Prediction Records

Every forecast, written down before the outcome

Each trading day the Regime Radar publishes its chance that the composite market regime changes within five sessions. The day it is published, that forecast is recorded here and never edited. Five sessions later the record shows what happened. Each week's Golden Ticket basket is recorded too, as a fingerprint written before its week closes and checked once its names are published.

Records since 25 Sep 2026

How a record works

Written once: the first time a day's forecast is served on the live site, it is stored with the date and time, and it is never changed afterwards.

Resolved against the record: once five later sessions are published, the record reads Changed if the composite regime differed from the recorded state on any of them, Held if it did not, and Unmeasured if any of them has no reading. The outcome is then written beside the forecast and kept.

Calibrated in the open: when enough records have resolved, the average forecast is shown beside the share that actually changed, with the Brier score.

The records

RecordDateChance of a change within 5 sessionsComposite regime thenIf it changesOutcome
#03176.1%CrisisToward risk-offOpen
#03035.5%CrisisToward risk-offOpen
#02976.1%Crisis—Open
#02876.1%Strong risk offToward risk-offOpen
#02776.1%Strong risk offToward risk-offOpen
#02676.1%Strong risk offToward risk-offChangedto Crisis on 4 Nov 2026
#025100%Strong risk offToward risk-offChangedto Crisis on 4 Nov 2026
#02476.1%Strong risk offToward risk-offChangedto Crisis on 4 Nov 2026
#02376.1%Strong risk offToward risk-offHeld
#02276.1%Strong risk offToward risk-offHeld
#02176.1%Strong risk offToward risk-offHeld
#02076.1%Strong risk offToward risk-offHeld
#01976.1%Strong risk offToward risk-offHeld
#01876.1%range 63.3%–100%Strong risk offToward risk-offHeld
#01779.2%range 70%–100%Strong risk offToward risk-offHeld
#01676.1%range 60%–92.9%Strong risk offToward risk-offHeld
#01576.1%range 61.1%–93.7%Strong risk offToward risk-offHeld
#01476.1%range 61.1%–93.2%Strong risk offToward risk-offHeld
#01376.1%range 60.7%–93.2%Strong risk offToward risk-onHeld
#01276.1%range 60.7%–93.2%Strong risk off—Held
#01176.1%range 60.7%–93.2%Strong risk offToward risk-offHeld
#01014.4%range 4.8%–26.9%Moderate risk onToward risk-onChangedto Strong risk off on 9 Oct 2026
#00914.4%range 4.8%–26.9%Moderate risk onToward risk-onChangedto Strong risk off on 8 Oct 2026
#0084.7%range 0%–11.1%Strong risk offToward risk-offChangedto Moderate risk on on 7 Oct 2026
#0074.7%range 0%–11.1%Strong risk offToward risk-offChangedto Moderate risk on on 6 Oct 2026
#00610.3%range 5.2%–15.9%Strong risk offToward risk-offChangedto Moderate risk on on 5 Oct 2026
#00510.3%range 5.2%–15.9%Strong risk offToward risk-offChangedto Moderate risk on on 2 Oct 2026
#00410.3%range 5.4%–15.9%Strong risk offToward risk-offChangedto Moderate risk on on 1 Oct 2026
#00312.3%range 6.9%–18.2%Strong risk offToward risk-offHeld
#00210.3%range 5.2%–15.9%Strong risk offToward risk-offHeld
#00110.3%range 6.1%–16.8%Moderate risk onToward risk-onChangedto Strong risk off on 28 Sep 2026
The records, as JSON →

Calibration

Resolved records
26
Average forecast
51.8%
Share that changed
42.3%
Brier score (0 is perfect)
0.546

Golden Ticket baskets: committed before the week closes, revealed after

Each Monday a new Golden Ticket basket is picked by rule, and its names are withheld from the public pages until the week closes (signed-in accounts see them that night). The first time the live site reads the new basket, its fingerprint (a SHA-256 hash of every name, side and weight, with a random salt) is recorded here with the time. The salt stays private while the week is open, so the names cannot be worked out from the fingerprint. When the week closes and the basket is published, the salt is published too and the fingerprint is computed again from the published basket. Had a single name, side or weight changed, the two would differ.

RecordWeekCommitted atFingerprint (SHA-256)Outcome
#W-0052026-W452 Nov 2026 to … · 10 nameseb2cb936cb…10992cOpen
#W-0042026-W4426 Oct 2026 to 2 Nov 2026 · 10 names2f2b309f05…87faa8Revealed: matchesrecomputed 2 Nov 2026: 2f2b309f05…87faa8salt: 5f6ef39b4a…bd17c1
#W-0032026-W4319 Oct 2026 to 26 Oct 2026 · 10 namesdf4732ff13…ad0747Revealed: matchesrecomputed 26 Oct 2026: df4732ff13…ad0747salt: 8788624900…eae4c2
#W-0022026-W4212 Oct 2026 to 19 Oct 2026 · 10 names0ecbbe3257…04662eRevealed: matchesrecomputed 19 Oct 2026: 0ecbbe3257…04662esalt: 13d95cd0bc…51799b
#W-0012026-W415 Oct 2026 to 12 Oct 2026 · 10 names3a4edfbb98…baf21cRevealed: matchesrecomputed 12 Oct 2026: 3a4edfbb98…baf21csalt: 71cd24aeee…e66737

Verify it yourself

  1. When the week closes, take its basket from /api/public/golden-tickets: the week whose slug matches, and each name's symbol, side and weight_pct.
  2. Write each name as [symbol, side, weight], the weight with exactly six decimals, rounded half up. Sort the list by symbol, then side, then weight. Put it under the basket's name, beside the record's salt from /api/public/prediction-records: {"baskets": {"equal": list}, "salt": salt}.
  3. Write that as JSON with no spaces and take the SHA-256 of its UTF-8 bytes, in lowercase hex. When the published basket is the one that was picked, it equals the fingerprint recorded before the week.

The canonical text looks like this:

{"baskets":{"equal":[["AAPL","LONG","10.000000"],["MSFT","SHORT","10.000000"]]},"salt":"0f1e2d3c4b5a69788796a5b4c3d2e1f0"}

In Python:

import json, hashlib, urllib.request
from decimal import Decimal, ROUND_HALF_UP
get = lambda p: json.load(urllib.request.urlopen("https://opulencealpha.ai" + p))
rec = next(r for r in get("/api/public/prediction-records")["gt_weeks"]["records"] if r["week"] == "2026-W44")
week = next(w for w in get("/api/public/golden-tickets")["weeks"] if w["slug"] == "2026-W44" and not w["open"])
six = lambda x: None if x is None else str(Decimal(x).quantize(Decimal("0.000001"), ROUND_HALF_UP))
rows = sorted([l["symbol"], l["side"], six(l["weight_pct"])] for l in week["legs"])
text = json.dumps({"baskets": {"equal": rows}, "salt": rec["salt"]}, sort_keys=True, separators=(",", ":"))
print(hashlib.sha256(text.encode()).hexdigest() == rec["forecast"]["fingerprint"])
The commitments, with the canonical form, as JSON →

The full record

Every closed Golden Ticket week is published in full on the Refutation Ledger: each name with its side, weight, entry, exit and return.

The Refutation Ledger →

Forecasts of the market regime, never of a price. The Golden Tickets are the platform's weekly benchmark engine, not a trading suggestion. Not investment advice.