Morning Brief: WTI at $99.25 Meets Crude Inventories and the Rate Test

An event window specifies which observations count as evidence. MacKinlay explains that announcements can arrive after the market closes, and that uncertainty about the event date can justify a two-day window. The lesson is about information timing. A window centered on a calendar label need not match when participants could react. Compare two explicitly editorial designs. A symmetric window includes sessions minus one, zero, and plus one. A forward window includes zero, plus one, and plus two. Both contain three sessions, but the first captures possible anticipation while the second allows more time for a delayed response. Keep the same abnormal-return calculation and success threshold in both. Suppose abnormal returns are 0.8, 0.1, 0.2, and negative 0.5 percent over those four sessions. The symmetric sum is positive 1.1 percent; the forward sum is negative 0.2 percent. This disagreement is informative about the chosen measurement interval. Selecting the first window only after seeing these values would hide that choice. For a scheduled cycle mark, state that both shapes are research alternatives; the announcement literature does not establish which one a Gann date deserves.

Try it

Synthetic abnormal returns for sessions -1, 0, +1, +2 are +0.8%, +0.1%, +0.2%, -0.5%. Using arithmetic cumulative abnormal returns, compute sums for [-1,+1] and [0,+2]. Which meets a predeclared +1% threshold?

Limits

This example does not recommend a winning window. Testing both is an additional specification choice that must be disclosed. MacKinlay's event-date discussion is not evidence that price reacts to calendar geometry.

Source

A. Craig MacKinlay, Event Studies, Section 2, announcement timing, printed pp. 14-15; Section 10.B, Inferences with Event-Date Uncertainty, p. 35. Peer-reviewed Journal of Economic Literature 35(1), 1997, pp. 13-39. Article reproduced in a public court exhibit; checked against the Boston University scan. Its earnings illustration covers January 1989 through December 1993. Timing-rule adaptations below are editorial, not Gann methods.

Today

At the Oct 18 UTC event mark, BTC versus ETH daily-close relative returns on Oct 17, 18, 19, 20 were +0.687 pp, +0.446 pp, -0.105 pp, -0.382 pp. The Oct 17–19 symmetric window sums to +1.028 pp; the Oct 18–20 forward window sums to -0.042 pp; their difference is +1.069 pp. The Oct 20 completed BTC daily close was $82,938.26; the Oct 21 candle is not complete. Both definitions are fixed before interpreting their disagreement and neither predicts a market turn. BTC-USD versus ETH-USD daily-close return proxy, Latest session 2026-10-21T00:00:00Z: sum(100*(BTC_close_t/BTC_close_t-1-1)-100*(ETH_close_t/ETH_close_t-1-1), t=Oct 17..19) - sum(same, t=Oct 18..20) = 1.069264 percentage points (symmetric minus forward) (inputs: btc_oct16_close_usd=81500.04, btc_oct17_close_usd=81378.21, btc_oct18_close_usd=82006.85, btc_oct19_close_usd=81837.69, btc_oct20_close_usd=82938.26, eth_oct16_close_usd=2419.92, eth_oct17_close_usd=2399.68, eth_oct18_close_usd=2407.51, eth_oct19_close_usd=2405.08, eth_oct20_close_usd=2446.62, forward_0_plus2_pp=-0.0415, symmetric_minus1_plus1_pp=1.027764). Data source 1 Data source 2

A geocentric Moon position uses Earth's center as the observer. A topocentric position uses a site on Earth's surface. The vector from that site to the nearby Moon differs from the vector drawn from Earth's center. This is parallax, not a change in the Moon's orbit and not an atmospheric correction. Horizons lets the user specify an observing site on a body with a known rotation model. To isolate the site effect, request the same target, instant, coordinate type, and frame for both origins. Do not compare an apparent topocentric result with an astrometric geocentric result and call the whole difference parallax. Suppose two matched hypothetical longitude rows read 100.20 degrees at the geocenter and 99.55 degrees at one site. The signed site-minus-center change is minus 0.65 degrees. A contact with a fixed 100-degree reference would be 0.20 degrees away in the first row and 0.45 degrees away in the second. A 0.30-degree tolerance admits only the first. This is why a local lunar timing claim needs an actual observer location.

Try it

Using the supplied rows, find the signed longitude change and classify both against the 100-degree reference with a 0.30-degree inclusive tolerance.

Limits

The numerical rows are hypothetical. Longitude differences do not equal full angular sky separation. Topocentric geometry supplies no evidence that one exchange responds differently to the Moon.

Source

JPL Horizons manual, Introduction, user-defined topocentric observing sites; General definitions, Center; observer quantity 31. Primary institutional astronomy reference. Worked examples are explicitly hypothetical, not observed market effects.

Today

At the completed Coinbase BTC-USD 10:55–10:56 UTC candle close (USD 83326.45), JPL Horizons Moon quantity 31 gives apparent ecliptic-of-date longitude 329.1352695° from Earth's geocenter and 328.6332286° from the New York site (74.006° W, 40.7128° N, 0.010 km). The wrapped site-minus-geocenter longitude is -0.5020409° at the same instant. This isolates the observer-location difference in matched sky coordinates; it is not the Moon's full angular separation and does not establish any BTC return predictability. BTC-USD, As of 2026-10-21T10:56:00+00:00: ((site_lon_deg-geocentric_lon_deg+180)%360)-180 = -0.5020409 degrees (inputs: candle_granularity_seconds=60, candle_start_unix_s=1792580100, geocentric_lon_deg=329.1352695, price_usd=83326.45, site_altitude_km=0.01, site_latitude_deg_north=40.7128, site_lon_deg=328.6332286, site_longitude_deg_east=-74.006). Data source 1 Data source 2 Data source 3