The jersey looks sharp, but it is all low-frequency detail, with crisp, clearly delineated edges on the decals/letters/numerals. I can see the eyelashes are rendered individually, but they are not well-defined, so there was good focus at the eye distance, but not much definition of the detail, so my feeling is that mostly, the ISO 5,000 is what hurt this shot the most. The shirt looks sharp because it relies on color-contrast, not fine details, but the skin and hair and eyes lack finely-rendered high-frequency details, so those things appear a bit soft.
Eye detection has become astoundingly good...there's a computer and software underlying and supporting the focusing system in modern cameras; switching various camera operating system features to OFF is at times, counter-productive, and slows us down. There's a rather pervasive on-line photo forum bias against using all of the technology modern cameras offer, with the underlying premise being that a human can perform complex,precise,critical functions faster and better than a dedicated computer can. Sometimes it makes sense to go it by hand-and-eye, while at other times, the computer, the software, and the hardware can out-perform the human.