of degree. A hard blocker zeroes the score outright; a penalty subtracts from the weighted total.

Hard blockers — score forced to zero

Excluded by youAn ecosystem you explicitly excluded in the constraints step is never recommended, regardless of how it would otherwise score.
Virtual-machine incompatibleWhen you set a hard VM requirement, chains that cannot satisfy it are blocked. A chain offering a compatible deployment path — for example Solidity support without being natively EVM — is not blocked, but scores lower on compatibility.
Not a deployment targetA data-availability layer is not somewhere application contracts deploy. It is blocked for every product category except infrastructure, with the reason stated.

Penalties — points subtracted

Security model below requirement−8 points. Critical security sensitivity against a sidechain or shared-security trust model.
Rewrite exceeds horizon−9 points. A time horizon measured in weeks against a target requiring a full contract rewrite.
Operational burden above capacity−7 points. High or very high operational complexity against a solo or small team.
Cost above budget−6 points. Minimal budget sensitivity against moderate or high transaction costs.
Liquidity below requirement−5 points. The product depends on deep liquidity and the ecosystem does not have it.
Insufficient data for value at risk−4 points. Very high value at risk against low-confidence ecosystem data.

Confidence

Confidence is a separate number from the score. A chain can score highly with low confidence — that combination means the recommendation rests on assumptions you should verify before acting.

It is derived from three inputs: how confident the knowledge-base record for that chain is, how many factors had to fall back to a neutral default, and how confident the Digital Twin extraction itself was. The specific gaps are listed under each chain in the scorecard rather than summarised into the number alone.

A report’s overall confidence is the score-weighted mean of its top candidates, so a weak long tail does not drag down a well-supported recommendation.

The chain knowledge base

Almost every field is a categorical band rather than an exact figure. This is deliberate. Throughput, fee and liquidity numbers move constantly, and any figure pinned into a knowledge base would be wrong within weeks while continuing to look authoritative. Bands are defensible and honest about their own precision. Where a live figure genuinely helps, it is fetched from a public source at runtime and always displayed with its source, its timestamp, and whether it is live, cached or seeded.

Chain	VM	Finality	Cost	Security model	Stablecoins	Confidence
Ethereum	EVM	Under a minute	High	Layer-1 consensus	Very high	high
Arbitrum One	EVM	Optimistic challenge period	Very low	Rollup with fraud proofs	Very high	high
Base	EVM	Optimistic challenge period	Very low	Rollup with fraud proofs	Very high	high
OP Mainnet	EVM	Optimistic challenge period	Very low	Rollup with fraud proofs	High	high
Polygon PoS	EVM	Seconds	Negligible	Sidechain validator set	High	high
Avalanche C-Chain	EVM	Sub-second	Low	Independent validator set	High	high
BNB Smart Chain	EVM	Seconds	Very low	Independent validator set	Very high	medium
Solana	SVM	Seconds	Negligible	Independent validator set	Very high	high
Sui	MoveVM	Sub-second	Negligible	Independent validator set	Moderate	medium
Aptos	MoveVM	Sub-second	Negligible	Independent validator set	Moderate	medium
NEAR Protocol	NEAR-VM	Seconds	Negligible	Independent validator set	Low	medium
Celestia	DA-layer	Seconds	Very low	Shared / delegated security	Very low	medium
Cosmos Ecosystem	CosmWasm	Seconds	Very low	Independent validator set	Low	medium
Scroll	EVM	Minutes	Very low	Rollup with validity proofs	Low	medium
Linea	EVM	Minutes	Very low	Rollup with validity proofs	Moderate	medium

The analysis pipeline

Seven stages, each with its own schema and its own validation. A stage that returns something invalid is retried with the validation error fed back; if it still fails, the analysis reports which stage failed rather than persisting malformed output.

01
Project extraction
Retrieved page text plus your answers → a validated factual profile.
02
Digital Twin
Profile plus constraints → the structured model everything else reads from.
03
Chain interpretation
Twin plus deterministic scores → explanation, advantages, trade-offs, unknowns.
04
Expansion sequence
Ranked scores → primary, secondary, ruled out, and a rollout order.
05
Architecture brief
Deployment model, components, messaging, token model, monitoring.
06
Risk register
Security, liquidity, operational, governance, UX and compliance risks.
07
Execution plan
Four weeks with milestones, tasks, dependencies and acceptance criteria.
08
Finalisation
Executive summary, technical brief, and the sources and assumptions record.

What this does not do

Routefold reads what you give it and what it can retrieve from public URLs. It does not read your codebase, your analytics, your counterparty list or your cap table. It cannot know how many of your users already hold assets on a given chain, and it does not pretend to.

It is not a substitute for a professional smart-contract audit, and it does not provide financial, legal, compliance or investment advice. Compliance items in every report are framed as questions to put to qualified counsel.

The scores are a structured argument, not an answer. The reason the entire factor table is exposed is so you can disagree with a specific line rather than with the number.

See it applied to an example Security posture