A unit-sum SST filter reduced 2024-2025 change error at three intervals at 30 moorings, but gave no clear warm-event overlap gain

We tested whether one SST correction can improve buoy temperature changes at several intervals and improve warm events. We fitted five-term unit-sum filters on 2022-2023 paired NOAA OISST v2.1 and buoy records, then scored 2024-2025 at 30 moorings. A filter with two future SST days reduced median station squared change error by 8.6%, 10.1% and 7.1% at 1, 7 and 30 days. Conditional spatial-group 95% intervals excluded unchanged error at all three intervals. One-day training supplied the same weights for all intervals. The constraint preserves constant input and the slope of linear input. A filter using current and past SST gave a 1.1% daily gain, but weekly and monthly intervals included unchanged error. The symmetric filter changed warm-event day overlap by only +0.0020 (0.0000 to +0.0238). Separate years and threshold and penalty checks gave no clear positive overlap gain. The result supports a common corrected-record benchmark, but does not validate corrected heatwave trends. Buoys feed OISST, only four tropical sites enter, and the test has two years. The warm-event diagnostic uses a short training reference, not a 30-year marine heatwave climatology. Filters and years are fixed in uncertainty estimates. Code, aggregate scores, exact source hashes and checks are included; daily records remain in the cited external archive.