Best way to parse and model formulas with dependencies on each other in a database and evaluate them given a set of data

Let's say we have a bunch of user defined formulas that depend on other formulas that look something like this (this could be any syntax).

Example: Give a Customer with multiple accounts

ExampleOne: Accounts.Sum() - ExampleTwo 
ExampleTwo: Accounts.CurrentAssets - 200 / CurrentYear 
...etc

I want to design a service that given a customer with accounts, evaluate these formulas and return the result for each formula item:

ExampleOne: 123
ExampleTwo: 999
...etc

What is the best approach about this given the following requirements within the c# .net environment. Please note that how the syntax would be is not important here):

• These formulas will be frequently changed by non technical people
• There are hundreds of these formulas that have depenedencies on each other (Acyclic)
• These formulas need to be stored in database
• Calculations performed to evaluate these formulas must be fast

I am trying to evaluate several options:

Graph Database

Parsing the formulas as an abstract Syntax Tree and storing them into a graph database. However, with this approach, it seems difficult to update formulas after creation.

Evaluate formulas as runtime

I could write a parser that would take a formula and evaluate it at runtime.

Rules Engine

I saw this similar question. But I'm not sure if this is the right approach for this.

Relational Database

How would I model formulas in the database with the flexibility of easily changing them?