This piece covers the completed daily session of 2026-09-02 across the 20 USD-quoted crypto pairs tracked on Coinbase Exchange candles. The session closed and the figures are fixed; the next session was still open at the time of writing.

The desk classified the session as dispersed — a label that describes the spread of outcomes across the 20 pairs on that single day, not a condition expected to persist or repeat.

What the Session Did

Eight of the 20 pairs closed higher on 2026-09-02. The median session change was -0.05%, a figure close enough to flat that the central tendency of the basket was essentially unchanged. The weight of the session, however, was not in the median: APT led at +7.18% while FET brought the rear at -1.69%, producing a dispersion figure — strongest minus weakest — of 8.87 percentage points.

That gap is the defining feature of the session. Most pairs clustered within roughly one percentage point of flat, but APT's move pulled the top of the distribution well clear of the pack. OP was the only other pair to post a gain above +2%, closing at +4.26%. The remaining six advancers ranged from +2.29% (ADA) down to a barely-positive +0.06% (DOGE).

Range positions — where each pair's close sat inside its own 30-day high-to-low band — added texture. STX closed at 83% of its 30-day range despite a modest +2.23% session gain, reflecting a strong prior month (+89.90% over 30 days). SOL and BTC sat at 73% and 78% of their respective 30-day ranges while posting small negative or near-flat session changes of +0.46% and -0.12%. APT, despite leading the session, closed at only 37% of its 30-day range, meaning its single-day surge still left it in the lower half of where it had traded over the prior month. The full pair-by-pair figures are in the accompanying table.

The Dispersed Regime Described

A dispersed session is characterised by an unusually wide gap between the strongest and weakest performers relative to a session where most pairs move together. On 2026-09-02 the 8.87-percentage-point spread between APT (+7.18%) and FET (-1.69%) drove that classification. In a broad-advance or broad-decline session the majority of pairs move in the same direction and the spread is narrower; in a compressed session both breadth and dispersion are tight. Dispersed sessions sit apart because individual pairs diverge sharply even when the median is near zero.

The breadth figure of 8 out of 20 is below the midpoint, meaning more pairs declined than advanced, but the magnitude of the top gainer was large enough to dominate the dispersion calculation. FET's -1.69% loss, while the weakest reading, was not an extreme move on an absolute basis; the asymmetry came from the top, not the bottom.

What Dispersion Means for Execution Cost

On a session where the median move was -0.05%, the round-trip cost of a standard maker-taker trade — typically in the range of 0.10% to 0.20% of notional on a major spot venue, depending on tier — would have consumed the entire median session move and then some. For any pair that closed within roughly ±0.50% of its open, the session change was of the same order of magnitude as execution cost, making the net outcome of any round-trip sensitive to the exact spread and fee tier rather than to the direction of the move.

The outlier at the top of the distribution is a different case. APT's +7.18% session change is large enough that a round-trip fee is a small fraction of the gross move. The practical implication is mechanical: dispersion widens the range of pairs where the gross move is large relative to cost, but it concentrates that condition in a small number of names. The majority of the 20 pairs on 2026-09-02 sat in the zone where execution cost is not a rounding error.

Slippage compounds this. A pair that moves sharply intraday — as APT did — may carry a wider effective spread at the moment of execution than its quoted spread implies, because liquidity providers reprice into fast-moving markets. The published close reflects the end-of-session price, not the price available during the move.

What One Session Does Not Establish

A single daily session is a sample of one. The dispersed label describes what the data showed on 2026-09-02; it carries no information about whether the following session will be dispersed, compressed, or anything else. Regime classifications are taxonomic, not predictive — they sort closed sessions into categories after the fact.

APT's +7.18% gain is the largest single-session move in the table, but its 30-day range position of 37% means it closed in the lower portion of where it has traded over the prior month. A strong session does not reset the longer-term context, and a single day's leadership has no established tendency to persist or reverse.

FET's position at 50% of its 30-day range despite a -1.69% session loss and a -9.65% seven-day decline illustrates a separate limitation: range position describes where a close sits relative to recent extremes, but it does not indicate whether those extremes are about to be tested. The measure is backward-looking by construction.

Breadth of 8 out of 20 is a weak signal on its own. It is below the midpoint, but not dramatically so, and it co-exists with a near-zero median. One session's breadth reading is not a reliable basis for characterising the health of the broader basket.

The 2026-09-02 session closed with its numbers fixed in the table; the dispersion that defined it was the arithmetic result of one pair moving sharply while the rest of the basket sat near flat.

Note: This explains how a mechanism works. It is educational and is not financial, investment or trading advice, not a recommendation, and not a signal service. Any figure describes a session that has already closed and says nothing about what happens next. Trading cryptoassets risks losing money, including all of it.