The issue is too pervasive, having a formula published won't solve it.

You will need the historical formulas, as well as all the underlying data, not even accounting for the publication delays and errors, missing informations, discretionary classifications, etc.

What we need is all that data and code in an introspectable flow of code and data:

<downvote>a blockchain, containing both your data and the smart contract used to perform the computations</downvote>.