Insider-Risk Bands and 24-Hour Outcomes in 2,259 Solana Tokens

Correction and methodology clarification (September 2, 2026): We clarified missing-data, selection, confounding, auditability and verification disclosures. The cohort and reported figures are unchanged.

Why insider risk, specifically

Our first outcome study compared overall Health Score bands with observed outcomes. This study describes 24-hour outcomes across one component: insider risk — an analysis-time reading based on how much of a token's supply sits with wallets identified as connected to each other or to the creator, including bundles bought in the launch block, wallets funded from one source and creator-linked holders. Those relationships are not visible from price movement alone, and inferred wallet connections do not by themselves prove common ownership or coordinated activity.

For the selected complete-case cohort, the MemeAssist screening engine computed an insider-risk reading (0–100, higher = worse) at analysis time and the tracker automatically recorded a usable price 24 hours later. Between July 23 and August 9, 2026, 2,259 tokens had both required records and were included.

Observed outcomes by historical insider-risk band

These figures apply to the selected complete-case Jul 23–Aug 9, 2026 observational cohort. The 60-point split was the historical band used for this analysis, not a validated decision threshold. Three descriptive results stand out:

• Large losses were more common in the 60+ band. Within this cohort, 14.2% of tokens in that band lost at least half their value within a day, compared with 2.4% below 60—an unadjusted ratio of about 6×. In the more severe tail, 8.2%—about 1 in 12—lost 80% or more. These price outcomes do not establish that holders coordinated exits.
• The group average does not describe every token. The −11.3% mean for the 60+ band summarizes varied individual outcomes, including a subset of large losses. This observational comparison does not identify a mechanism or show that insider concentration caused those losses.
• Large-gain rates were numerically close. Within this cohort, 50%+ gains occurred in 2.5% of tokens in the 60+ band and 2.6% below 60. That single unadjusted sample does not prove identical upside, a risk-reward trade-off or how future tokens will perform.

What "insider risk" actually measures

The reading combines connected-supply evidence our engine reconstructs from on-chain data at analysis time: bundled wallets that bought in the launch block, holders funded from a common source, insider concentration around the creator, and launch-window accumulation patterns. In this selected complete-case Jul 23–Aug 9, 2026 cohort, readings of 40–59 did not show the same descriptive outcomes as the historical 60+ band. That pattern is unadjusted and does not validate 60 as a cutoff, identify who controls the wallets, establish coordinated selling or predict future-token outcomes.

Honest caveats on the insider risk data

The following limits define what this study can and cannot establish:

• Cohort: tokens that reached our screening engine and completed a 24-hour outcome check with a usable price. Tokens whose market stopped quoting entirely are excluded rather than assumed dead. The direction and size of the resulting bias cannot be established from this complete-case analysis.
• One horizon. The study measures price change over 24 hours only; it does not evaluate losses or other events outside that window.
• Association, honestly framed. The comparison is unadjusted, and insider-risk readings may be associated with other traits such as liquidity and creator history. It describes an association in this cohort; it does not establish coordination, causation, a mechanism or future-token performance.
• Missing outcomes: tokens without both an analysis-time insider score and a usable 24-hour price were excluded; they were not treated as flat or as losses. If missing prices are more common after severe failures, the reported loss rates are biased downward.
• Selection and uncertainty: this is the set of tokens submitted to the MemeAssist analyzer, not a random sample of all Solana launches. This period-specific selected cohort may not generalize to other periods or to all Solana launches. The comparison is descriptive, unadjusted for market cap, liquidity, token age or overlapping risk signals, and no confidence interval or causal effect is claimed.

Methodology

Cohort: every token scored by the MemeAssist screening engine between July 23 and August 9, 2026 whose 24-hour outcome check completed with a valid 24-hour price and whose analysis-time snapshot recorded an insider-risk sub-score (n = 2,259). The insider-risk reading is computed at analysis time from live on-chain data; the outcome is the price 24 hours later versus the price at analysis, recorded automatically by the same tracker that powers our public track record. No token was excluded for performing badly. Every claim in this study rests on a cohort of at least 30 outcomes — our minimum bar for publishing a number.

Auditability and verification: public row-level records for this historical cohort are not available. The source facts and outcomes were generated from the internal production tracker and were not manually verified row by row. Accordingly, the aggregate tables cannot be independently row-audited from public data.

Where the insider-risk reading appears

The reading used in this study appears in MemeAssist reports alongside the Overall Health Score, Rug Risk Rating, AI Verdict and Detailed Risk Breakdown, including the connected-wallet evidence behind the reading. It is one descriptive input rather than a proven trading rule or validated prediction of an individual token's future performance.