Personal finance app database design

Question

After searching around for a minimalistic money tracking/budgeting app, I decided to build one for my own personal use.

However I'm unsure with part of the database design. Basically at the moment, I have an entries table which obviously stores data about each transaction, whether is credit or debt etc.

The dilemma that I have is, I don't know if I should create another table to store the current balance of each account or if I should populate it dynamically by subtracting the debits from the credits.

Part of me is saying that as the entries table grows the ability to generate the balance for each account will get slower (yes premature optimization is supposedly evil), but it also seems unnecessary to add another table when I can calculate the data from existing tables.

Thanks

EDIT: Sorry I may not have been clear, I understand how to implement either method of creating the account balance. I was more looking the advantages/disadvantages of either method as well as what would be the 'best practice'. Thanks very much for the replies!

Answer 1

If I were to design a minimalistic accounting application, I would probably do something like

ledger
-------------
   key          INT(12) PRIMARY KEY
   account_id   INT(10)
   category_id  INT(10)
   trans_type   CHAR(3)
   amount       NUMERIC(10,2)

account
------------
   account_id   INT(10) PRIMARY KEY
   created      DATETIME
   name         VARCHAR(32)
   ...

category
------------
   category_id  INT(10)
   name         VARCHAR(32)
   ...

The column key would consist of a date and a zero-padded numeric value (i.e. 201102230000) where the last 4 digits would be the daily transaction id. This would be useful to track the transactions and return a range, etc. The daily transaction id 0000 could be the account balance at the beginning (or end) of the day, and the id 0001 and up are other transactions.

The column trans_type would hold transaction codes, such as "DEB" (debit), "CRE" (credit), "TRA" (transfer) and "BAL" (balance), etc.

With a setup like that, you can perform any kind a query, from getting all the "credit" transactions between any given date, to only the account balance at any given date, or date range.

Example: fetch all credit and debit transactions between 2011-01-01 and 2011-02-23

SELECT ledger.*, account.name, category.name
  FROM ledger
  JOIN account
    ON ledger.account_id = account.account_id
  JOIN category
    ON ledger.category_id = category.category_id
 WHERE (ledger.trans_type = "CRE"
     OR ledger.trans_type = "DEB")
   AND ledger.key BETWEEN 201101010000 AND 201102239999
 ORDER BY ledger.key ASC

Example: fetch all transactions (except balances) between 2011-01-01 and 2011-02-23 for the account #1 (ex: Mortgage)

SELECT ledger.*, account.name, category.name
  FROM ledger
  JOIN account
    ON ledger.account_id = account.account_id
  JOIN category
    ON ledger.category_id = category.category_id
 WHERE ledger.trans_type <> "BAL"
   AND ledger.key BETWEEN 201101010000 AND 201102239999
   AND account.id = 1
 ORDER BY ledger.key ASC

So there you go, flexibility and extensibility.

Answer 2

For a personal financial database today's relational database systems are plenty fast enough to calculate the balance of multiple accounts dynamically. You don't need a column to hold the current balance. Even Microsoft Access is fast enough. I know this because I built and use a personal financial database in Access. It might even be what you were originally looking for. You can read about it and download it at http://maiaco.com/software/ledger/index.php

Answer 3

I am actually working on just this website idea right now and the way I've setup my database is:

TABLE account
    id
    account_name
    current_balance

TABLE transaction
    id
    account_id
    payee
    date
    amount
    category

And whenever a new transaction is added I update the account's current balance.

FYI, I hope to launch my site within a month and if you're interested in using person's site, just check out my profile.