I decided I wanted to do the multiple file approach, in my case I keep one “ledger”[0] file for each bank account, credit card, AR / AP account (or whatever).
I wrote some helper functions (in Emacs Lisp) to facilitate changing ledger files easily, do transfers, set/change default ledger file and/or “entity” (company or person, etc.).
One of my favorite is probably the “snarf” function which will relocate the txn at point to the appropriate ledger file and date location.[1]
This one probably needs more cleanup (well, just need to write some documentation, really) than others to publish, but I could perhaps make the effort if there is any interest.
[0] Sorry, I realize this is probably an overloaded term, but not sure what else to call it, within the context of my package.
[1] Yes I realize date order does not “really” matter to the Beancount parser, but I’m OCD and like to keep my files in order anyway.
I think what I do is pretty typical (using an intermediate ‘Transfers’ account) perhaps best illustrated by a simple example:
# Beancount
2025-12-20 ! "(self)" "transfer A to B" ^xfer_A_B_2025-12-20_100.00
date-txn: 2025-12-20
date-entered: 2025-12-20
source: "my-beancount-transfer-insert"
Assets:CapitalOne:Checking-A -100.00 USD
Assets:Transfers
2025-12-20 ! "(self)" "transfer A to B" ^xfer_A_B_2025-12-20_100.00
date-txn: 2025-12-20
date-entered: 2025-12-20
source: "my-beancount-transfer-insert"
Assets:Chase:Checking-B 100.00 USD
Assets:Transfers
Some things to point out (in case anyone is not familiar with Beancount syntax):
^xfer_A_B_2025-12-20_100.00
This is a “link” in Beancount (which links the 2 transactions together). I made a simple function to dynamically generate this (based on account “letter” shorthand, date, and amount, which combine to make a “unique enough” identifier).
I also include some metadata, which I feel is relevant (of course you don’t have to).
Above is generated by an (Emacs Lisp) function (my-beancount-transfer-insert) which prompts for “from account”, and “to account” (using completing-read), and amount, and then generates the txns accordingly. It then calls the bmf-txn-snarf-refile function (from this library) (twice) which relocates them to the appropriate location (by ledger file and date) which I find really handy.
