GobiPractice, seen country by country.
Gobi is a trading practice companion. These pages explore simulated trading, with virtual money rather than real investment returns.
Live views show measured aggregates from simulated trading. Figures update as new records arrive and may be revised as coverage improves.
The People map shows all-time community reach: each existing known user is counted once, including countries with just one person. Someone who has chosen a community country in the app, and kept it for a week, is counted there. Everyone else is counted in their first observed country. It is independent of performance coverage and is not daily or monthly active users. Countries and territories follow the location provider’s geographic codes. Unknown countries and known staff/test accounts are omitted. Performance views use a separate measured sample. That sample includes participants who filled a trade or held a position during the displayed period. Each person is counted once, across devices and fresh runs. Ordinary paper accounts are included; known staff and test accounts, Atlas and real-money accounts are excluded. Performance is shown from one measured participant, with the actual sample size alongside the figure. Missing source records remain excluded; a small sample is not a claim of statistical significance. Country is the community country a person chose in the app once that choice has stood for a week, and otherwise the first trusted network-location observation. Neither is nationality or verified residence; unknown countries are omitted. A monthly view uses the country in effect when the month began, so a change counts from the following month. All time uses the current country, so a person’s whole record moves with them. Published monthly editions are not revised when someone changes country. Backfilled activity is grouped by the first observed country, which may have been recorded later than the trades. It does not establish where those historical trades took place.
Signups are different from trading participants. Country comparisons describe the participating Gobi sample; they do not measure a whole population’s ability, and they do not predict real-money performance.
One combined portfolio for each country.
We add together trading profit and loss, then divide by the combined allocated capital. This measures percentage return, not total dollar profit, and it gives each unit of allocated capital equal weight.
With equal starting balances, the result equals the mean participant return. With unequal balances, it does not. The global return uses the same calculation across the countries shown, rather than averaging country percentages.
For example, two accounts starting with 10,000 each and making +1,000 and −400 have a combined return of +3%. Earlier gains and losses remain when someone starts a fresh run. The old run ends at its recorded reset valuation. Opening equity plus fresh virtual funding forms the denominator; resets and bonuses add capital, never trading profit. Trading P&L includes fees, funding payments and unrealized gains and losses.
How people practice matters, too.
Average entry leverage
The arithmetic mean of leverage selected at trade entry, weighted by the number of trade entries. Executed opening orders, including additions, count once across partial fills. Closing quantities are excluded. At least 95% of entries must have verified measurements before the metric is published.
Typical position size
The median opening trade value in US dollars: executed opening quantity multiplied by its execution price. Partial fills of the same opening order are added together; additions count as new orders. On a reversal, only the quantity opening the new side counts, and closing quantities and fees are excluded. Orders belong to the period of their first opening fill. At least 95% of entry candidates must have verified measurements. The global median belongs to the pooled set of entries; it cannot be reconstructed by averaging country medians.
Higher leverage and larger positions are not ranked as better decisions. The switches sort these values from highest to lowest to make comparisons easy.
Monthly and all-time views.
Connected views default to All time, beginning at the earliest covered date shown, with a This month option for the current UTC calendar month. Historical imports extend this coverage when source records are available. All-time return uses its own capital baseline, not an average of monthly returns. The first partial month is labeled. Aggregates update about once a minute; delayed inputs retain the last verified timestamp. Boundary valuations use the first verified sample within 90 seconds of the UTC month boundary. Historical boundaries use source ledger records through the cutoff and historical price marks less than one minute before it. Accounts with missing historical evidence are omitted from the affected dataset and the view indicates incomplete coverage; recovered accounts are a measured subset, not a complete census.
Global figures cover the countries published for that metric.
The big picture, with exact numbers beside it.
Mint and rose show positive and negative return. Purple shows leverage and position size. The People map uses purple for every known country, starting with one user. Grey means no known country observation, not proof of zero activity. In performance views a dash means the metric is unavailable; the country remains listed. Small countries are also accessible through search and the ranking table.
Geography uses Natural Earth country boundaries distributed by World Atlas, in a Natural Earth projection. Boundaries are for visual orientation, not a statement about political status.