Filters

News Pulse runs every potential mint through a chain of filters. Failed filters mean no alert — or, for some filters, alert-but-no-trade.

1. Pump.fun mint check

Address must end in pump. Non-pump.fun mints (Raydium-direct, launchlab, PumpSwap) have different MC math and pump.fun’s API can return junk values for them.

No alert, no trade.

2. 2-hop launder pattern

A sophisticated obfuscation: fund an OG wallet (which has prior pump.fun deploys), have it route SOL through a fresh single-use shell wallet, which then funds the actual dev wallet.

The bot traces this chain via Helius’s funded-by endpoint. If:

• Dev was funded by a wallet with ≤ 2 lifetime txs (shell signature)
• That shell received and forwarded the same amount in the same slot (atomic pass-through)
• The shell’s funder has ≥ 1 pump.fun deploy in their history

…then we have a confirmed 2-hop launder. The mint is silently rejected — no alert sent, no trade attempted. The pattern detection is logged internally for analysis.

Applies to all deploy counts.

3. Tumbler-chain funding

The most sophisticated rug pattern: dev wallet funded through 5+ single-use intermediate wallets, often laundered through MEV/DEX activity so no single OG is identifiable.

The bot walks 5 hops up the funding chain. If ≥ 60% of intermediate wallets are single-use (≤ 3 lifetime txs), the mint is silently rejected — no alert sent, no trade attempted.

This catches the JENNER-style 40-hop laundering chains that escape the 2-hop filter.

4. First-deploy half-sizing

If the creator has never deployed before (deploy_count <= 1), the alert still fires and the trader still buys — but at half size (0.1 SOL instead of 0.2 SOL).

Brand-new deployers have a higher rug rate but also include real launches, so we reduce exposure rather than skip outright. Combined with the launder/tumbler filters above (which block the actual rug patterns), this gives most of the upside with a fraction of the risk.

5. Creator blacklist

A small static list of known-rug-pattern creator addresses. These mints don’t fire News Pulse and aren’t auto-traded.

6. Mint age (trader only)

The internal trader bot skips mints older than 14 days. You still get the alert; you just won’t see a position open.

7. MC threshold

The fundamental filter — must cross $20,000 USD market cap. See Tiers & Thresholds.

8. MC sanity check

If Pump.fun’s API returns an obviously glitched market cap (e.g., $2M reported on a token with zero real SOL reserves), the mint is rejected before firing. Defends against the API’s occasional 100×–1000× MC corruption.

How they compose

Filters run in order, and failures short-circuit later checks:

1. Pump.fun mint check       (free, instant)
2. 2-hop launder check       (3 API calls)
3. Tumbler chain check       (5-10 API calls, only if 2 didn't flag)
4. First-deploy detection    (1 API call, drives sizing not skip)
5. Creator blacklist         (constant-time lookup)
6. Mint age (trader only)
7. MC threshold              (live, continuous)
8. MC sanity check           (cross-validation)

The order is optimized for cheap-failures-first: rejections that can be made from local data run first, expensive API-call chains last.

All filter activity is logged

Every skip is persisted internally with: mint, symbol, creator, deploy count, reason, source tweet — so we can later analyze which filters cost real upside and which save real losses. Filter tuning is data-driven.