Open data

ARC AI Data

Two monthly reports built only from data ARC AI recorded at the time: the Rug Risk Index for promoted Solana launches, and the graded accuracy of our AI agents, published even in bad months. Each edition is frozen when the month closes, so the numbers you cite never change.

How every number is computed: methodology.

Editions

The first edition (October 2026) is published on 1 November 2026.

October 2026 so far

Preview: October 2026 is still in progress, so these numbers change until the month closes. Cite the frozen edition, published on the 1st.

Launch Radar data starts on 02 Oct 2026; earlier days of October 2026 are not covered.

Promoted Solana launches

Tokens from DexScreener's new-profile and boost lists that reached $10K liquidity and $5K daily volume within 7 days of launch. Promoted launches only, not every token created.

Tracked
1,755
Risk score 50+ (RugCheck)
6.2% (107 of 1,730)

Share of the tokens RugCheck checked. Latest score, not the score at launch.

Risk score 75+ (RugCheck)
1% (17 of 1,730)
Wash-trading pattern
14.3% (251 of 1,755)

Volume over 50× the pool, or $100K+ from under 50 trades.

25 tracked tokens were never checked by RugCheck and are left out of the risk shares.

RugCheck risk scores, October 2026. Number of checked tokens in each RugCheck risk-score range (0 = no issues found, 100 = highest risk).
  • Score 0–191,499
  • Score 20–3999
  • Score 40–5973
  • Score 60–7950
  • Score 80–1009

Most common RugCheck findings

Most common RugCheck danger and warning findings. Number of tracked tokens with each finding.
  • Single holder ownership109
  • Low Liquidity61
  • High holder concentration48
  • Large Amount of LP Unlocked46
  • Creator history of rugged tokens42
  • Top 10 holders high ownership33
  • Copycat token32
  • Low amount of LP Providers23

What happened after ARC AI's warnings

Our Sentinel agent warns about risky tokens that can be traded. Each warning is checked 24–48 hours later against the pool's liquidity and price at the moment of the warning: 3,232 tokens checked this month, one result per token. There is no comparison group: this shows what followed a warning, not how warned tokens compare with others.

Lost 50%+ of liquidity or price, or the pool died
90.7% (1,167 of 1,286)

Of the 1286 warnings with a clear result.

Held 70%+ (warning wrong)
9.3% (119 of 1,286)
Mint authority open at warning
46.7% (1,508 of 3,232)
Freeze authority open at warning
19.7% (638 of 3,232)

1,946 were in between or lacked data and count as neither. A further 437 warnings repeated a token already checked and are not counted again.

72 hours after launch

Liquidity under 50% locked at first sight
7% (3 of 43)

Tokens first seen within 6h of launch: 54.

Lost 50%+ of liquidity, or the pool died, by 72h
not enough data (0 of 0; under 20)

54 tokens first seen in the last 3 days are left out: they had not had 72 hours yet.

Median liquidity kept at 72h
not enough data