Trail Heat Methodology 2026: How FarmDash Scores DeFi Opportunities

TLDR: Trail Heat is FarmDash's 0–100 research-prioritization heuristic, not an airdrop forecast, payout model, or automatic allocation instruction. The live quantitative score has four components: 40 raw points TVL scale, 25 seven-day momentum, 15 chain diversification, and 10 category outperformance. The 90-point raw maximum is normalized to 0–100. Editorial notes are metadata with 0 quantitative weight.

What Is Trail Heat?

Trail Heat compresses several observable protocol-market signals into one comparable number.

Its job is narrow:

Help an operator or agent decide which protocols deserve deeper investigation first.

It is not designed to answer:

• whether a token will launch;
• whether a wallet is eligible;
• how large an allocation will be;
• when a snapshot will occur;
• whether a trade will be profitable;
• whether an agent has authority to transact.

Those questions require different evidence.

What Is the Current Quantitative Formula?

The canonical live methodology is:

Component	Raw points	Effective share after normalization	Evidence
TVL scale	40 / 90	~44.4%	Current calibrated DeFiLlama TVL
Seven-day momentum	25 / 90	~27.8%	Observed seven-day TVL change when history is available
Chain diversification	15 / 90	~16.7%	Current DeFiLlama chain distribution
Category outperformance	10 / 90	~11.1%	Protocol TVL versus live category baseline
Editorial notes	0	0%	Descriptive metadata only

The headline calculation is:

[ ext{Trail Heat} = operatorname{round} left( rac{ ext{TVL} + ext{Momentum} + ext{Diversification} + ext{Category}}{90} imes 100

ight) ]

bounded to the published 0–100 range.

This is why “40/25/15/10” should not be described as percentages. They are raw point maxima out of 90. Their effective normalized shares are approximately 44.4%, 27.8%, 16.7%, and 11.1%.

Component 1: TVL Scale — 40 Raw Points

TVL provides a size/liquidity context.

FarmDash uses a calibrated logarithmic curve rather than a simple linear ranking. The current machine methodology describes roughly:

• around $1M TVL as approximately 10/40 on the calibrated curve;
• $1B+ as reaching the upper end of the 40-point component.

Why logarithmic scaling?

Because the economic difference between $1M and $10M is usually more meaningful for protocol maturity than the difference between $1.001B and $1.010B. A logarithmic curve prevents the largest protocols from consuming the entire score simply because they are large.

What TVL does not prove

High TVL does not prove:

• strong token economics;
• a current reward program;
• sustainable revenue;
• deep exit liquidity for every asset;
• a future airdrop;
• low smart-contract risk.

TVL is one input.

Component 2: Seven-Day Momentum — 25 Raw Points

Momentum measures change, not size.

A protocol can have modest TVL but strong recent inflows, or huge TVL with flat/negative growth.

FarmDash uses observed seven-day TVL movement when the required historical data is available.

When the source history is unavailable, the system should preserve that uncertainty through evidence/confidence state rather than inventing a worst-case zero or a fake live observation.

Why momentum matters

Momentum can identify:

• accelerating deposits;
• capital rotation;
• newly growing markets;
• declining protocols that still look large on an absolute TVL basis.

It can also be noisy.

A temporary liquidity campaign, whale deposit, token-price move, or accounting change can move TVL without proving durable demand.

Component 3: Chain Diversification — 15 Raw Points

This component uses DeFiLlama's current chain-level TVL distribution where available.

Diversification can matter because deployment across several ecosystems may reduce dependence on one chain or expand addressable liquidity.

But “more chains” is not automatically safer.

A multi-chain protocol can add:

• bridge dependencies;
• fragmented liquidity;
• more upgrade/admin surfaces;
• cross-chain messaging risk;
• inconsistent product maturity.

FarmDash therefore treats diversification as a bounded component, not a quality verdict.

For live single-chain protocols, the methodology uses a neutral baseline rather than automatically treating one-chain deployment as zero-quality.

Component 4: Category Outperformance — 10 Raw Points

Protocols should be compared with economically similar peers.

A $100M lending protocol and a $100M niche derivatives protocol may occupy very different competitive positions.

The category component compares protocol TVL with live category baselines for supported sectors and aliases.

This asks:

How strong is this protocol relative to the market structure it actually competes in?

It does not ask:

“Which category deserves an airdrop?”

Editorial Notes Are 0% of the Quantitative Score

This separation is deliberate.

FarmDash can maintain:

• campaign notes;
• research commentary;
• risk flags;
• status labels;
• referral disclosures;
• first-party source links.

Those fields can be useful, but they are not allowed to silently alter the live quantitative Trail Heat headline.

An agent should be able to distinguish:

observed input → derived score → editorial context

rather than receiving one blended number whose provenance cannot be audited.

What Are Data-Evidence States?

A score is more useful when the caller knows where each number came from.

FarmDash's machine contracts distinguish evidence states such as:

• observed/live;
• derived calculation;
• cached/stale where a previously validated observation is reused;
• unavailable;
• editorial.

The exact vocabulary is published by the current machine contract.

For agent use, the rule is simple:

Preserve evidence state and freshness when repeating a score.

Do not copy “Trail Heat 84” into another system and discard the timestamp, source, caveats, and component evidence.

What Happens When Data Is Missing?

Missing data should reduce certainty.

It should not produce a synthetic live value merely because a dashboard looks cleaner with a number.

Examples:

• missing historical TVL → momentum evidence is incomplete;
• no mapped live source → do not invent current telemetry;
• stale cache → retain age/stale label;
• upstream provider error → surface the typed degraded state.

An autonomous system should become more conservative as evidence quality falls.

Trail Heat vs. Static Catalog Scores

FarmDash has historically exposed both live Trail Heat and lighter static/catalog discovery heuristics.

They are not interchangeable.

The canonical quantitative Trail Heat surface is the live Trail Heat API. Static tracker/catalog pages can use a deterministic catalog fallback for discovery when live enrichment is not available, but that fallback must remain labeled separately.

An agent should not compare a static editorial/catalog score with a live quantitative Trail Heat value as if both were produced by the same formula.

How Should You Interpret a High Score?

A high score means:

Under the current quantitative inputs and methodology, this protocol ranks strongly enough to deserve more research.

The next questions should be:

• Is the underlying data fresh?
• Is the protocol's campaign status actually documented?
• What is the smart-contract/oracle/liquidity risk?
• What does my wallet already hold there?
• What are fees, slippage, funding, or borrow costs?
• What is the exit path?
• Does the action make sense without a speculative future reward?
• Is the relevant execution capability live and authorized?

Trail Heat prioritizes the investigation. It does not replace it.

How Should an Agent Use Trail Heat?

A robust agent loop is:

1. call the live status contract;
2. fetch Trail Heat;
3. retain component scores, evidence state, timestamp, and confidence;
4. filter by user policy and chain constraints;
5. inspect first-party campaign evidence;
6. inspect wallet exposure;
7. estimate all-in costs and risk;
8. return INVESTIGATE, WAIT, BLOCKED, or READY FOR APPROVAL;
9. never convert score confidence into wallet authority.

A score of 90 does not authorize a transaction any more than a weather forecast authorizes an airplane to take off.

Why Does the Method Use TVL at All?

TVL is useful because it can provide context for:

• market size;
• capital depth;
• adoption;
• protocol scale;
• category position.

But FarmDash does not claim TVL is “the strongest historical predictor of airdrop dollar value.”

That would require a validated statistical model over a representative historical population, which Trail Heat is not.

The current contract explicitly labels the model a heuristic not statistically validated.

Why Does the Model Exclude Editorial Bonuses?

Because editorial conviction and quantitative market state answer different questions.

If a researcher believes a protocol is strategically important, that can be written down and sourced. It should not be disguised as ten extra quantitative points.

This makes the score easier to:

• audit;
• reproduce;
• expose to external agents;
• test for drift;
• compare over time.

Frequently Asked Questions

Is Trail Heat a machine-learning prediction?

No. It is a deterministic quantitative heuristic based on documented components.

Does a high Trail Heat score guarantee an airdrop?

No. It does not prove a token, snapshot, eligibility rule, allocation, price, or profit.

Does editorial research affect the headline score?

Not in the current quantitative methodology. Editorial fields are separate metadata with zero quantitative weight.

How often does Trail Heat update?

The live score updates when its upstream data is refreshed and available. Consumers should inspect the returned timestamp/freshness fields rather than assuming a fixed schedule from this guide.

Can I access Trail Heat through the API?

Yes. The canonical live quantitative surface is:

GET /api/v1/trail-heat

Use the current OpenAPI contract for parameters and response schema.

Does Trail Heat work for Solana protocols?

The methodology itself is chain-agnostic where the required data inputs are available. Chain-specific execution support is a separate capability and must be checked through live status.

Source of Truth

For operational use, prefer:

• the live Trail Heat API response;
• lib/api/agent/data-quality.ts methodology metadata;
• src/config/capabilityRegistry.ts capability/evidence definitions;
• the current OpenAPI/MCP contracts;
• the current Methodology page.

If an older guide, generated page, or cached AI answer disagrees with those sources, the current machine contract wins.

---

Trail Heat helps decide what deserves investigation. It does not decide what deserves your money.