Analyzing Pumpfun Sniper Tip Accounts

On Pumpfun, being first to buy a new token can mean the difference between a 10x and getting dumped on. We analyzed 23,635 tokens launched in the last 24 hours to discover which tip accounts (fee recipients) the fastest snipers use.

Methodology

For each token, we find the first trade after creation - defined as:

  • Same slot as creation but tx_idx > creation_tx_idx, OR
  • Any trade in a later slot

This excludes dev self-buys that are bundled in the same transaction as token creation.

Result: 22,129 tokens (93.6%) had a snipe after creation.

Top 25 Tip Accounts (Last 24 Hours)

RankTip AccountSnipesAvg DelayMed DelayMin Delay
1astrazznxsGUhWShqgNtAdfrzP2G83DzcWVJDxwV9bF1,2257,8693,3681
2astra9xWY93QyfG6yM8zwsKsRodscjQ2uU2HKNL5prk8434,4532,0201
3astra4uejePWneqNaJKuFFA8oonqCE1sqF6b45kDMZm4327,52311
4astraRVUuTHjpwEVvNBeQEgwYx9w9CFyfxjYoobCZhL3563,03911
596gYZGLnJYVFmbjzopPSU6QiEV5fGqZNyN9nmNhvrZU533116,1966,8181
6ADuUkR4vqLUMWXxW9gh6D6L8pMSawimctcNZ5pGwDcEt306152,1477,8621
7ste11p5x8tJ53H1NbNQsRBg1YNRd4GcVpxtDw8PBpmb25944216814
8DfXygSm4jCyNCybVYYK6DwvWqjKee8pbDmJGcLWNDXjh25211,0167,7661
9ADaUMid9yfUytqMBgopwjb2DTLSokTSzL1zt6iGPaS4924712,5007,7261
10HFqU5x63VTqvQss8hp11i4wVV8bD44PvwucfZ2bU7gRe24015,6067,8421
11DttWaMuVvTiduZRnguLF7jNxTgiMBZ1hyAumKUiL2KRL23816,9077,8561
12Cw8CFyM9FkoMi7K7Crf6HNQqf4uEMzpKw6QNghXLvLkY2388,9587,5571
133AVi9Tg9Uo68tJfuvoKvqKNWKkC5wPdSSdeBnizKZ6jT22026,8917,8651
14DiTmWENJsHQdawVUUKnUXkconcpW4Jv52TnMWhkncF6t2195502949
15HWEoBxYs7ssKuudEjzjmpfJVX7Dvi7wescFsVx2L5yoY21530521
1695cfoy472fcQHaw4tPGBTKpn6ZQnfEPfBgDQx6gcRmRg1906,12723511
17astraZW5GLFefxNPAatceHhYjfA1ciq9gvfEg2S47xk18919,28611
18Eb2KpSC8uMt9GmzyAEm5Eb1AAAgTjRaXWFjKyFXHZxF318737,66164223
19astraubkDw81n4LuutzSQ8uzHCv4BhPVhfvTcYv8SKC18321,68911
20FjmZZrFvhnqqb9ThCuMVnENaM3JGVuGWNyCAxRJcFpg91731,5312246
21astrawVNP4xDBKT7rAdxrLYiTSTdqtUr63fSMduivXK17279,74411
22astraEJ2fEj8Xmy6KLG7B3VfbKfsHXhHrNdCQx7iGJK1715,58811
23soyasDBdKjADwPz3xk82U3TNPRDKEWJj7wWLajNHZ1L16686843711
24noz9EPNcT7WH6Sou3sr3GGjHQYVkN3DNirpbvDkv9YJ1531,20426412
25DXfkEGoo6WFsdL7x6gLZ7r6Hw2S6HrtrAQVPWYx2A1s914658,40529,4469,848

Delay in tx_idx units. Same slot: delay = tx_idx difference. Cross-slot: delay = (slot_diff × 2000) + tx_idx_diff

Key Observations

The “Astra” Dominance

The astra* family of tip accounts holds 7 of the top 25 spots, including #1-4. Combined stats:

MetricValue
Total first snipes3,571 (16% of all snipes)
Accounts in top 257
Most have median delay1 tx_idx

These accounts consistently land in the very next transaction after token creation - indicating sophisticated infrastructure with low-latency mempool monitoring.

Speed Tiers

Looking at median delays, we see distinct performance tiers:

Tier 1 - Instant (Med ≤ 2 tx_idx): Ranks 3, 4, 15, 17, 19, 21, 22

  • Landing 1-2 tx_idx after creation
  • Same-slot sniping with sub-100ms reaction

Tier 2 - Fast (Med 100-500 tx_idx): Ranks 7, 14, 16, 20, 23, 24

  • ~50-200ms reaction time
  • Still competitive, likely geyser-based

Tier 3 - Standard (Med 2000-8000 tx_idx): Ranks 1, 2, 5, 6, 8-13, 18

  • ~1-4 seconds reaction time
  • High volume but not the fastest

Tier 4 - Slow (Med 29,000+ tx_idx): Rank 25

  • ~15+ seconds after creation
  • Catching tokens with no competition

Notable Accounts

#7 ste11p5x8tJ53H1NbNQsRBg1YNRd4GcVpxtDw8PBpmb: 259 snipes with avg delay of only 442 tx_idx and minimum 14. Consistently fast without any instant snipes.

#15 HWEoBxYs7ssKuudEjzjmpfJVX7Dvi7wescFsVx2L5yoY: 215 snipes with median delay of just 2 tx_idx. True instant sniper.

#25 DXfkEGoo6WFsdL7x6gLZ7r6Hw2S6HrtrAQVPWYx2A1s9: Minimum delay of 9,848 tx_idx (~5 seconds). Getting first position only on low-competition tokens.

Technical Details

Finding First Snipes

# For each token, find first trade AFTER creation
for trade in trades:
    trade_slot = trade['slot']
    trade_tx_idx = trade['tx_idx']

    # Trade must be strictly after creation
    is_after = (trade_slot == creation_slot and trade_tx_idx > creation_tx_idx) or \
               (trade_slot > creation_slot)

    if is_after:
        first_snipe = trade
        break

Calculating Delay

if snipe_slot == creation_slot:
    # Same slot: just tx_idx difference
    delay = snipe_tx_idx - creation_tx_idx
else:
    # Cross-slot: account for slot difference
    delay = (snipe_slot - creation_slot) * 2000 + (snipe_tx_idx - creation_tx_idx)

Extracting Tip Recipients

def extract_fee_recipient(top_level_transfers_json):
    """The smallest SOL transfer is typically the tip."""
    transfers = json.loads(top_level_transfers_json)
    sorted_transfers = sorted(transfers, key=lambda t: int(t.get('lamports', 0)))
    return sorted_transfers[0].get('to') if sorted_transfers else None

Implications

For Snipers

  1. The astra infrastructure dominates: If you’re not using similar infrastructure, you’re at a disadvantage for first position
  2. Median delay of 1 is achievable: Multiple accounts consistently hit the next tx_idx
  3. Volume vs speed tradeoff: Top accounts by count aren’t always the fastest

For Token Creators

  1. 93.6% of tokens get sniped: Expect a sniper in the first slot
  2. Consider bundling your first buy: To capture value before snipers
  3. Know who’s sniping: The tip account reveals the infrastructure

For Traders

  1. Competing for #1 is hard: Requires specialized infrastructure
  2. Positions 2-3 are more accessible: Let the astra bots fight for first
  3. Study the tip accounts: Knowing the landscape helps strategy

Data collected from Pumpfun on-chain transactions. Analysis reflects the last 24 hours.