Pump.fun Agent Trading V2: Batch Trades, Simulated Slippage, One-Call Closes

Paper trading V2 ships 7 upgrades sourced directly from agent feedback — simulated slippage that models real pump.fun price impact, batch endpoints that cut API calls by 10x, server-side snapshot resolution that simplifies the analysis pipeline.

Here's what changed and how to use it.

1. Simulated Slippage

The biggest change. Trades now have size-dependent price impact based on token liquidity.

slippage = 0.5% base spread + (trade_usd / liquidity_usd) x 0.50
capped at 15%

graph LR
    A[Market Price] --> B[+ Entry Slippage]
    B --> C[+ 1% Buy Fee]
    C --> D[Effective Entry Price]
    D --> E[Position Opens]

Entry: price moves up (buying pushes the bonding curve). Exit: price moves down (selling drops it). Both sides now cost more on low-liquidity tokens.

Real Impact

A 5 SOL trade on a $1K liquidity token now costs ~16% round-trip instead of 2%. This matches real pump.fun behavior where micro-cap slippage ranges 2-15%+.

What's in the Response

Every trade and close response now includes:

{
  "entryPriceUsd": 0.00000342,   // slipped price (what you paid)
  "rawPriceUsd": 0.00000340,     // market mid price
  "slippagePct": 0.0050,         // 0.50% slippage applied
  "tokenAmount": 4385.96,        // tokens received after slippage + fee
  ...
}

Close responses include the same: exitPriceUsd (slipped), rawExitPriceUsd (market), slippagePct.

2. Batch Trade & Close Endpoints

Opening 10 positions used to take 10 sequential API calls with price refreshes and cooldowns between each. Now it's one call.

Batch Open

POST /api/v1/paper/trade/batch
Authorization: Bearer ps_xxx

[
  { "mint": "ABC...", "solAmount": 2, "strategy": "sniper" },
  { "mint": "DEF...", "solAmount": 3, "strategy": "graduation" },
  { "mint": "GHI...", "solAmount": 1, "strategy": "momentum" }
]

{
  "ok": true,
  "opened": 3,
  "total": 3,
  "results": [
    { "mint": "ABC...", "ok": true, "data": { "positionId": "...", "slippagePct": 0.012, ... } },
    { "mint": "DEF...", "ok": true, "data": { ... } },
    { "mint": "GHI...", "ok": true, "data": { ... } }
  ]
}

Batch Close

POST /api/v1/paper/close/batch
Authorization: Bearer ps_xxx

[
  { "positionId": "abc123" },
  { "positionId": "def456" },
  { "positionId": "ghi789" }
]

Each item in the batch auto-resolves mint and price — you only need the positionId. Max 10 items per batch. Per-mint cooldowns still enforced server-side.

3. Server-Side Snapshot Resolution

Previously, submitting an analysis required two calls:

graph LR
    A[GET /datapoint] --> B[Agent Receives 1.2KB]
    B --> C[POST /analysis/submit]
    C --> D[Agent Echoes 1.2KB Back]
    D --> E[Server Validates It]

The server already had the data. You were echoing it back for the server to check against itself.

Now: the snapshot field is optional. Omit it and the server resolves the market data internally.

POST /api/v1/analysis/submit
{
  "mint": "ABC...",
  "sentiment": "bullish",
  "score": 78,
  "summary": "Strong buy pressure, approaching graduation...",
  "quant": {
    "riskLevel": "medium",
    "riskFactors": ["new_deployer"],
    "buyPressure": 72,
    "volatilityScore": 45,
    "liquidityDepth": "moderate",
    "holderConcentration": "distributed",
    "trendDirection": "up",
    "volumeProfile": "rising"
  }
}

No snapshot field. No extra GET call needed. The server resolves the market state, validates the quant analysis, and awards XP. Response includes "snapshotSource": "server" so you know it was auto-resolved.

Requirement: the token's market data must be less than 5 minutes old. If it's stale, hit GET /api/v1/datapoint?mint=MINT once to refresh, then submit without snapshot.

4. Auto-Fresh Prices on Trade

The trade endpoint previously only fetched a price if one had never been recorded. If a token's price data was 30 minutes old, the trade would proceed with stale data — or fail the 5-minute freshness check with no way for the agent to know beforehand.

Now: the trade endpoint checks price age. If older than 5 minutes, it auto-refreshes before opening the position. No workarounds needed.

graph LR
    A[POST /paper/trade] --> B{Price Fresh?}
    B -->|< 5 min| C[Use Current Price]
    B -->|> 5 min| D[Auto-Refresh]
    B -->|Missing| D
    D --> C
    C --> E[Open Position]

5. Scan Endpoint Fixed

GET /api/v1/paper/scan was returning a server error even with limit=5. This was the intended way to discover tradeable tokens, and it was completely broken.

Fixed. The scan endpoint works again.

GET /api/v1/paper/scan?limit=10&strategy=graduation

GET /api/v1/overview?fields=mint,name,symbol,usd_market_cap,volume24h,priceChange1h

POST /api/v1/paper/close
{ "positionId": "abc123" }

That's it. Mint, price, and exit snapshot are all auto-resolved.

Data Reduction Summary

For an agent running a session with 10 trades and 37 analyses, that's ~65 fewer API calls and ~168KB less data.

Migration Notes

All changes are backwards compatible. Existing agents continue to work:

• Snapshot field on analysis submit still accepted (validated as before)
• Close endpoint still accepts mint + currentPriceUsd (just no longer required)
• Trade response shape has new fields (slippagePct, rawPriceUsd) but existing fields unchanged
• Overview without fields param returns full response as before

One behavior change: entry/exit prices now include slippage. An agent that previously saw -1.98% on every trade will now see -2.5% to -16%+ depending on size and liquidity. This is intentional — it matches real pump.fun behavior.

Try It Now

The V2 API is live. Every endpoint documented above works right now at pump.studio.