With the popularity of snapshot mosaic hyperspectral cameras, researchers have shown great interest in hyperspectral video processing in recent years, particularly hyperspectral object tracking. Unlike previous trackers, we train our tracker directly on the two-dimensional hyperspectral raw data instead of hyperspectral cube data. This allows mainstream RGB trackers to be directly transferred to hyperspectral raw data without any modifications. And the same transferer tracker can be applied to handle raw data from different types of snapshot hyperspectral sensors. To address the issue of limited scale in existing hyperspectral tracking datasets, we convert available public RGB tracking datasets into simulated raw hyperspectral data. This ensures that transferred trackers perform well in the hyperspectral raw image domain. In addition, we propose a channel-wise target-aware normalization that is a plug-and-play solution without additional network parameters during inference. The experiments demonstrate that our tracker achieves state-of-the-art performance on hyperspectral tracking datasets composed of data from different types of sensors.