The Raid Score.

01

The formula

Each factor runs from 0 to 1 and is published per target. The product is deliberately unforgiving: a token that trades at a deep discount but has no capture path scores zero, and so does a capturable token whose holders are mostly one wallet cluster.

Reach is a fourth published metric. It decides whether a raid is affordable and is used as a gate, but it is not multiplied in.

// every score starts from one ratio
r = (treasury.sol + treasury.liquidTokensSol) ÷ market.capSol

Treasury value counts SOL plus liquid tokens in wallets the agent controls. Market cap is fully diluted, in SOL.

02

Discount

How far below the value of its own treasury the token trades. Zero at or below 0.5×, one at or above 1.5×, linear in between.

Discount = clamp((r − 0.5) ÷ (1.5 − 0.5), 0, 1)
bands: full ≥ 1 · none = 0 · partial otherwise

03

Capturability

Whether a designed mechanism lets a token majority actually move the treasury, and who holds the key. Raider takes the best path available.

Capturability = base(path) × custodyMult × timelockMult
requiresExploit → 0, and a never-touch hit

Timelock

Governance only; other paths use 1.00. A long vote timelock gives the target time to defend.

Why platform custody is a hard zero: on today’s largest agent launchpad, per-coin treasuries sit behind the operator’s own signer service. Holders have no vote and no redemption, so there is nothing a token majority can capture.

04

Cleanliness

Whether the holders and the volume are real. A weighted mean of five linear sub-scores, where 1 is clean, followed by two hard-zero rules.

05

Reach

Whether the stake needed to cross the line can be bought at today’s depth. Published per target and used as a gate. Not multiplied into the Raid Score.

stakeShare     = 0.51 (governance, buyback) · 0.25 (arbitrage)   // assumption, tunable
requiredBuySol = stakeShare × cap × (1 + 0.5 × stakeShare × cap ÷ max(liquidity, 0.1))
capacitySol    = 3 × daily volume
lockFactor     = clamp((tradable% − 55) ÷ (85 − 55), 0, 1)
Reach          = clamp(capacitySol ÷ requiredBuySol, 0, 1) × lockFactor

Thin pools cost more: the slippage term grows with the stake relative to liquidity. A token whose float is mostly locked cannot be reached at any price.

06

Gates

Eligible for a live raid only if every gate passes. A high Raid Score that fails a gate stays on the board as WATCH.

• Capturability ≥ 0.6
• Treasury (SOL + liquid tokens) ≥ 50 SOL
• Reach ≥ 0.5
• No never-touch hit

07

Never touch

Hard rules. Any hit makes the verdict NEVER, whatever the score says, and the simulator refuses the target even when forced.

• Dev-controlled treasuryNothing to capture; you would fund the dev’s exit.
• Launchpad-operated custodyKey held by the launchpad operator; no holder path.
• Living community of ~2,000+ real holdersThe fight costs more than the treasury.
• Capture requires exploiting a bugThat is a hack, not M&A. It ends the project.

08

The prime-target pattern

Six checks that describe the best kind of target: hype gone, treasury intact, rules in a program, nobody home. A target is scored without them; they are published as a count out of six.

1. Launched through an agent launchpad 4–10 weeks ago
2. Treasury 50–200 SOL
3. Market cap below the SOL value
4. Governance or treasury rules live in a program, not a dev wallet
5. Dev stopped posting; agent still on autopilot
6. No single cluster above 20% of supply

09

Verdicts

10

First-raid economics

Whether a raid is worth doing at all. Under about 30 SOL of captured value the first raid is too small to matter, so Raider waits for a bigger target or stacks two.

capturedValue = treasury × (1 − 0.5% fee)
tenderPremium = 35% tendered × 12% premium × cap
netToRaider   = capturedValue − requiredBuy − tenderPremium
clearsBar     = capturedValue − tenderPremium ≥ 30 SOL

netCapturedValue = max(0, netToRaider)
burn             = 70% × netCapturedValue
warChest         = 30% × netCapturedValue

Settlement after a capture: the tender premium is honoured first and the stake Raider bought is paid back. What is left is net captured value, the profit on the raid. 70% of it buys $RAIDER and burns it, and 30% stays as the war chest. A raid that does not profit burns nothing, and a raid that does leaves the vault richer by exactly the war chest. Burning a share of the gross capture instead would burn tens of SOL on a raid that left only a few and leave the vault poorer, so the stake is netted. No dividend, no revenue share. In the calculator below, Net to Raider is that profit: load the 84 vs 41 example and read the burn and war chest rows. See also the worked example on the token page.

11

Worked examples

Four fictional targets that differ in one input at a time. The first is the source document’s case: 84 SOL in the treasury, 41 SOL market cap. Scores come from the engine when it is online.

Scored locally with the published constants.

12

Calculator

Every input the engine reads. Change one and watch which factor moves. For the full instrument, with the opportunity map and the paper-raid simulator, open the Score Lab.

13

Assumptions & limits

Engine assumptions, as served

• Stake share needed: 0.51 for governance and buyback paths, 0.25 for redemption arbitrage (tunable assumption).
• Capacity = 3 × daily volume.
• Economics: 0.5% fee on captured value; 35% of holders tender at a 12% premium.
• Treasury value counts SOL plus liquid tokens in agent-controlled wallets.
• All demo targets are fictional. Scores are model output on synthetic data.