if you are going to use those markers for SDS-PAGE then you can select commonly found proteins in pure form and mix them 1:1 or with different ratio if you want to emphacise any particular mol weight. add loading buffer and they are ready. freeze at -80 and aliquots at -20 for frequent use. its simple.
You need cheap, stable, non-glycosylated proteins free of protease contamination whose electrophoretic mobilities are not aberrant and are evenly spread across the range of molecular weights spanned by your samples.
Honestly, I prefer to buy premade markers and then worry about the million other things that can and do go wrong in the lab, but if short of money/in a hurry (we've all been there), a mixture of egg lysozyme, BSA, beta-galactosidase and any monoclonal antibody will do (you'll get 14 kDa, ~25 kDa, ~50 kDa, 67 kDa and 116 kDa, plus good controls for how well your DTT/bME is working -the BSA and mAb heavy and light chains- as well as a positive control for HRP- or AP-conjugated secondary antibodies, if doing Western blots). It will look ugly as hell, but will get the job done.
Usually a laboratory works mainly on 2-3 kinds of protein. So, when you or any other lab member purifies these for their experiments, generally we don't use most of these proteins, which get wasted. You can pool these different size of proteins and can use these to make marker. In this way you can ascertain the size of homologous protein in future with more accuracy than with conventional markers and other protein with somewhat equal accuracy (conventional markers) .