The Lab

Python, SQL, Julia, C++. Graded on the result.

More than a hundred exercises on four shared datasets. Write the code, run it, and the answer is compared with one computed from the same rows, to four decimals.

Open the app

Parametric VaR · MSFT

from scipy.stats import norm

r = prices[prices["symbol"] == "MSFT"] \
      .sort_values("day")["close"].pct_change().dropna()
result = -(r.mean() + norm.ppf(0.05) * r.std(ddof=1)) * 100

Correct · 95% VaR: 0.8594%

Four languages

Each one where it belongs

Python

In your browser

pandas, NumPy, SciPy, statsmodels and scikit-learn, loaded on first import. No install.

SQL

In your browser

A real PostgreSQL in the page: joins, windows, dates, retention, and data-writing tasks checked in a rolled-back transaction.

Julia

On the QuantPad runner

Numerical functions on market data: returns, volatility, EWMA, drawdown, correlation.

C++

On the QuantPad runner

The same functions in the language pricing libraries are written in, compiled and run for you.

Cohort retention · SQL

select date_trunc('month', u.signup_date) as cohort,
       count(distinct u.id) as users,
       round(100.0 * count(distinct e.user_id) / count(distinct u.id), 1) as active_pct
from users u
left join events e on e.user_id = u.id and e.event_date >= u.signup_date + 30
group by 1 order by 1;

EWMA volatility · Julia

function ewma_vol(close::Vector{Float64})
    r = close[2:end] ./ close[1:end-1] .- 1
    v = r[1]^2
    for x in r[2:end]; v = 0.94v + 0.06x^2; end
    return sqrt(v * 252) * 100
end

Correct · 12.451%

How it is graded

One correct answer, checked like a reviewer would

Read
A statement that fixes the method: which rows, which ddof, which seed.
Write
Your code, in an editor with completion. Tables are already loaded.
Run
Python and SQL run in the page; Julia and C++ are compiled on the runner.
Compare
Your result against the one computed from the same data, value by value.

Languages

Where each language runs

Language	Where	Why a team uses it
Python	Browser	The common language of research and data.
SQL	Browser	Every tick, trade and P&L row lives in a database first.
JavaScript, TypeScript	Browser	Nothing to install; the norm for trading front-ends.
C++	Runner	Where latency matters: pricing, gateways, HFT.
Julia	Runner	Fast numerical code that reads like the maths.
C#, OCaml	Runner	Order management and typed pricing models.
R, kdb+/q, Rust, Java	Read	Econometrics, tick databases, crypto infrastructure.

Free · English and French

Start with a spread. Finish with an offer.

No card, no trial. Sign in and solve your first exercise in under a minute.