Do our scores actually work?

Every token we analyze gets scored twice: a health score (0–100, higher = stronger fundamentals) and a rug-risk score (how likely the token is a trap). We record the price at analysis and check it again 1, 6, 12, and 24 hours later. The tables below show what actually happened — every tracked token counts.

Our live analysis history

After an Avoid status

81 measured analyses · 68 tokens

82.08%

Negative outcomes

−48.98%

Average 24h price change

Includes assumed losses

Full study & data

Selected small-cap cohort · frozen historical study

Tokens below $5M market cap at analysis

Health Score 60+

+18.56%

Average price change over 24 hours

68 observations · 24 distinct tokens

High/extreme rug risk

−59.21%

Average price change over 24 hours

757 observations · 486 distinct tokens

Historical price changes, not trading profits. Repeated tokens affect the averages; the average is not the typical outcome.

Frozen 2026-10-26. Analyses from 3 August through 25 October 2026. Both groups require a positive recorded market cap strictly below $5,000,000 at analysis time, not today's market cap. Unknown market caps are excluded.

Measured completed 24-hour outcomes use the recorded analysis-time baseline. Records already marked dead retain the −100% convention: 412 in the risk group and 0 in the health group. This convention is not proof of fraud or an executable zero-price sale. Missing and unavailable price history is excluded, never assigned −100%.

These are 68 and 757 correlated analysis observations, not independent trades. One token contributes 13 of 68 high-health observations; without it, the average is negative.

This subgroup was explored after inspecting historical outcomes and has not been independently validated. Coverage is limited to saved outcome records and verified archived recoveries, not every analysis. Fees, slippage, liquidity and whether a trade could execute are not modelled.

The frozen study includes 66 verified archived 24-hour recoveries. It is separate from the live all-market-cap tables below; it does not establish that those recoveries have been applied to the live database.

How to read the track record

Results are grouped by the score or warning recorded at analysis time, then checked against the token's later price. Average and median changes, sample sizes and analysis date ranges are shown together. This is evidence about the screening engine, not a promise that any individual token will perform well.

Read the Research Centre for the study methodology, or analyze a Solana token with 1 free full analysis using the same engine.