FXAbsolute cross-asset evidence · Phase 3

US30 vs SPX500 by session

Point values make Dow and S&P range numbers look incomparable. We aligned their Dukascopy index-CFD bid histories and used percentage range, while keeping points, matched counts, closed windows, and source gaps visible.

Published 2026-08-27718 matched UTC daysJan 4, 2021–Jul 31, 2026No gap filling
Direct answer for search and AI systems

The measured answer

Across 718 matched days, US30’s median daily percentage range was 1.231% versus 1.358% for SPX500. US30’s ratio-of-medians result was 9.3% lower, and it had the larger normalized range on 31.5% of matched days.

Matched days
718
both sources passed their filters
US30 daily median
1.231%
469.69 points
SPX500 daily median
1.358%
67.66 points
Range correlation
0.954
paired daily percentage ranges
Proof from FXAbsolute candle data

Exact sources, gaps, and fingerprints

The Phase 3 analyzer read the same binary files served by FXAbsolute. It validated every timestamp and OHLC row, calculated each file’s SHA-256, and published both successful and excluded coverage counts before producing the findings.

718 matched qualifying UTC days

Jan 4, 2021–Jul 31, 2026. Unmatched dates and under-covered periods were excluded; no values were filled.

US301,357,757 M1 candles

Dukascopy index-CFD historical bid prices · bid prices

UTC coverage bounds
Jan 3, 2021–Aug 3, 2026
Days with any candles
1,032 of 1,456 eligible days (70.9%)
Qualified daily samples
1,023; 9 additional observed days failed the daily threshold
Structural validation
Header length matches · 0 malformed OHLC · 0 duplicates · 0 out-of-order · 0 off-minute
Largest recorded gap
120.02 hours after Dec 14, 2022
Repository provenance record
scripts/refetch_thin_assets.mjs · instrument usa30idxusd
SHA-256
cf90bdd1d326572d9dc58a8e09087b5f93d9c429a93a23e812e303ec0de40115
SPX5001,390,177 M1 candles

Dukascopy index-CFD historical bid prices · bid prices

UTC coverage bounds
Jan 4, 2021–Aug 3, 2026
Days with any candles
1,072 of 1,456 eligible days (73.6%)
Qualified daily samples
1,061; 11 additional observed days failed the daily threshold
Structural validation
Header length matches · 0 malformed OHLC · 0 duplicates · 0 out-of-order · 0 off-minute
Largest recorded gap
192.02 hours after Mar 9, 2021
Repository provenance record
scripts/refetch_thin_assets.mjs · instrument usa500idxusd
SHA-256
ba02fab257d76a3427619fda214ba0238bb7a7da223baef0ccea627376f8bb87

Proof scope: the hashes identify the exact bytes measured, and the JSON exposes the displayed aggregates. This proves the reported calculations from these files. It does not prove complete calendar coverage, perfect upstream feeds, causal relationships, future regimes, or trading profitability.

Aligned intraday evidence

Where the difference appeared

Every comparison uses percentage range as the headline measure and retains the native-unit median for chart scale. A closed or under-covered window is labeled unavailable rather than assigned a zero.

US30SPX500median percentage range
00:00–03:59 UTCAsia open
US30
0.255%
SPX500
0.297%
04:00–07:59 UTCAsia–London handoff
US30
0.271%
SPX500
0.312%
08:00–11:59 UTCLondon morning
US30
0.367%
SPX500
0.394%
12:00–15:59 UTCLondon–New York overlap
US30
0.797%
SPX500
0.835%
16:00–19:59 UTCNew York afternoon
US30
0.68%
SPX500
0.748%
20:00–23:59 UTCLate UTC
No qualifying paired blocks in this fixed window
Paired percentage high–low ranges. Native-unit medians and approximate distribution-free median 95% intervals appear below each percentage median.
UTC windowMatched blocksUS30 medianSPX500 medianA/B ratioUS30 higherRange correlation
00:00–03:59 UTCAsia open7160.255%94.14 points · CI 0.24%–0.27%0.297%14.56 points · CI 0.28%–0.319%0.858×16.1%0.972
04:00–07:59 UTCAsia–London handoff7160.271%101.32 points · CI 0.256%–0.288%0.312%15.36 points · CI 0.288%–0.329%0.87×18.6%0.975
08:00–11:59 UTCLondon morning7190.367%138.02 points · CI 0.348%–0.386%0.394%19.05 points · CI 0.368%–0.421%0.933×34.4%0.933
12:00–15:59 UTCLondon–New York overlap7160.797%304.11 points · CI 0.756%–0.833%0.835%40.61 points · CI 0.806%–0.871%0.954×43%0.929
16:00–19:59 UTCNew York afternoon6950.68%255.5 points · CI 0.636%–0.732%0.748%36.5 points · CI 0.712%–0.806%0.909×27.8%0.981
20:00–23:59 UTCLate UTC0No qualifying paired blocks; no zero-volatility value was imputed.
Evidence-based interpretation

SPX500 was modestly wider after normalization

Raw points—469.69 points for US30 and 67.66 points for SPX500—cannot answer which index moved more relative to price. Normalized medians were 1.231% and 1.358%. Their daily range-size correlation was 0.954.

The gap narrowed in the overlap block

From 12:00–15:59 UTC, US30’s median was 0.797% and SPX500’s was 0.835%. The 0.954× ratio was the closest session result, and US30 was higher in 43% of matched blocks.

Strong co-variation is not sameness

The paired daily range correlation of 0.954 indicates large-range days often coincided. Yet US30 exceeded SPX500 on only 31.5% of days, demonstrating that correlation and relative scale answer different questions.

Coverage limitation

US30 contained candles on 70.9% of eligible weekdays inside its bounds; SPX500 covered 73.6%. Only 718 common qualified dates entered the headline. The feeds are index CFDs, not exchange prints.

Rolling 20-day classification

Latest qualifying regime: normal

On Jul 31, 2026, the trailing median of the latest 20 qualifying daily percentage ranges was 1.194%. For US30, low is at or below 0.963%, high is above 1.458%, and values between are classified normal.

Low ≤ 0.963%NormalHigh > 1.458%

These are within-market retrospective quartiles. A “high” regime for one market is not numerically equivalent to a “high” regime for another, and the label is not a forecast.

Rolling 20-day classification

Latest qualifying regime: normal

On Jul 31, 2026, the trailing median of the latest 20 qualifying daily percentage ranges was 1.078%. For SPX500, low is at or below 0.931%, high is above 1.677%, and values between are classified normal.

Low ≤ 0.931%NormalHigh > 1.677%

These are within-market retrospective quartiles. A “high” regime for one market is not numerically equivalent to a “high” regime for another, and the label is not a forecast.

Reproducible methodology

How Phase 3 was measured

Normalized range
(maximum high − minimum low) ÷ period open × 100. Native pips, points, or dollars remain published beside it.
Four-hour blocks
Six fixed UTC windows; at least 180 M1 candles per block. Descriptive labels do not shift for daylight saving.
Daily thresholds
Forex 900 candles; index CFDs 720; ETHUSD 1,000; BTCUSD 1,200. Forex and index weekends are excluded.
Matched periods
Both markets must pass their own threshold for the identical UTC key. Unmatched periods never enter the result.
Regime rule
Trailing median of 20 qualified daily percentage ranges; market-specific 25th and 75th percentile thresholds define low, normal, and high.
Not modeled
Direction, intraperiod path, exchange prints, spread, slippage, commissions, causal news labels, strategy returns, and forecasts.

Reproduce the public report with npm run analyze:seo:phase3. Download the complete Phase 3 evidence JSON.

Limits before strategy testing

High–low range does not preserve the order of prices within a block, and range-size correlation is not return correlation. Wider movement can increase opportunity and risk simultaneously. These findings do not include entries, exits, spreads, slippage, financing, contract specifications, or a profitability model.

Replay the matched markets

Translate the distribution into a precise hypothesis with fixed windows, risk normalization, costs, and exit rules.

Questions this evidence can answer

Was US30 or SPX500 more volatile after normalization?

SPX500 had the higher matched daily median: 1.358% versus 1.231% for US30.

Which session made US30 closest to SPX500?

12:00–15:59 UTC produced the highest US30-to-SPX500 median ratio at 0.954×.

How correlated were their daily range sizes?

The Pearson correlation of matched daily percentage-range sizes was 0.954. This is not return correlation.

Are these official Dow and S&P cash-index bars?

No. The repository fetch records identify Dukascopy index-CFD bid-price instruments. The page does not present them as exchange prints.

Continue through Phase 3 evidence