Pumphouse bonding-curve tracking 2026-06-22: wallet cohort analysis and database freshness

The Pumphouse Pump.fun bonding curve tracker hit the 2026-06-22 07:31 UTC refresh window with indexer freshness reported at 0.0 minutes old, while edge access remained locked for the cohort of 74 wallets active in the live database.

Live indexer state and fresh board distribution

The live database indexer at /root/Pumphouse/pump_events_live.db was verified active and fresh as of 2026-06-22 07:31 UTC. The latest event was recorded at 0.0 minutes old, which satisfies the system safety threshold requiring data updates to be under 30 minutes old. Fresh tokens entering the board within the trailing six-hour window totaled 200, categorized by the grading pipeline as 7 GREEN, 152 AMBER, and 41 RED tokens. In addition, the read-only trending panel active at the 15-minute resolution monitored 30 wallets.

The underlying tracking mechanism connects to the blockchain network using resources outlined in the Solana RPC WebSocket documentation, mapping state transitions as tokens trade along the bonding curves. Details of the curve mathematics are documented in the Pump.fun documentation. While the indexer remains fully operational, edge access remains locked (unlocked=False) for the current wallet set. A strict verification claim rule governs token promotions, requiring a minimum of a 2.5x lift on at least 2 fresh, non-overlapping live windows before any token claim is publicized.

Cohort wallet profiles and graph architecture

The system monitors relationships and group associations to assist with Pump.fun cabal detection. The research maps wallet nodes and transaction links to trace cluster risk, utilizing the Solana holder scanner to audit holder distribution profiles across newly launched contracts. The active database infrastructure relies on multiple files, including /root/Pumphouse/narrative_engine.db, /root/Pumphouse/funder_intel.db, /root/Pumphouse/narrative_media.db, and /root/Pumphouse/pump_exec.db.

Specific API and state outputs generated during the current analysis window include /var/www/memes/api/wallet_finder.json, /var/www/memes/api/confluence.json, /var/www/memes/api/narratives.json, and /var/www/memes/api/wallet_finder_public.json. These documents trace the wallet connections that form cluster risks, providing researchers with evidence of coordinated accumulation.

Paper model replay lanes and performance ledger

Performance metrics are tracked in the memecoin paper demo ledger with a GBP 50.0 bankroll and GBP 0.5 stake. Momentum lanes (MEDIA_VELO, MEDIA_VELOR, VELO, VELOR) utilize a GBP 2.5 stake.

The historical results for each replay lane, measured since the GBP 50 reset, are logged directly in the system configuration:

• BUNDLE_V: GBP 320.73 (+270.73), 1009/1589 entries (12.1% win), best 6371.5%, worst -93.1%.
• BUNDLE_Q50: GBP 309.15 (+259.15), 515/986 entries (11.6% win), best 9797.5%, worst -93.8%.
• BUNDLE: GBP 241.59 (+191.59), 510/1114 entries (11.1% win), best 9797.5%, worst -93.8%.
• VELO: GBP 157.98 (+107.98), 298/298 entries (50.7% win), best 223.9%, worst -91.4%.
• VELOR: GBP 94.29 (+44.29), 208/218 entries (41.3% win), best 344.5%, worst -91.4%.
• CORE_V: GBP 67.1 (+17.1), 359/359 entries (25.9% win), best 957.2%, worst -95.5%.
• MID: GBP 56.51 (+6.51), 280/337 entries (34.8% win), best 721.7%, worst -95.7%.
• BW_SENTINEL_SOLO: GBP 56.28 (+6.28), 299/380 entries (45.9% win), best 2670.0%, worst -92.3%.
• CORE: GBP 51.32 (+1.32), 334/378 entries (27.5% win), best 796.2%, worst -98.2%.
• MEDIA_VELOR: GBP 47.22 (-2.78), 4/4 entries (25.0% win), best 17.5%, worst -91.4%.
• MEDIA_VELO: GBP 46.78 (-3.22), 4/4 entries (0.0% win), best -0.1%, worst -91.4%.
• FRESH: GBP 39.71 (-10.29), 317/417 entries (22.9% win), best 1625.9%, worst -98.7%.
• FRESH_V: GBP 36.92 (-13.08), 693/747 entries (28.5% win), best 1390.8%, worst -76.0%.
• ELITE: GBP 32.73 (-17.27), 363/390 entries (8.5% win), best 575.4%, worst -94.4%.

These results show momentum tracking variance. Velocity-based filters (like VELO) maintain higher win rates in backtests compared to broader basket methods that trade frequency for lower accuracy.

Technical caveats and safe research boundary

Several parameters are not finalized or validated. First, edge access remains locked, meaning no automated claims or live execution strategies are currently active. Replay ledger results are generated purely under simulation conditions; they do not reflect actual trading performance or real-world execution capacity. System alerts and tracking metrics are delayed and redacted on the public feed.

The data presented is intended for analytical research. Outbound context links are available for the Pumphouse platform, the Pumphouse method overview, the broader Atlas notes archive, and the main FreedomCore portal.