Session Read 2026-09-12: 14 of 20 Pairs Higher

This post covers the daily session that closed on 2026-09-12 across the 20 USD-quoted crypto pairs tracked by Skia Paper's session desk, sourced from Coinbase Exchange daily candles. The table accompanying this piece shows each pair's close, session change, 7-day and 30-day change, and where the close sat inside that pair's own 30-day range.

The session was classified dispersed — the label applied when the spread between the strongest and weakest mover is wide enough to distinguish the session from a broad uniform advance or decline. That label describes what happened on the 12th; it says nothing about what follows.

What the Session Did

Fourteen of the 20 pairs closed higher on 2026-09-12. The median session change was +0.48%, a figure pulled upward by a handful of pairs and kept modest by the six that closed in the red. Dispersion — the gap between the strongest and weakest mover — came in at 8.00 percentage points, stretching from STX at +5.89% down to ATOM at -2.12%.

STX was the clear outlier on the upside, closing at 80% of its own 30-day range — meaning the 12th close sat well toward the top of the range that pair had traced over the prior month. That 30-day figure of +115.09% for STX is the largest in the table by a wide margin and provides the context for why a single-session reading of +5.89% still left it below its 30-day high. ATOM, the weakest pair, closed at 34% of its 30-day range, near the lower half, with a 30-day gain of only +6.96% — the smallest in the table.

Among the larger-cap pairs, BTC added just +0.07% on the session while sitting at 75% of its 30-day range, and ETH gained +0.37% at 82% of its range — both near the upper end of their recent ranges despite modest session moves. SOL and AVAX both closed lower, at -0.68% and -0.84% respectively. DOT was the second-weakest on the session at -1.99%, a notable contrast with its 7-day reading of +11.36%, the largest 7-day figure in the table. The table contains the full picture; the figures here are selected only to illustrate the shape of the session.

What "Dispersed" Describes

The dispersed label is applied when the spread between the session's strongest and weakest pair is wide relative to the typical session, and when that spread is not explained by a uniform directional move — in other words, when the market is not simply all up or all down together. On 2026-09-12, 8.00 percentage points separated STX from ATOM, and the six declining pairs included names from different segments of the table rather than clustering at one end of the market-cap spectrum.

A dispersed session differs from a broad-advance or broad-decline session, where most pairs move in the same direction and the spread is narrower, and from a compressed session, where nearly all moves are small regardless of direction. It also differs from a mixed session, where breadth is roughly even. Here, breadth was tilted positive at 14 of 20, but the distribution of moves was uneven enough that the dispersion reading, rather than the breadth reading, is the more informative descriptor of the session's character.

Range positions on the 12th were spread across the full spectrum: ETH and BTC sat in the upper quartile of their 30-day ranges (82% and 75%), while ATOM and OP sat in the lower third (34% and 35%). That spread of range positions is consistent with the dispersed classification — pairs were not uniformly extended or compressed relative to recent history.

What This Session Meant for Execution Cost

On a dispersed session, the cost picture varies sharply depending on which pair is under consideration. For a pair like STX, whose session move was +5.89%, a standard maker fee of roughly 0.40–0.60% of notional (the range published by Coinbase Exchange for its tiered fee schedule) represents a small fraction of the gross move on a round trip — perhaps 10–20 basis points of the session's range as a share of the move. For a pair like BTC at +0.07% or OP at -0.02%, the same round-trip fee is larger than the session move itself, meaning any strategy that needed to open and close within that single session would have faced a fee that exceeded the gross price change.

Spread and slippage compound this asymmetry. Larger-cap pairs such as BTC, ETH, and SOL tend to carry tighter bid-ask spreads and absorb size with less slippage than smaller-cap pairs; STX, whose move was the largest, is a thinner market where the quoted spread can widen during volatile moves, partially eroding the gross gain for a taker. A session move of +5.89% is large enough that even a wider spread is a modest fraction of it, but the comparison is different for pairs whose session moves sat in the +0.25% to +0.75% band — the cluster of pairs near the median.

The three ways a strategy clears its round-trip cost are frequency, size, and holding period. A strategy that held through the 30-day window — where moves ranged from +6.96% for ATOM to +115.09% for STX — would have a very different cost-to-move ratio than one that opened and closed on the 12th alone. Nothing here recommends a holding period; the observation is that the 30-day figures in the table are large enough, for most pairs, that a round-trip fee is a smaller fraction of them than of any single session's move.

What One Session Does Not Establish

A single daily session is a sample of one. The dispersed classification describes the character of 2026-09-12 accurately, but it carries no implication about the character of the session that follows. Dispersed sessions can precede broad advances, broad declines, compressed sessions, or further dispersion — the label is retrospective, not predictive, and the historical base rate of any transition is not derivable from one observation.

The 30-day figures in the table — some of them very large, such as STX's +115.09% or NEAR's +44.59% — describe price changes over a window that has already closed. They do not indicate that those moves will continue, reverse, or stabilise. Range position tells where a close sat inside a completed range; it does not establish where the range's boundaries will be in future sessions.

The breadth figure of 14 of 20 is a majority, but a majority in a 20-pair sample is a weak signal. A coin-flip model would produce 14 or more heads in 20 flips roughly 6% of the time — uncommon but not rare. The median of +0.48% is a central tendency for this session only. The structural failure of any reading drawn from a single session is that the session is too short a window to distinguish signal from noise in an asset class with the volatility profile of spot crypto.

The 2026-09-12 session closed with its character settled — 14 advancers, 8.00 points of dispersion, and a wide scatter of range positions — and that record is now fixed; what the next session does with it is a separate question entirely.

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.