The observation: size and breadth diverged
A market map answers several questions at once. The largest shape shows where market capitalization is concentrated. Its colour usually shows a price move. Count the rising assets instead, and you are asking a different question: how widely that move is shared.
We captured the 20 assets returned by Spectre’s public coins feed at 09:33:42 UTC on 15 September 2026. Bitcoin represented 69.19% of the market capitalization in that returned set, while Bitcoin and Ethereum together represented 82.72%. Yet nine of the twenty assets had positive reported 24-hour price changes. Eleven were negative.
The practical finding is straightforward: a view dominated visually by a few large assets can feel different from a count of individual winners and losers. Neither reading replaces the other. They describe concentration and breadth, respectively.
What we measured
The sample’s returned market capitalizations summed to approximately $2.231 trillion. Its five largest constituents accounted for 93.59% of that total. These are shares of this specific sample. They are not estimates of Bitcoin dominance across the entire crypto market.
- 20 assets in the returned sample; nine positive and eleven negative reported 24-hour changes.
- Bitcoin share of sample capitalization: 69.19%. Bitcoin plus Ethereum: 82.72%. Largest five: 93.59%.
- Median reported 24-hour change: −0.44%. Equal-weight average: −0.29%.
- Average reported 24-hour change weighted by current sample capitalization: −1.13%.
Why the averages differ
In an equal-weight average, every returned asset contributes one twentieth of the result. In the capitalization-weighted calculation, an asset contributes in proportion to its current market capitalization. Bitcoin’s reported 24-hour change was −1.19096%, and Ethereum’s was −1.83047%. Their large weights pulled the weighted result below the simple average.
The median answers another question: what sits in the middle of the twenty reported changes? It is less influenced by a single very large gain or loss than the arithmetic mean. Here, the median was negative even though nearly half the sample was positive.
Our weighted calculation is a descriptive statistic from a snapshot. It is not a backtested portfolio return. We used capitalization at observation time, not weights fixed at the start of the 24-hour period, and we did not model supply changes, trading costs or rebalancing.
A repeatable way to read the map
Begin by identifying the universe. A screen can show a selected group, a watchlist or a ranked sample rather than the whole market. Then inspect concentration, count positive and negative changes, and compare the median with the headline move.
Keep the measurement window consistent. This article uses the response’s reported 24-hour percentages; it does not reconstruct those returns from individual trades. We recorded each asset’s timestamp, which ranged from 09:31:24 to 09:32:38 UTC. The observations were close in time but not simultaneous.
Finally, keep visual sizing separate from arithmetic. Spectre’s public bubbles use a compressed size scale for readability. Their displayed diameters should not be used to calculate capitalization shares. Use the underlying values, as we did here.
Sources, limits and reproduction
This is original analysis of a first-party Spectre data response, not a rewritten news report. The frozen source file includes all twenty returned records, observation timestamps, derived statistics and calculation definitions. The response labelled its source “Spectre Market Data”; we did not independently reconcile every quote against an exchange.
The selected set is not a complete market census. Market capitalization is not liquidity or the amount of money invested, and a 24-hour snapshot does not explain what caused the moves. The cover illustrates concentration conceptually and contains no measured data.
To reproduce the concentration figures, divide each selected capitalization by the sum of all twenty values. Count the signs of the reported changes for breadth; sort those changes for the median. Rounding happens only when presenting the results. This article is a dated research example, not a live market update or a trading recommendation.
Sources
See a discrepancy? Send the article URL and supporting evidence to spectre@spectreai.io.
Explore Spectre