Econometric strategy research workspace

Turn a market hunch into evidence you can trust.

Write the hypothesis, backtest it with real costs, then put it through the statistical tests an economist would ask for. All in the browser, all in your own private workspace.

Runs in the browser
Private workspace per account
Paper trading only

Most backtests answer the wrong question.

The usual shortcut

IdeaTune until it looks goodDeploy

Parameters chosen on the full sample, no costs, no out-of-sample test. The equity curve looks great and means little.

The OpenQuant way

HypothesisFrozen rulesHoldout + tests

Decide what you are testing first, evaluate once on unseen data, and check the return series before believing it.

The key shift: treat a strategy as an econometric claim about a time series, and make it survive the tests.

One workflow, three disciplined steps.

Hypothesize

Write the idea down before touching data
Freeze rule parameters and the evaluation window
Save it to your research notebook

hypothesis: momentum persists 3–12m
rule: tsmom(lookback=126)
holdout: last 30% of sample
status: frozen

Test

Backtest with commission and slippage
Rank parameters on training data only
Score once on a chronological holdout

train Sharpe   0.91
holdout Sharpe 0.64
max drawdown  −18.2%
trades        142

Validate

Stationarity, autocorrelation and normality tests
GARCH volatility and ARIMA forecasts
Bootstrap intervals and Newey–West errors

ADF p = 0.003  ✓ stationary
Ljung–Box p = 0.41
JB p < 0.001  fat tails
β (HAC) = 0.38 ± 0.07

1Hypothesize
2Test
3Validate
4Refine

A complete research desk.

MKT

Markets & data

Daily candles, watchlists, Yahoo adjusted prices, CSV import and FRED macro series.

LAB

Strategy lab

SMA crossover, RSI mean reversion and time-series momentum with realistic costs.

ECO

Econometrics

ADF, KPSS, Ljung–Box, Jarque–Bera, GARCH(1,1), ARIMA and benchmark OLS.

OOS

Honest evaluation

Train/holdout search, walk-forward model selection and Monte Carlo intervals.

LIB

Model library

Transparent templates, side-by-side comparison and a signals hub for saved models.

PPR

Paper portfolio

$100,000 of virtual capital with cash and position checks. No broker, no real money.

Who it's for

Economists and students

Bring time-series discipline to market questions, with every test result on screen.

Independent researchers

Go from idea to defensible evidence without building a backtester first.

Small teams

Share a common, reproducible way to evaluate strategies before any capital is at risk.

Every version of an idea, on the record.

v1
Baseline crossover
SMA 20/50 on daily closes, no costs.
Sharpe 1.12
v2
Add realistic costs
1 bp commission and 5 bp slippage per fill.
Sharpe 0.71
v3
Holdout check
Parameters frozen, scored on unseen data.
Sharpe 0.48
v4
Volatility targeting
GARCH-scaled position size, re-tested.
pending

Illustrative example. Note how the Sharpe ratio shrinks as the test gets more honest.

Questions

Is this a trading platform?

No. It is a research workspace. Orders go to a paper portfolio with virtual capital; there is no broker connection.

Do I need to install anything?

No. The workspace runs in your browser. An optional Python engine is available for heavier computation.

How do I sign in?

Click Sign in and use Google or a one-time code sent to your email. Each account gets its own private workspace.

Where does the data come from?

Yahoo adjusted daily prices, FRED macro series, your own CSV files, or clearly labeled synthetic samples.

Can I trust a good backtest?

Only after it survives a chronological holdout and the statistical checks. The workflow is built around that.

Is it investment advice?

No. Results are historical simulations for research and education.