> Do you really need full double-entry accounting for personal finances?
Probably not, which is why I'm not actually using it right now. My main use cases are keeping track of various types of assets and liabilities (bank accounts, stocks, apartment deposit, etc) to track my net worth as well as having a general idea of where my money goes, e.g how much I spend on groceries as opposed to eating out.
The statistical sampling approach sounds very interesting, thanks for the tip!
I would love to see an accounting system that was designed for this kind of uncertainty or incomplete data in general. For instance, allow me to say that there's about $5000 in the odd account with no API access or that I spend about half the cash I withdraw from the ATM on restaurants.
In tools that I've tried, they've all been finely tuned to down-to-the-penny precision and capturing all data.
- you need it for tax/legal/bureaucratic purposes, or
- you are really short on money and you need to keep track of it down to the last cent.
When either of those two conditions are true, double-entry is amazing.
If the above are not true (and it sounds that way based on your comment) then I would do it statistically.
Determine how much time you would like to spend recording transactions. Let's say it's 1/11 of the time it takes when you are doing a full accounting. (Eleven is good because it's prime, so it's less likely to accidentally line up with cyclic transactions.)
Enter all expenses/incomes only every eleventh day. Nothing the other days. You have to be really strict about this!
Instead of entering the actual number, though, multiply it by 11.
Tadaa! After a while, you have practically the same result but at a fraction of the effort (roughly three times per month instead of 30.)
Note that the number you get will have some sampling error. For bonus points, you can calculate the size of this. Depending on how large the variation of your expenses is and how often you enter them, it's not unreasonable to get a sampling error where three standard deviations are ±50 % -- are you okay with a rough estimation like that? If not, you have to sample more often.
More technical note: what I've described is actually a stratified sampling scheme, only once you've sampled a day, you do a full accounting of the transactions within that day. Depending on the practical structure of your transactions, you might find that you can optimise your sampling scheme (lower sample error and/or less effort) by selecting more days, but only entering one transaction from each day.
Read up on sampling! It's powerful!