Coding Typing Test

Real code in six languages, with symbol accuracy and code-specific mistakes reported.

Six languagesJavaScriptPythonSQLSymbols tracked

Loading coding passages...

Typing code is a different skill from typing prose. The letters are the easy part; the symbols, capitals and line breaks are where speed goes. This hub mixes all six languages, measures each one, and shows how a coding score is calculated, so you know which part of your typing to work on.

What the hub serves

Ten passages drawn from the six language tests: 2 JavaScript, 1 TypeScript, 2 Python, 1 Java, 2 SQL, 2 HTML & CSS. Each opens with a comment naming its language, so you can see what is coming. Here is the start of the first one:

JavaScriptCoding Hub 1 - Parser Profile Audit
// Mixed coding hub passage 1
// Type this focused javascript snippet from the main coding practice page.

const profileEvents = [
  { id: 101, actor: 'Asha', type: 'draft', delta: 14, value: 33 },
  { id: 102, actor: 'Ravi', type: 'verified', delta: 21, value: 44 },
  { id: 103, actor: 'John', type: 'verified', delta: 28, value: 55 },
  { id: 104, actor: 'Noah', type: 'draft', delta: 35, value: 66 },
  { id: 105, actor: 'Isha', type: 'verified', delta: 42, value: 77 },
  { id: 106, actor: 'Kabir', type: 'verified', delta: 49, value: 88 },
];

function parseProfileEvent(event) {
  return {

Why code typing is its own skill

We measured the passages behind every test on this site with the same rule the coding scorer uses for what counts as a symbol. Ordinary English prose averages 2.6 symbols per 100 characters, mostly full stops and commas. The coding passages run from 6.5 (SQL) to 17.2 (HTML & CSS), between 2.5 and 6.6 times as many. Most of those symbols sit off the home row, and many need Shift.

Code also breaks lines constantly: prose passages hardly do, while every code passage averages a line break every few dozen characters. That is why a typist who is fluent on prose can be noticeably slower on code without being any less skilled with letters. The skills to train are different.

Each language's passages, measured. Every figure is recalculated whenever the site is built.
LanguageSymbols / 100 charsMost frequent runCapitalsAvg lineAvg name
HTML & CSS17.2</2%37 chars5.5 chars
Java13.7()4.4%39.1 chars6.9 chars
JavaScript16.9=>2.4%36.6 chars6.2 chars
Python14.3",0.5%40.3 chars5.8 chars
SQL6.5--22%32.6 chars6.3 chars
TypeScript16.1',4%53.4 chars6.5 chars
English prose (for comparison)2.6—2.1%——

How a coding score is calculated

Speed uses the standard rule that five characters make one word, counting every character you type, including spaces and line breaks. Net WPM then takes off one whole word for every error. Because code errors cluster around symbols, a handful of bracket or quote mistakes can cost more speed than they seem to. Here is a worked example of a three-minute run:

Worked example: a three-minute run on a coding passage.
Characters typed1,050 in 3 minutes
Gross WPM1,050 ÷ 5 ÷ 3 = 70
Errors18, of which 11 were on symbols
Net WPM(210 words − 18) ÷ 3 = 64
Accuracy(1,050 − 18) ÷ 1,050 = 98%
Symbol accuracy160 symbols in the passage so far, 149 correct = 93%

Overall accuracy of 98% looks excellent, but symbol accuracy of 93% shows that most of the 6 lost WPM came from symbols: 11 of the 18 errors. That gap is the most useful number on the coding report card, and the reason it is reported separately. The report card also counts bracket, quote, semicolon, whitespace and new-line mistakes on their own, and lists the keys you mistyped most.

How indentation and Enter work

Indentation is part of every passage, but you do not type it. Press Enter at the end of a line and the next line's leading spaces are filled in, the way a code editor indents for you. Tab inserts two spaces, and a space typed where the passage expects a line break is ignored, so a reflexive space at the end of a line does not cost you. What does count is Enter itself: it has to come exactly where the passage ends the line.

Every field on the coding report card

When the timer ends, the report card shows these figures, the passage you were given and your typing side by side. Each mistake is counted once overall and, where it applies, once more in the category of the character the passage expected, so a missed closing quote is both an error and a quote error.

FieldWhat it measures
Net WPMCharacters typed ÷ 5, minus one word per error, per minute. The headline speed.
Gross WPMThe same without the error deduction: raw output.
CPMCharacters per minute, the same raw output without dividing by five. Useful for code, where “words” are a poor unit.
AccuracyCharacters typed correctly as a share of all characters typed.
Symbol accuracyOf the symbols you have reached in the passage (brackets, quotes, operators, punctuation), the share typed correctly.
Bracket errorsMistakes where the passage had ( ) { } [ or ].
Quote errorsMistakes where the passage had a double quote, single quote or backtick.
Semicolon errorsMistakes where the passage had a semicolon.
Whitespace errorsMistakes where the passage had a space or a tab, usually inside a line.
New-line errorsMistakes where the passage ended a line: Enter missed, or pressed early.
Main mistakesThe keys you got wrong most often, with what you typed instead.

What each duration covers

The coding tests run for 1, 2, 3 or 5 minutes. How much code that is depends on the language, because line lengths differ. At a steady 40 WPM (200 characters a minute) a five-minute run covers about 1,000 characters, which works out as follows:

Approximate lines of each language's passages typed at 40 WPM, from measured characters per line.
Language1 minute3 minutes5 minutes
HTML & CSS5 lines15 lines25 lines
Java5 lines14 lines24 lines
JavaScript5 lines15 lines26 lines
Python4 lines13 lines22 lines
SQL6 lines18 lines29 lines
TypeScript4 lines11 lines18 lines

Use one minute to warm up or to test a single habit, such as closing brackets in order. Use three to five minutes for a score worth comparing from week to week: short runs flatter a fast start and hide whether your accuracy holds up. Whatever you choose, keep it the same between sessions so the numbers are comparable.

Which language test to practise next

After a hub run, look at the report card and match your biggest mistake to the language that trains it hardest. The routes below follow from the measurements in the table above.

If your report showsPractiseWhy
Low symbol accuracy overallHTML & CSSThe most symbols per 100 characters of any test here.
Bracket errorsJavaA brace or parenthesis on almost every line, often several closing at once.
Quote errorsPythonDouble quotes are its most frequent symbol, from dictionary keys and f-strings.
Semicolon errorsTypeScriptThe highest share of lines containing a semicolon.
New-line errorsPythonLine ends carry meaning: 42% of lines end with a block-opening colon.
Errors on capitalsSQLThe highest share of capital letters, switching case several times a line.

If none of these stands out and your letters are the slow part, the problem is general typing rather than code: the Learn Typing course or the symbol typing test will help more than another coding run.

The six language tests

Frequently asked questions

What is a coding typing test?

A typing test where the passage is code instead of prose. It scores the same speed and accuracy as a normal test, but code is full of symbols, capitals and line breaks, so it also reports symbol accuracy and separate counts for bracket, quote, semicolon, whitespace and new-line mistakes.

How is this hub different from the language pages?

The hub has ten passages drawn from all six languages, so one session covers several syntaxes. Each language page has its own set of passages in that language only, with technique and measurements specific to it. Start here to find your weakest language, then switch.

Do I have to type the indentation?

No. Press Enter at the end of a line and the next line’s indentation is filled in for you. Tab inserts two spaces if you need it, and a space typed where the passage expects a line break is ignored. Every other character, including the line break itself, must match.

Why is my WPM lower on code than on prose?

Code has far more symbols: between 6.5 and 17.2 per 100 characters in these passages, against 2.6 in ordinary English. Most symbols are reaches off the home row, many need Shift, and each mistake costs a whole word in net WPM. A noticeably lower code WPM is normal.

What is a good coding typing speed?

There is no published standard for programmers, and developers are rarely typing-tested. A more useful target than a WPM figure is symbol accuracy: if it stays at 95% or more across a five-minute run, the mechanics of typing code are no longer what slows you down.

Can I choose one language on the hub?

Yes. The language menu above the passage filters the hub’s passages to a single language. For more passages in that language, and a page of technique and measurements to go with them, open the dedicated language test.

Are my coding results saved?

No. The coding tests keep nothing between visits. The report card at the end of each run shows your figures, the expected code and your typing side by side; note or screenshot anything you want to track.

Which keyboard layout do the tests assume?

Any layout works for typing, but where the pages count Shift presses or describe where a symbol sits, they assume a standard US QWERTY keyboard. On other layouts some symbols move and need different fingers.

Is the coding typing test free?

Yes. Every coding test runs in your browser with no account and no download.