How the simulation works

Quotes and candles come from public market-data feeds for US stocks, major crypto pairs, and forex. Data can be delayed by up to a few minutes and occasionally wrong or missing — treat prices as realistic, not exchange-grade. When a live quote is unavailable, the app says so (or shows your cost basis, clearly labeled) rather than inventing a number.

Spot market orders fill from a server-fetched quote with your configured adverse spread and slippage assumptions. Quotes older than five minutes are rejected, and stocks fill only in the provider’s regular session. Limit orders require a fresh observation after placement and a fill price within the limit after modeled price costs. There is no queue or partial fill. Leveraged stop-loss and take-profit levels are checked on observed quotes, not every exchange tick. Background checks can be delayed; they are not a guaranteed execution schedule.

In Research & review, configure a broker profile for each spot account’s asset class. Explicit fees can be zero, as with many stock/ETF brokers. Fixed, per-unit and notional fees are separate from modeled spread and slippage. Fees affect cash, cost basis and net results; each transaction preserves its applied assumptions. Existing fills are not recosted. Leveraged positions retain their separate existing model. These assumptions do not establish real execution fidelity: liquidity, settlement, dividend credits and corporate actions are absent.

Backtests replay a strategy's exact live logic bar by bar over recent history. Three rules keep them honest:

• No look-ahead: at every decision point the replay sees only bars that had fully closed at that moment — never data from the future.
• Costs are charged: every backtest trade pays an estimated round-trip spread + slippage (about 1 pip on forex majors, ~0.1% on crypto, ~0.05% on liquid stocks). These estimates are independent of your spot account’s broker profile and do not reproduce changing real-market costs.
• Stops first: when a bar touches both the stop and the target, the trade is scored as a loss.

Backtests still overstate real-world results — history is one path, and live spreads widen at the worst moments. Use them to compare strategies, not to predict returns.

Your performance chart is a time-weighted return: adding virtual cash never counts as profit — only what your positions actually did. The leaderboard ranks traders by percentage return on the money put in since each account's last reset, valued at live prices. There is no way to buy an asset at a made-up price or backdate a trade.

The AI scanner runs on your own Anthropic API key — Poshkan never bills you for it and can't spend on your behalf. Every AI-proposed trade must pass the same validation as any other signal (real symbol, sane stop/target geometry, price near the live quote, minimum reward:risk) before it can execute, and it always trades inside the risk limits you configure.

Your account needs an email address and a display name — nothing else. No payment details, no broker connection, no real money anywhere in the system. Trades, balances, and returns are 100% virtual; nothing can be won or lost, and nothing on Poshkan is financial advice.