I have a historical list of index components, so I'm just adding up market caps of those to get a proxy for the index. However, when I check it against Bloomberg, my returns are consistently higher - is there anything else I'm missing? Alternatively, is there a way to get index divisor timeseries from some source?
UPDATE: Thanks to @MatthewGunn for pointing it out: one has to account for the corporate events: on a given day I can compare the total market cap using today float and today's vs yesterday's prices and use this ratio to adjust the overall market cap. That removes jumps in index return series, but still getting much higher returns that the benchmark. Is there anything else I should be aware of?