Methodology

This page is the public definition of every label MadeOnSol puts on a wallet, a deployer or a token. The numbers are the ones the code enforces — a build-time test fails when the two drift apart — and every change bumps the version below. Labels are descriptive, not advice: a tier says what a wallet has done on-chain in our window, nothing about what it will do.

Coverage and lineage

Solana mainnet, observed through our own gRPC capture (two regions). Launch-pipeline datasets (deployers, bonds, snipers, almost-bonded, surges) are pump.fun first, with bonk/LetsBonk LaunchLab where stated; every launch-pipeline API response carries a coverage object naming exactly which launchpads and programs are and are not covered. Robinhood Chain (Arbitrum Orbit L2, chain 4663) is observed through our own node. Trade history starts 2026-04-12; deployer snapshots 2026-04-07; cross-source wallet-flag snapshots 2026-08-28.

Bonds (graduations)

A pump.fun bond is credited only after an economic gate: the curve migrated at least 30 SOL, or — on paths without the migration transaction — the token trades at ≥ $8,000 market cap. LaunchLab bonds use the same $8,000 gate. Bond-time and peak market caps are read from 1-minute candles with at least $1,000of liquidity — never from an API price mark (pump.fun resets a curve's virtual reserves at migration, which made every curve-priced reader report the launch market cap for a few seconds). A bond is instant when create → migrate took ≤ 1 minute — a bundle that filled the curve itself; instant_bond is exposed on bond rows and events.

Deployer tiers

• elite — ≥ 5 deploys, ≥ 5 bonds, lifetime bond rate ≥ 40 %, recent (last 10) bond rate ≥ 50 %.
• good — ≥ 3 deploys, ≥ 3 bonds, lifetime ≥ 25 %, recent ≥ 30 %.
• moderate — ≥ 3 deploys, ≥ 1 bond, lifetime ≥ 15 %, recent ≥ 15 % (not tracked for alerts).
• rising — 1–3 deploys with a 100 % resolved bond rate (a still-pending launch is excluded), and at least one non-instant bond or exactly two bonds.
• cold — ≥ 10 recent outcomes with a 0 % recent bond rate over ≥ 3 deploys.
• unranked — everything else, including any wallet whose bonds are all instant (elite/good/moderate require at least one earned bond).

Tiers are recomputed on every bond and by a daily pass. runner_rate= share of a deployer's labeled tokens that were runners (peak market cap reached ≥ 60 minutes after deploy); gate on labeled_tokens ≥ 3.

Dump / runner outcomes

dump: peak market cap reached < 15 minutes after deploy. runner: peak market cap reached ≥ 60 minutes after deploy. The 15–60 minutes band is excluded. Universe: tracked-deployer tokens with ≥ 15 early buyers, ≥ 24 h old; dump-cluster stats use a trailing 42-day window, runner_rate uses all of a deployer's labeled tokens (gate on labeled_tokens ≥ 3). Dump-cluster membership is a rolling ensemble, not a static list — validated out-of-sample (wallets with ≥ 5 dump cohorts and 0 runner cohorts predict forward dumps with dose-response).

Sniper and bundler wallets

sniper: early buyer (first 20 wallets) on ≥ 5 distinct tokens and sold on ≥ 80 % of them; recomputed daily over the wallet's full history (a wallet can leave the set). bundler: bought two or more different tokens in the same block at least once; append-only (a wallet never leaves the set).

Alpha wallets and bot confidence

Alpha statistics come from early-buyer PnL (first 20 buyers per token). bot_confidence is high when any of: buy-size stddev < 0.001 SOL over ≥ 3 tokens (uniform sizes); bundle rate > 30 %; > 5 tokens per active hour; 100 % win rate over ≥ 10 tokens; > 10 tokens inside < 6 active hours (burner). medium when any of: bundle rate > 15 %; buy-size stddev < 0.05 SOL over ≥ 3 tokens; 3–5 tokens per active hour; ≥ 95 % win rate over ≥ 10 tokens. Otherwise low/none. Leaderboard eligibility: net PnL ≥ 0 SOL, ≥ 1 winning token, ≥ 10 tokens traded (the alpha leaderboard also excludes high/medium bots by default).

Token surges and revivals

Gates for every tier: liquidity ≥ $1,500, liquidity ÷ market cap ≥ 2 %, buy volume ≥ 3 % of the market-cap gain (a move must be paid for), age ≥ 20 s, and the level must hold for 10 s. Tiers: early ≥ 3× baseline and ≥ $12,000 within 10 min; strong ≥ 6× and ≥ $30,000 within 30 min (≥ 2× over the last 3 min); breakout ≥ 8× and ≥ $45,000 within 2 min. revival: ≥ 24 h dormant on the candle tape, then ≥ 5 buys and ≥ $500 of buy volume in 15 min at ≥ $5,000 and ≥ 1.5× the pre-dormancy market cap.

Point-in-time answers

/api/v1/wallet/{address}/flags?as_of= returns the flags that were current at a timestamp (write-on-change snapshots, daily); /api/v1/deployer-hunter/{wallet}/history the deployer tier trajectory; /api/v1/manifests a nightly fingerprint and schema hash per dataset. See the API docs.

Changelog

• 2026.08.28-2 (2026-08-28) — Deployer tiers: a wallet whose bonds are all instant (create → migrate within 1 minute) is excluded from elite/good/moderate; rising requires a non-instant bond or exactly two bonds (migs 355/356).
• 2026.08.28-1 (2026-08-28) — Bond-time market cap is read from the liquidity-gated candle tape (≥ $1,000 liquidity) instead of API price marks; pump.fun's reserve reset at migration no longer produces $3k "bonds" (mig 354). Peak market caps candle-authoritative since mig 351.
• 2026.08.02 (2026-08-02) — Coverage disclosure on every launch-pipeline response (which launchpads and programs are and are not covered).
• 2026.06.08 (2026-06-08) — Deployer runner_rate (validated as-of predictor) and dump-cluster wallet stats (rolling 42-day ensemble) introduced.
• 2026.04.20 (2026-04-20) — Alpha-wallet bot classifier tightened (throughput, perfect-winrate and burner-wallet signals).

Questions or a definition you need stated differently for a review: [email protected].