While global climate modeling algorithms are able to capture the mean behavior of the earth's climate system, sub-scale processes are usually unable to be resolved. In particular, Sea Surface Temperature (SST) anomalies can represent information that is not captured in model climatologies or reveal unaccounted changes in climate dynamics. Historically, these anomalies have been studied in a Linear Inverse Model (LIM) framework with some success in analyzing monthly, regional anomaly predictability. In this framework, parameters and uncertainties have typically been given point estimates. In this talk, we first examine the decadal predictability of global SST anomalies and discuss the ways in which typical point-estimates are unreliable in this setting. Next, we show how using a Bayesian setting can incorporate model assumptions and improve parameter estimates and forecast skill. Finally, we explore other statistical techniques to model global SST anomalies.