September 2026: what it cost to sell $100,000

We track 1,073 assets. 582 have a contract on a covered chain, and 345 of those returned no on-chain price when we asked — an asset whose real market is a centralised exchange, or a receipt token never meant to trade, is invisible to this measurement rather than illiquid. The index is only ever a statement about the 86 we can see.

The index fell. The tokens barely moved.

The headline went from 14.99% in August 2026 to 14.53% this month, a move of -0.46 points. That is not the same statement as “selling got easier”, and the difference matters more than the figure.

52 tokens were measured in both months. Across those, the median change was -0.19 points: 23 improved, 13 deteriorated, and 16 stayed within half a point. So the index moved in the opposite direction to the market it is measuring. The sample grew from 78 tokens to 86, and what came in is worse than what was already there — which is what happens when a measurement reaches further down the market, not what happens when the market changes.

This is a caveat on our own headline. We publish it because somebody checking the figure against the CSV would find it in an afternoon, and it is worth more coming from us.

The same ticker is not the same market

16 tickers were measured on more than one chain this month. They are not one asset with one exit cost; they are separate pools with separate depth, and the gap between them is routinely larger than the whole spread of the index.

The practical reading: the figure you have seen quoted for a token is a figure about a market in it. If you hold the bridged version, the number that applies to you is the one for the chain your balance is actually on.

A deep pool is not a sellable one

Liquidity is the figure most sites publish instead of measuring an exit, because it can be read off a contract. These tokens each carry a pool of over a million dollars and still lost a fifth or more of a $100,000 sale:

• GT on Ethereum — pool $99,008,074, median loss 71.86%
• BASEDHYPE on Base — pool $3,278,617, median loss 99.11%

Depth is not the only thing that decides an exit. Where the depth sits in the curve, how concentrated it is in one venue, and what the router can actually reach all matter, and none of them are visible in a total. That gap is the whole reason this series exists.

There is no typical token

17.4% of measured tokens lost under 2%, and 25.6% lost more than 50%. The distribution has two ends and not much of a middle, which means the median is a weaker summary of it than a median usually is: there is no typical token here, there are deep markets and there are thin ones.

The two ends, by name. Dearest: BABYCLAW 99.97%, FT 99.72%, BASEDHYPE 99.11%, FLUID 99.04%, 1INCH 98.49%. Cheapest: AERO 0%, USAT 0%, WM 0.01%, GHO 0.03%, DAI 0.1%.

What we could not see

We track 1,073 assets. 582 have a contract on a chain we cover, and 86 of those returned a price we could route against. The rest are not illiquid — they are invisible to this method. An asset whose real market is a centralised exchange, and a receipt token that was never meant to trade, land in the same bucket.

That is why the headline says “of the tokens with a live on-chain market” and never “of the top 500”. The second sentence is unsayable from this data, it is the sentence this index was originally going to use, and finding out why it was wrong is the reason the denominators are printed on every month.

Check it, then cite it

Every figure above comes from one table. The series is downloadable, every token named links to its own measurement history, and the method is written down. If a number here is wrong we would rather hear it from you than not know.

Suggested citation: Cryptominium Exit-Friction Report, September 2026. https://cryptominium.com/exit-index/2026-09

The index and the full series Download the series (CSV) → · Download the chart (PNG, free with attribution) · JSON Every measurement behind it → How it is measured →