When Bitcoin Had Its Worst Days in October, Ether Had One Too — Every Single Time

A threshold-breach analysis of Coinbase USD spot data for Bitcoin, Ether, and Solana through October 27, 2026 finds that Ether breached its September-calibrated daily loss cutoff on all five dates in October when Bitcoin also breached its own — a 100% overlap on a five-observation conditional group. Solana breached on four of those five dates. On the remaining 22 October dates when Bitcoin did not breach, Ether breached three times and Solana never did. The finding is methodologically precise but emphatically limited: five Bitcoin breach dates are not a sufficient basis for probability inference, the calibration window is short and retrospectively chosen, and the author is explicit that these observed coincidences carry no trading signal and do not establish a lasting market relationship.

What This Analysis Measures — and What It Does Not

Before reading the data, it is essential to understand what this analysis is and is not trying to show.

This is not a standard correlation analysis. Correlation measures the degree to which two assets move together across all observations in a sample — giving equal weight to every day, large or small. The threshold-breach coincidence measured here asks a different question: on the specific days when one asset had an unusually large loss (strictly below its calibrated threshold), how often did the other assets also have unusually large losses?

This is also not a beta analysis. Beta measures the magnitude of one asset's returns relative to a benchmark, fitting a linear relationship across the return range. This analysis ignores magnitude once a breach is classified — a day that falls barely below the threshold is treated identically to a day with a catastrophic loss.

The contribution of this analysis is locating unusually weak closes on common dates under fixed, asset-specific definitions — identifying whether tail-loss events in Bitcoin tend to coincide with tail-loss events in Ether and Solana at a frequency that differs meaningfully between the Bitcoin-breach group and the Bitcoin-nonbreach group. That is a specific and useful question for tail-risk assessment, and the answer this data produces is striking — tempered by the small sample that sits behind it.

The Methodology: How Thresholds Are Defined and Breaches Classified

The analysis uses a rigorous, reproducible methodology with all parameters fixed before October observations enter the evaluation.

The Data

All prices use Coinbase Exchange USD spot closes — specifically, the Coinbase candle close defined as the last trade within each UTC daily bucket. The bucket timestamp represents the bucket's start; the close is the final trade price before 00:00 UTC begins the next bucket. Under this convention, the October 27 return uses the October 27 bucket's last trade, with that bucket completing at 00:00 UTC on October 28.

The input window covers 57 matched completed daily buckets per asset, dated September 1–October 27, 2026. September 1 provides only the opening price baseline — it generates no return itself. This leaves:

• 29 September returns (September 2–30): used exclusively to calibrate each asset's threshold

• 27 October returns (October 1–27): used exclusively to classify breaches

Buckets starting October 28 or later are excluded from all counts.

Logarithmic Returns

Daily returns are calculated as logarithmic returns: log return = ln(current close / previous close). The logarithmic percentage is expressed as 100 × ln(price ratio) — not the more common simple percentage change (which would be expressed as (current − previous) / previous × 100).

The distinction matters for precise comparison. A simple return and a logarithmic return are approximately equal for small price moves but diverge as moves become larger. The equivalent simple return can be recovered from a logarithmic threshold by the formula: 100 × (exp(logarithmic threshold) − 1).

Both the logarithmic and equivalent simple-return thresholds are reported below for each asset. All actual classifications use unrounded logarithmic values; the rounded figures are provided for readability only.

Logarithmic returns are preferred in financial analysis for several reasons: they are time-additive (multi-day returns can be calculated by summing daily log returns), they are symmetric (a gain followed by an equal loss returns to the start), and they produce a more normal distribution for statistical analysis than simple returns, which are bounded at −100% but unbounded on the upside.

Threshold Definition

For each asset, the threshold is the sixth order statistic of the 29 September log returns — the sixth-lowest value when the 29 September returns are sorted from lowest to highest. The sixth order statistic of 29 is the nearest-rank empirical 20th percentile: 6/29 ≈ 20.69%, which is the closest rank position to the 20th percentile in a 29-observation sample without interpolation.

This means each asset's threshold is independently determined: it represents that asset's own definition of an unusually bad September day, calibrated to the left tail of its own return distribution using only prior-period data.

A breach is defined as a daily log return strictly below the threshold. Equality — a return exactly equal to the threshold — does not count as a breach. This strict inequality ensures that any ambiguity in rounding does not distort classifications.

Bitcoin's Threshold and October Breach Dates

Bitcoin's September-calibrated threshold and October results:

Metric Value Logarithmic threshold −1.197497% Equivalent simple return −1.190356% October breaches 5 of 27; 18.52% October nonbreaches 22 of 27

Bitcoin's five threshold breaches in October occurred on: October 7, 8, 12, 14, and 22, 2026.

These five dates define the Bitcoin-breach group — the conditioning set for evaluating Ether's and Solana's breach behaviour. The remaining 22 October dates form the Bitcoin-nonbreach group.

An important note on interpretation: 18.52% of October days produced Bitcoin threshold breaches (5 of 27). Approximately 20% of September days exceeded the threshold by definition (the threshold was calibrated to the 20th percentile). The near-alignment is not surprising but is not guaranteed: September's 20th percentile calibration tells us about the September distribution, not the October one. The October breach rate reflects what actually happened in October.

Ether: 100% Coincidence Within the Bitcoin-Breach Group

Ether's September-calibrated threshold and October results:

Metric Value Logarithmic threshold −1.098994% Equivalent simple return −1.092978% On BTC-breach dates 5 of 5; 100% On BTC-nonbreach dates 3 of 22; 13.64% Frequency difference 86.36 percentage points All October dates 8 of 27; 29.63%

Ether's threshold breaches in October occurred on: October 2, 7, 8, 12, 14, 22, 24, and 25, 2026.

Ether breached its own threshold on every single date that Bitcoin breached its threshold — five for five. Within the 22 Bitcoin-nonbreach dates, Ether breached three times independently (October 2, 24, and 25).

The frequency difference of 86.36 percentage points — the gap between 100% breach rate within the Bitcoin-breach group and 13.64% outside it — is the headline statistic. It shows that Ether's large-loss days were highly concentrated on the dates when Bitcoin also had a large-loss day.

What Ether's independent breaches reveal: The three Ether breaches on Bitcoin-nonbreach dates (October 2, 24, 25) matter for a complete picture. They confirm that Ether can have unusually bad days without Bitcoin also having one — the relationship is not deterministic in both directions. October 2, 24, and 25 are dates when something affected Ether specifically, without rising to the level of a Bitcoin threshold event. Any risk framework that relied solely on monitoring Bitcoin for signals about Ether would have missed these three independent Ether tail days.

Solana: Four-of-Five With Zero Nonbreach Coincidences

Solana's September-calibrated threshold and October results:

Metric Value Logarithmic threshold −2.467787% Equivalent simple return −2.437586% On BTC-breach dates 4 of 5; 80% On BTC-nonbreach dates 0 of 22; 0% Frequency difference 80.00 percentage points All October dates 4 of 27; 14.81%

Solana's threshold breaches in October occurred on: October 7, 8, 12, and 22, 2026.

Solana breached on four of the five Bitcoin breach dates — missing only October 14, the date when Bitcoin and Ether both breached but Solana did not. On all 22 Bitcoin-nonbreach dates, Solana did not breach once.

The threshold difference requires careful interpretation: Solana's logarithmic threshold is −2.467787%, versus Bitcoin's −1.197497% and Ether's −1.098994%. Solana's threshold is approximately twice as large in absolute magnitude as Bitcoin's and more than twice Ether's. This reflects the difference in the three assets' September volatility profiles — September's 20th-percentile loss day was simply a much larger move for Solana than for Bitcoin or Ether.

This is why the article explicitly notes that "Solana's lower event count does not establish smaller losses or lower overall risk; its September-calibrated cutoff required a substantially larger simple loss." Solana's four breaches each represented a daily loss exceeding approximately −2.44% (simple return equivalent). The five Bitcoin breach days each required only a loss exceeding approximately −1.19%. The assets are being assessed against their own volatility profiles, not against each other's absolute thresholds.

October 14 — the exception: October 14 is the one Bitcoin breach date on which Solana did not breach. Bitcoin fell below its threshold (−1.197497% log return), Ether fell below its threshold (−1.098994%), but Solana did not fall below its own threshold (−2.467787%). The numbers are consistent with Solana experiencing a negative day on October 14, but not a day negative enough to reach its own high-volatility threshold — the threshold that required a move roughly twice as large as Bitcoin's.

The Complete October Picture: Day-by-Day Classification

The exhaustive daily record of threshold breaches across all three assets for October 1–27, 2026, using UTC daily buckets:

• October 1: None

• October 2: Ether only

• October 3–6: None

• October 7–8: Bitcoin, Ether, and Solana

• October 9–11: None

• October 12: Bitcoin, Ether, and Solana

• October 13: None

• October 14: Bitcoin and Ether (Solana does not breach)

• October 15–21: None

• October 22: Bitcoin, Ether, and Solana

• October 23: None

• October 24–25: Ether only

• October 26–27: None

There were no threshold ties for any asset during October — no day produced a return exactly equal to any asset's threshold.

The four dates on which all three assets breached simultaneously — October 7, 8, 12, and 22 — are the most significant concentrations of joint tail risk in the sample. These are the dates when investors holding any combination of BTC, ETH, and SOL faced large losses across their entire crypto portfolio in a single session.

What "86.36 Percentage Points" Actually Means

The frequency difference figures — 86.36 points for Ether and 80.00 points for Solana — deserve precise interpretation because they are the central empirical result.

The 86.36 percentage point gap for Ether means that the breach rate in the Bitcoin-breach group (100%) exceeded the breach rate in the Bitcoin-nonbreach group (13.64%) by 86.36 percentage points. It is the difference between two conditional frequencies — not a percentage of something, not a correlation coefficient, not a p-value.

To understand what would make this gap narrow or disappear: if Ether breached at the same rate whether or not Bitcoin breached (say, 30% in both groups), the difference would be zero. If Ether breached on fewer Bitcoin-breach dates than Bitcoin-nonbreach dates, the difference would be negative. The observed positive difference of 86.36 points reflects the concentrated coincidence: Ether's threshold breaches were much more likely to fall on Bitcoin-breach dates than on Bitcoin-nonbreach dates.

The relevant baseline for this comparison is the assets' unconditional October breach frequency — how often they breached across all 27 October dates. Ether's unconditional October frequency is 29.63% (8 of 27). Solana's is 14.81% (4 of 27). These are the reference points against which the conditional frequencies within the Bitcoin-breach and Bitcoin-nonbreach groups should be compared.

The Caveats: What This Analysis Cannot Support

The BTCUSA comment that accompanies the data is unusually candid about the limitations of the finding, and those limitations are significant enough to reproduce in full detail.

The Fundamental Sample Problem

The Bitcoin-breach conditional group contains five observations. Five. Statistical inference from five observations is inherently precarious — the confidence intervals around any probability estimate from a five-observation sample span most of the probability range. A finding of 5-of-5 (100%) is literally compatible with true conditional probabilities ranging from roughly 56% to 100% at standard confidence levels, depending on assumptions.

The BTCUSA comment notes this directly: "The strongest counterargument is sample sensitivity." Five Bitcoin threshold breaches is not enough to establish a stable empirical relationship. A month with seven Bitcoin breach dates and only four Ether coincidences would show dramatically different ratios. The finding is vivid precisely because the conditional group is small enough that 100% is achievable.

The Retrospective Calendar Split

The September-calibration / October-evaluation split was chosen after the fact — September and October were selected, and then the methodology was applied. This is standard in demonstrating how a methodology would function, but it introduces the risk that the apparent clean split was partly influenced by awareness of what happened in each month. The finding could look different with a different split date (August-September calibration, for instance, or a rolling window).

The Short Window

29 September observations is a short calibration window. If September's return distribution was unusually benign or unusually volatile relative to longer-run norms, the 20th percentile threshold it produces may not accurately represent the asset's typical 20th percentile. A 90-day or 180-day calibration window would produce more robust threshold estimates.

What Daily Candles Cannot Show

Daily closing prices obscure everything that happens within each trading session. Two assets can close at their thresholds having taken completely different intraday paths — one might have gapped down at open and recovered, while the other gradually declined throughout the session. "Joint tail loss" at the daily level does not mean the losses occurred at the same time intraday, were experienced by the same investors, or involved any common mechanism.

The analysis explicitly notes: "Daily candles cannot reveal intraday joint-loss paths, investor flows or execution outcomes."

No Trading Signal

The BTCUSA comment concludes with an unambiguous statement: "The comparison offers no trading signal." This means that observing Bitcoin breach its threshold on any given day cannot be used to predict Ether or Solana breach with sufficient confidence for a trading decision, because the sample is too small and the mechanism too unclear to support that application.

What the Finding Does and Does Not Establish

What the analysis does establish:

• In October 2026, using Coinbase USD spot closes and the defined methodology, Ether and Solana breached their September-calibrated thresholds substantially more often within the Bitcoin-breach group than outside it

• The specific dates of joint breaches are precisely identified and fully auditable

• The methodology is fixed, reproducible, and applied symmetrically across all three assets

What the analysis does not establish:

• A causal relationship between Bitcoin's bad days and Ether's or Solana's bad days

• A stable or persistent probability of Ether or Solana breach conditional on Bitcoin breach

• A relationship that would hold across different time periods, calibration windows, data venues, or daily boundary definitions

• Any basis for trading or hedging decisions

The Next Checkpoint

The October evaluation period extends through October 31, 2026. The next checkpoint is November 1 at 00:00 UTC, when the October 31 bucket completes and four additional observations (October 28–31) become available for evaluation.

The September thresholds remain frozen — they will not be recalibrated for any additional data. The extension applies the same fixed thresholds to the four remaining October dates, producing an extended October record.

The continuation test is whether each asset's breach frequency remains higher on Bitcoin-breach dates than on nonbreach dates in the expanded window. If Bitcoin produces additional breach dates in October 28–31 and the other assets do not match them, the conditional frequencies converge; if Bitcoin produces nonbreach dates and Ether or Solana independently breach, it widens the evidence for independent Ether or Solana tail risk. An empty Bitcoin-breach group in the final four days would leave the conditional frequency undefined for that sub-period.

Historical classifications from October 1–27 are fixed by the reported data and can only be changed by corrected input data or arithmetic, not by subsequent observations.

Summary

A threshold-breach coincidence analysis of Bitcoin, Ether, and Solana using Coinbase USD spot closes for October 1–27, 2026 against September-calibrated 20th-percentile daily log-return thresholds finds highly concentrated coincidence: Ether breached its own threshold on all five dates when Bitcoin breached its threshold (100% conditional rate, versus 13.64% on the 22 non-breach dates); Solana breached on four of those five dates (80% conditional rate, versus 0% on non-breach dates). On October 14, Bitcoin and Ether both breached while Solana did not. Ether also showed three independent breach dates (October 2, 24, 25) absent any Bitcoin breach. The finding is limited by the five-observation conditional group, a retrospectively chosen calendar split, a short calibration window, and the fundamental opacity of daily candles about intraday dynamics. The BTCUSA author describes the result as "concentrated coincidence under fixed, asset-specific definitions" and explicitly confirms it carries no trading signal and does not establish a lasting market relationship. The complete October evaluation through October 31 will be available after the November 1 UTC checkpoint.