Forecast Accuracy

OCBC Research — forecast accuracy

Settled 12Awaiting prices 3Direction 33.3%Valid direction sample 12Excluded 366Mean target error —

Every forecast below was published publicly by OCBC Research with a stated target and date. Tlines records the first same-source observation on or after the target date, marks the call settled, and does not adjust the record afterwards. Calls that could not be settled are counted as excluded above rather than dropped silently. Settlement method · Dispute a record

Settled forecasts

Asset	Forecast	Actual	Direction	Error	Target date
Gold	—	4,348.4618	Miss	—	09/12/2026
Gold	—	4,348.4161	Correct	—	09/11/2026
Gold	—	4,317.245	Miss	—	09/10/2026
Gold	—	4,428.9981	Correct	—	09/05/2026
Crude Oil (WTI)	—	94.21	Miss	—	09/05/2026
Crude Oil (WTI)	—	92.69	Miss	—	09/04/2026
Crude Oil (WTI)	—	92.69	Correct	—	09/04/2026
Gold	—	4,474.1535	Miss	—	09/03/2026
Crude Oil (WTI)	—	92.55	Miss	—	09/03/2026
Gold	—	4,326.4733	Miss	—	09/02/2026
Crude Oil (WTI)	—	87.03	Correct	—	08/31/2026
Gold	—	4,448.8013	Miss	—	08/31/2026

Wilson 95% interval: 13.8%–60.9%. 0 repeated same-target report(s) are excluded from independent metrics.

89 rate and macro calls are excluded from this score. "Bullish on 10Y Treasuries" means yields fall, and a policy or inflation stance has no price to settle against, so scoring them against the available series would invert the verdict rather than measure it.

Method: first same-source observation on/after target date, with a matching observation on/before forecast date; maximum gap is configurable and defaults to 7 days.