Benchmarks
What a normal pool looks like
A benchmark here is a distribution, not a ranking. It answers one question: is the number in front of you ordinary or unusual for its chain? Nothing on this page describes a single token, and nothing on it is a verdict on a project — a pool at the tenth percentile is a thin pool, which is a fact about depth and not an accusation.
Who is in this population
The 19 tokens people have analyzed on MoonMath that cleared the quality bar for an indexable page: at least $5M of market capitalisation and at least $100K of pool liquidity when they were recorded, with a small curated set admitted from $10K. This is not every token on these chains, and it is not a random sample of them. Two consequences worth stating before you read a figure:
- The population is cut off at the bottom on the very metric published here. The low percentiles are a floor set by that rule, not a floor found in the market — no token below the threshold is in this sample, however many of them exist.
- Each figure is that token's own snapshot from the last time somebody analyzed it, between 2026-07-29 and 2026-09-03. It is not a synchronised reading of every pool at one moment, and it does not become one by being put in a table.
- Selection is by attention: a token is here because somebody looked it up. That makes this a distribution of what gets analyzed, which is the honest thing it can be a distribution of.
- Observations
- 19
- tokens in the population
- Chains published
- 0 / 6
- the rest are withheld
- Minimum sample
- 30
- observations per chain, or nothing
No chain has reached 30 observations yet, so every distribution below is withheld. That is the page working, not the page broken: the sample sizes are shown so you can see how far off publication each chain is. They are deliberately not combined into one cross-chain figure — pooling six thin samples to clear a gate is a way around the gate, not a way past it.
Pool liquidity by chain
Total reported pool value in USD, both sides of the pool, as each token was last observed. A chain publishes only once it has 30 observations; below that the figures are withheld rather than shown with a caveat.
| Chain | Sample | p10 | p25 | Median | p75 | p90 |
|---|---|---|---|---|---|---|
| Solana | 6 | Withheldsample too small — 6 of 30 observations | ||||
| Ethereum | 3 | Withheldsample too small — 3 of 30 observations | ||||
| BSC | 2 | Withheldsample too small — 2 of 30 observations | ||||
| Polygon | 3 | Withheldsample too small — 3 of 30 observations | ||||
| Avalanche | 1 | Withheldsample too small — 1 of 30 observations | ||||
| Base | 4 | Withheldsample too small — 4 of 30 observations | ||||
What a $10K buy costs
Price impact of a single $10K market buy against the pool, under constant product, using the same calculation the liquidity impact calculator runs. This is the liquidity distribution above restated in the unit a trade is actually felt in — the same population, not a second measurement, so the deepest tenth of pools is the cheapest tenth of trades by construction rather than by agreement.
| Chain | Sample | p10 | p25 | Median | p75 | p90 |
|---|---|---|---|---|---|---|
| Solana | 6 | Withheldsample too small — 6 of 30 observations | ||||
| Ethereum | 3 | Withheldsample too small — 3 of 30 observations | ||||
| BSC | 2 | Withheldsample too small — 2 of 30 observations | ||||
| Polygon | 3 | Withheldsample too small — 3 of 30 observations | ||||
| Avalanche | 1 | Withheldsample too small — 1 of 30 observations | ||||
| Base | 4 | Withheldsample too small — 4 of 30 observations | ||||
How this is computed
- Percentiles use linear interpolation between closest ranks — the method NumPy, R type 7 and spreadsheet
PERCENTILE.INCall use. It is named so the figures can be reproduced: percentile conventions disagree at the edges by a whole observation. - Only p10, p25, median, p75 and p90 are published. There is no minimum and no maximum, because those are individual observations — publishing them would publish one token's value under the name of an aggregate.
- The withholding happens in the calculation, not in this page. A chain below the sample threshold has no percentiles to render, so no future caller can ask for one.
- Price impact treats the reported pool value as both sides of the pool, so a buy consumes quote inventory worth half of it. Constant product with no fee, no routing and one pool: it is a floor on the cost, not a quote.
What is deliberately not here
- Holder concentration percentiles. MoonMath fetches holder distributions live for the token you are looking at and does not keep them for tokens nobody saved, so there is no population to describe. Building one would mean collecting holder data for its own sake, which is a different decision from publishing what is already measured.
- Fee cost per swap size. A swap fee is a constant per chain — 0.25%–0.30% depending on the dominant venue — so its “distribution” would be one number repeated as many times as there are tokens. A table of that would look like a measurement and contain none.
- Anything about a single token. If you want one token's numbers, run its calculation; that page says what was observed and when. This page exists so that number has something to be compared against.
What MoonMath is used for · What MoonMath does and does not process