Tesseract on Windows: Outputs are always innacurate

Question

So I'm trying to OCR this image: (This are actually usernames)

using this command on the Windows command prompt:

tesseract screenshot.png out 

However the out.txt always brings innacurate results. In this case it is:

belsahx

What am I doing wrong? Is it the white text on a black background bringing problems. Im using the English language pack by the way.

To confirm the problem, this image:

produces this:

1akvDZ\ 5