Automated Solana Meme Coin Trading Bot | pump.fun Auto-Trading from $39/mo

Most bots wait for the chart to confirm a move. By then, the entry is gone. Bot 10 watches the raw on-chain buy stream via gRPC and detects explosive launches as they happen — before DexScreener updates, before the chart moves, before anyone else reacts.

• ▸
  3 buys · 3 unique wallets · 30-second windowCoordinated shill farms use the same wallets — the unique-wallet gate filters them out automatically
• ▸
  0.5–10 SOL bonding curve gateToken must have real traction (0.5 SOL floor) but not be already pumped (10 SOL ceiling)
• ▸
  Bypasses the normal scoring queueJump-drained immediately — no 30-second hold, no queue wait. Speed is the edge.
• ▸
  Full safety check still firesHoneypot and rug checks run in parallel — momentum speed without sacrificing safety gates

Coordinated shill farms are the #1 cause of bad entries. X Scout runs a forensic analysis of a token's X/Twitter presence — not just follower counts, but the quality, originality, and authenticity of the community behind it. Claude AI handles narrative analysis. The result: a STRONG / WATCH / RUG_FILTER verdict that feeds directly into the Confluence Engine.

• ▸
  Social Graph (25%) — real community vs. coordinated farmAccount age, follower quality, network diversity
• ▸
  Forensic History (20%) — prior rug involvement detectionPast token associations, shill phrase frequency, wallet patterns
• ▸
  Engagement Authenticity (20%) — organic vs. scriptedRepetitive phrases, bot-like cadence, copy-paste shill detection
• ▸
  Professionalism (10%) — project presentation qualityBio completeness, linked assets, verified presence signals
• ▸
  Narrative + Claude AI (25%) — meme originality & loreClaude Haiku scores community language, creativity, and shill density in real time

When real alpha is circulating, it shows up in multiple groups at once. TG Scout monitors a curated list of Solana alpha Telegram groups 24/7. A single mention means nothing. When 2 or more independent groups mention the same token within 60 seconds — that's a confluence signal. It fires immediately as a priority boost into the scoring engine.

• ▸
  2+ independent groups · 60-second windowCoordinated shills post to the same group — independent groups posting simultaneously is a different signal entirely
• ▸
  Priority signal fires instantlyBypasses the normal scan cycle — confluence tokens jump the queue for immediate scoring
• ▸
  CA extraction from links and raw textParses DexScreener links, Jupiter links, raw contract addresses, and query-param formats automatically
• ▸
  Feeds the Confluence Engine as a weighted inputTG Scout signal adds directly to the token's social score in the 8-pillar board

The best edge in meme coin trading is knowing what smart money is buying before the chart shows it. SolBot's Whale Intelligence engine discovers, scores, and tracks profitable wallets automatically — no manual curation, no static lists that go stale.

• ▸
  Self-growing discovery — runs every 4 hoursScans pump.fun on-chain activity, identifies wallets that bought tokens before graduation, promotes those with 40%+ win rate
• ▸
  Second-degree network mappingWallets that co-buy alongside tracked whales are automatically added — the network compounds itself
• ▸
  0–100 confidence score per walletWin rate (50 pts) + recency decay (25 pts) + trade volume (25 pts). Inactive whales are deprioritized automatically
• ▸
  Live copy-trading + signal feedWhen a tracked whale buys, SolBot copies the trade (if it passes your safety threshold) and sets the WHALE signal for all subscribers
• ▸
  Add your own wallets — up to 15Know a wallet worth watching? Pin it alongside the system list. Your additions run the same confidence scoring.

Card counters don't bet the same amount every hand. When the count is high — when the odds are in their favor — they size up. When it's not, they size down. SolBot applies the same math to every trade. A score-60 entry and a score-85 entry are not the same bet. Kelly treats them differently.

• ▸
  Per-user, per-score-bucket Bayesian modelYour trades train your own model. Tony's win rate doesn't affect your position sizes — and yours don't affect anyone else's
• ▸
  Quarter-Kelly default — conservative by designFull Kelly maximizes long-run growth but creates violent drawdowns. Quarter-Kelly captures the edge with a fraction of the volatility
• ▸
  Self-calibrates from your live trade historyAfter 20+ closed trades per score bucket, Kelly updates its win-rate and win/loss-ratio estimates from your actual results
• ▸
  Auto-recalibrates when you change entry filtersChange your min score threshold and Kelly resets to conservative sizing — your old priors no longer apply to the new population

Price can hold while signals collapse underneath. Negative Divergence fires when Bot 2 detects that live confluence signals are deteriorating while you're still in profit — before the chart reflects the weakness. It's the difference between locking in gains and watching them evaporate.

• ▸
  Score-adaptive partial exit — not a flat percentageHigher conviction entries exit less aggressively, letting winners run longer. Lower conviction entries exit faster to protect gains.
• ▸
  Only fires when you're in profitNegative Divergence is a profit-protection signal — it never triggers a loss exit. That's what the stop-loss is for.
• ▸
  Re-entry allowed immediatelyNo exit cooldown — this is a momentum exit, not a stop-loss. If signals recover, the bot can re-enter the same token.