DXOs Lens test results are so useless. They rate it less than the Otus, as they should, however all of the measures they choose to exhibit would otherwise indicate that the new Sigma 50 should be the better lens. Comparatively, it has the same resolution, better transmission, less distortion, and less CA than the Otus. Only in a footnote do you actually learn why DXO rates the Otus higher: It has sharper corners.
Bleh. DXO. Bleh. It's like they just barf up test results and let the chunks & giblets remain where they plop.
I think the world would be well-served if DXO just gave up on lens tests alltogether, nuked their lens tests database, and just stuck with sensor tests. (And furthermore, I think the world would be better served if DXO did away with scalar test "scores"...just as useless as the chunks and giblets that is their lens tests.)
Um... I'm pretty sure DxO scores are based on a specific "best" setting, where the "best" setting represents the highest achievable score on a given lens. The sigma score is based on f/2, and the Otus f/1.4. They both acheive similar sharpness at that setting, however that gives the Otus a full stop advatage on toward the score. You should maybe figure out how they score before trying to put down their scoring method......
If you look at all of DXO's lens tests, it's very clear that they are VERY HEAVILY weighted based on the T-stops value. That's a problem, because it makes comparing the overall quality of lenses with different maximum apertures practically impossible...it's why a 50mm f/1.4 lens scores higher than Canon's 600mm f/4 L II supertele. The 50/1.4 doesn't even come remotely close to comparing to the 600/4 II, but it scores quite a bit higher. Why? Because it's f/1.4, and the 600 is f/4.
The T-stops weighting effectively nullifies much of the value that could potentially exist in DXO's lens tests. There is a certain value to testing lenses that way...but not if the most important benchmark is T-stops.
BTW, DXO does not mix ratings for measures from different apertures. If they choose the measures for a lens when it is tested at f/1.2, then that means sharpness, distortion, vignetting, and CA are all based on the testing at that aperture. As far as I know, DXO tests all lenses at all of the full-stop apertures (plus the max aperture, in case it isn't a full stop faster), but when they score, it's based on whatever they deem is the "best" aperture. Since they put such a significant weight on T-stops, that is usually the maximum aperture, although not always.