It would be nice if the math used was high precision. I have run into issues with this. You may want to use [big.js](https://github.com/MikeMcl/big.js)
