Pump.fun Bonding Curve Tracker Notes From a Live Indexer and Paper Ledger

A Pump.fun bonding curve tracker is only useful when it separates fast tape, paper evidence and wallet mythology before the casino starts calling luck a system.

Live Indexer Freshness And Board Counts

The 2026-06-26 07:31 UTC Pumphouse source pack shows a working scanner, not a sleeping dashboard. /root/FreedomCore_Tools/bin/pumphouse_note_figures.py read the public board at 07:32 UTC and reported the latest indexed event as 0.0 min old, with live=True against a 30 min stale threshold. That matters because delayed analytics are acceptable; stale analytics wearing a live badge are not.

The public Pumphouse board carried 200 fresh tokens in the last 6h: 8 GREEN, 161 AMBER and 31 RED. The read-only trending slice showed 30 wallets in the last 15m. Edge access remained unlocked=False, with a 74 wallet cohort and 74 active wallets in the live DB. The claim rule is deliberately dull: >=2.5x lift on >=2 fresh non-overlapping live windows before any public token claim. That is not degen poetry. That is a stool bolted to the floor.

For feed plumbing, the context pack points to /var/www/memes/api/fresh.json, /var/www/memes/api/trending.json, /var/www/memes/api/signals.json, /var/www/memes/api/copy_demo_public.json and /var/www/memes/api/manifest.json, all refreshed around 2026-06-26T07:32 UTC. The Solana side still depends on fast event intake, the kind of stream model documented in the Solana RPC WebSocket docs, but the public note should judge what the board has actually captured, not what a feed vendor brochure implies.

Paper Ledger Evidence, Not Victory Laps

The memecoin paper demo ledger is still the cleaner evidence surface because it can say ugly things without moving real funds. The paper model in the pack uses GBP 50.0 bankroll, GBP 0.5 default stake and GBP 1.0 momentum lanes for MEDIA_VELO, MEDIA_VELOR, VELO and VELOR. It is marked paper only, delayed and redacted for the public feed.

The lane replay since the GBP 50 reset shows why the board needs scepticism baked into the furniture. BUNDLE sits at GBP 458.49, up GBP 408.49, with 886/1925 entries and an 11.1% win rate across 1925 trades. The best mark is 9797.5%, the worst is -93.8%. That is not a calm distribution. That is a roulette wheel with a PhD thesis taped to it.

VELOR reads GBP 142.91, 316/316 entered, 40.2% wins, best 416.1%, worst -91.4%. VELO reads GBP 102.59, 90/90 entered, 40.0% wins, best 876.4%, worst -98.9%. MEDIA_VELO is smaller but cleaner on paper: GBP 73.16, 31/31 entered, 58.1% wins, best 468.9%, worst -98.9%.

Those figures belong in research, not sales copy. The Pumphouse method is useful only if it keeps the paper ledger visible next to the scanner badges.

Wallet Graph Research And Cabal Filters

The better direction in the 2026-06-25 ledger is entity-first. The source pack records live DB counts of pump_events=2936172, wallet_early=3501771, candidate_scored=598433, token_seen=603697, copy_entry_decisions=938939, copy_lanes=2791 and copy_wallet_tiers=496. Funder intelligence adds wallet_funders=106957, funder_events=186122, funder_scores=71578, with 22485 cross-token reused funders and 498 same-token multi-wallet funders.

That is the useful shape of Pumphouse wallet graph research: identify funder reuse, fresh-child wallets, crew behaviour and launch shape before scoring a token. Pump.fun cabal detection is not a magic blacklist. It is a repeated-contact problem across funder_intel.db, pump_events_live.db, wallet_early, candidate_scored and related scanner tables.

The narrative layer stayed cautious. /root/Pumphouse/reports/narrative_confluence_replay/latest.json covered 522,672 Runner Truth rows, 1,040 runners, 2,390 matched controls, 7,240 recent narrative rows and 717 unique narrative mints. No variant earned KEEP_PAPER. The strongest watch row, NARRATIVE_PLUS_FUNDER, had 199 entries, 3 runners, 2 controls, 1.50 lift and 0.4372 rug rate. A Solana holder scanner can count a room; it still has to ask whether the room is full of the same people in different hats.

Mayhem Research And Curve Survival Scars

The Mayhem work is worth exposing only as research residue. /root/Pumphouse/reports/mayhem_velocity_scan/MAYHEM_VELOCITY_SCAN.md scanned 31,010 tokens and found 20 fast candidates using a local PumpHouse tape rule: first local market cap <= $8k and $100k reached within <=300s. That is a useful survival filter for post-graduation behaviour, especially around PumpSwap-style exits, but it is not proof of a public playbook.

Named examples in the context include Girls, Toad, BEEP, WENDYGELES, AKIRA, FomoPacks, Human, IPO.fun and QPAY SOL. Their value here is diagnostic: drift gates, fill basis, paper trigger marks and runner handling can disagree violently within seconds. The clean public lesson is that bonding-curve speed creates measurement error as much as opportunity.

For broader market notes, Atlas notes and FreedomCore keep separate jurisdictions. Pumphouse stays with token tape, holders, paper ledgers and wallet clusters.

Caveats And Public Boundary

This note does not publish live Solana buying or selling instructions, snipe-bot execution steps, private wallet funding details, private tracker addresses or operator console mechanics. The service spine at /root/FreedomCore_Tools/data/system_spine/latest.json was degraded for observability:not_all_ok with deadman:degraded, so scanner freshness and board figures should not be confused with whole-system perfection.

The paper lanes are not live public outcomes. The narrative replay is not approved for promotion. Mayhem remains research evidence with route scars. Any public claim still needs the stated >=2.5x lift across >=2 fresh non-overlapping live windows before it deserves daylight.