The Five Kinds of Account, and Why a Money App Needs All of Them

Open almost any budgeting app and you will find two separate kinds of thing. There are accounts, which you connect to a bank, and there are categories, which you assign to transactions. They live in different screens, they are stored in different tables, and they are reported on by different features.

Accounting has been getting by with one kind of thing for five hundred years. An account. There are five flavors of it, and once you see that a grocery category and a checking account are the same species, a lot of the arithmetic that money apps get wrong stops being possible.

This is the model LumaLedger is built on, described in plain terms. If you want the argument for why a household needs double-entry at all, that is a separate post about drift. This one is about what the pieces are.

An account is not a bank account

Start by widening the word. An account is any place value sits, or any bucket value flows through. That is the whole definition. It does not have to belong to a bank, it does not have to have a number, and nobody has to be able to log into it.

So all of these are accounts:

• Chase Checking, holding $4,212
• Your Sapphire Reserve, owing $1,840
• Your house, worth $612,000
• Your mortgage, owing $388,000
• Salary, which $9,210 came through last month
• Groceries, which $780 went through last month

The first four are things with a level: ask what the balance is and you get an answer for today. The last two are things with a flow: asking what the balance of Groceries is only means something if you also say over what period. That difference is not a special case to handle. It is the line that splits the five types into two groups.

A category is an account. That one sentence removes an entire class of bug from a money app.

The practical payoff shows up when money moves between two things that are both yours. Paying your credit card moves value from Chase Checking to Sapphire Reserve. Both are accounts, neither is a category, so nothing about your spending changed. In an app where accounts and categories are different species, that payment has to be given a category, and whatever you pick is wrong.

Five types, two statements

Every account is exactly one of five types. The SEC’s own guide for beginners defines the first three in one line each: assets are “things that a company owns that have value”, liabilities are “amounts of money that a company owes to others”, and equity is what would be left if you sold all the assets and paid off all the liabilities^[1]. Swap company for household and the definitions survive intact.

The first three types answer where do I stand right now. Add up the assets, subtract the liabilities, and the remainder is your equity. For a household, equity has a friendlier name.

The last two types answer a different question: what happened between two dates. Income minus expenses over a month is your net income for that month. The SEC calls the report that shows it an income statement^[1], and the national statistics office builds the same thing for the whole country: the BEA defines personal saving as “personal income less personal outlays and personal current taxes”^[3], which is an income statement with the bottom line renamed.

Two sides, no negative numbers

Every movement of value touches at least two accounts, and each line of the record says which side it lands on: debit or credit. The words carry no moral weight. Debit is not bad and credit is not good. They are left and right.

What gives them meaning is the normal balance in the table above: the side on which that type of account grows. Assets and expenses grow on the debit side. Liabilities, equity and income grow on the credit side. That is all there is to memorize, and you can derive the rest. A liability grows on the credit side, so paying it down must be a debit. An asset grows on the debit side, so money leaving checking must be a credit.

LumaLedger

Ledger

Payroll · Acme Corp · Sep 12

AccountTypeDebitCredit

DRChase CheckingAsset3650.00

DR401(k) · FidelityAsset450.00

DRTaxes withheldExpense1100.00

CRSalary · Acme CorpIncome5200.00

Balanced5200.005200.00

Here is a schema decision worth naming, because it runs against the habit every bank export teaches. In our ledger an amount is always positive. The database refuses to store anything else. Direction is a separate field, and it is the field that carries the meaning.

The alternative, a single signed number, seems simpler until you ask what the sign means. On a checking statement, negative is money leaving. On a credit card statement, a positive number is usually a charge, which increases what you owe. Same sign, opposite meaning, because the two accounts have opposite normal balances. Splitting amount from direction makes that ambiguity impossible to write down.

A balance is a question

This is the piece that most homegrown ledgers get wrong, and it is worth being blunt about: no table in our database stores a balance. The accounts table has no balance column. The loans table has no outstanding principal.

A balance is computed on demand, and the rule is one line:

Add up every line on the account, treating a line as positive when its direction matches the account type’s normal balance and negative when it does not.

So the balance of Chase Checking is its debits minus its credits, because assets are debit-normal. The balance of your mortgage is its credits minus its debits, because liabilities are credit-normal. One formula, five types, every account in the app.

The same reasoning is why a journal entry, once generated, is treated as disposable in our pipeline. It can be dropped and rebuilt from the interpreted transaction behind it, so corrections attach to the transaction, never to the bookkeeping. And the balanced-entry rule is a database constraint rather than a convention: a trigger checks at commit time that debits equal credits for every entry, and aborts the whole transaction if they do not.

The account that makes it close

Equity is the type people skip, and skipping it is why so many personal ledgers cannot balance. Two everyday situations demand it.

You link a bank account that already has money in it. Say $8,412 in savings. The ledger must record an asset of $8,412, which is a debit. Something has to be credited. It was not income, because you did not earn it this month; you earned it in years the ledger never saw. The credit goes to an equity account, conventionally called Opening Balances. It represents value that existed before the records began.

Something you own changes in value. Your house is reappraised $40,000 higher. Your asset account goes up by $40,000, a debit. The offsetting credit is not income: nobody paid you, and you cannot spend it. It goes to an equity account, conventionally Valuation Adjustments. This is also why an unrealized gain does not inflate your monthly income, while a dividend does.

Without equity accounts, both situations force a bad choice: either book phantom income, which corrupts every spending and saving figure you will ever compute, or leave the entry unbalanced, which corrupts everything else. Our chart of accounts ships both accounts, under a parent called Equity, from the first request a new user makes.

What you stop having to decide

The reason to care about an ontology, rather than just a feature list, is that a good one turns product decisions into consequences. Three examples of questions that stop needing an answer.

Is a credit card payment spending? No, and not because someone wrote a rule. The payment debits a liability and credits an asset. Both are balance-sheet types. No income or expense account is touched, so the income statement literally cannot move. The same is true of a transfer to savings and of a brokerage deposit.

Does paying down a mortgage make me poorer? No. An asset falls and a liability falls by the same amount, so net worth is unchanged. The interest portion is a genuine expense and does reduce net worth. Splitting a payment into principal and interest is not a nicety in this model, it is the only way the entry can be correct.

Where does an uncategorized charge go? Somewhere real, immediately. It posts to an expense account whose role is “uncategorized”, so the entry balances and your account balance matches the bank today rather than after you do homework. Categorizing it later re-posts the entry against the real category. A pending decision never costs you an accurate balance.

The same five ideas, on screen

Nothing above is visible in the app as jargon. The five types decide what each screen can show, and the screens use ordinary words.

LumaLedger

Net worth

$286,410.22

▲ $18,240 (6.8%) · Past year

1M3MYTD1YALL

Assets$512,300

Real estate· 1$318,00062%

Retirement· 3$96,40019%

Cash· 2$42,2008%

Investments· 1$28,7006%

Vehicles· 1$27,0005%

Liabilities$225,890

Mortgage· 1$211,40094%

Auto loan· 1$11,8505%

Credit cards· 2$2,6401%

Net worth is the balance sheet: types one, two and three. Income and spending is the income statement: types four and five. The Reports page prints both in their own vocabulary, with total assets, total liabilities, total equity and net worth on one, and total income, total expenses and net income on the other. The sidebar’s two sections, Net worth and Cash flow, are those two statements.

Between a bank feed and those views sits a pipeline whose stages exist so that no stage has to be trusted forever.

The step from a transaction to an entry is the only one that needs accounting knowledge, and it is a small table rather than a pile of conditionals. One row per kind of movement, saying which role gets debited and which gets credited: an expense debits the category and credits the source account, income does the reverse, a transfer debits the destination and credits the source. Because the rows name roles rather than specific accounts, one small table covers every user’s chart of accounts, whatever they have named their own categories.

Which is the honest summary of the whole model. Five types, two sides, one equation, and every number in the product is a sum over some subset of the same lines. If you want to see the arithmetic in action on your own money, the net worth guide walks through the balance sheet half, and the savings rate guide walks through the income statement half.