PrecisionEvaluate() would be for math equations to stop binary to decimal conversion errors.
Follow to LDEV-2863 bug and you will see my rounding test program.
numberformat() seems to be the best to round.
Lucee 6 fixed most of the problems. I think I see one error from trycf.com for Lucee 6. I guess I will look into it.
ACF still isn’t very good either.

Thanks for the reply, kabutotx. However, with numberformat() , there is another range of numbers that are not accurate when using this function. For example, 237453.525 will return 237453.52, not 237453.53