YDLib / POLICIES · UPDATED 21 OCTOBER 2026
Risk & research methodology
Research is a way to investigate uncertainty. It does not remove trading risk.
Financial risk
Trading can lose some or all committed capital. Leveraged instruments can produce rapid losses and, depending on the product and protections, losses beyond the initial deposit. Software failure, connectivity, broker differences and execution conditions can change outcomes. Evaluate suitability independently; material here is general research and education.
Simulated versus live results
Backtests, optimizations, forward simulations and verified live results are different evidence classes. Simulated outcomes benefit from assumptions and may be affected by hindsight, selection and overfitting. An out-of-sample window must follow optimization chronologically and remain untouched during fitting. Repeated inspection or tuning weakens its independence. Starter and demo content is illustrative, not a performance record.
Read the whole report
Inspect the instrument, session, timeframe, data quality, sample size, test dates, optimization and OOS windows, commissions, spread, slippage, sizing and ambiguous-bar assumptions. Win rate alone is insufficient. Profit factor, expectancy, drawdown, tail losses and time-based risk matter. Sharpe requires a stated return interval and annualization convention; a per-trade R proxy is not an annualized account-return Sharpe.
Prop firm simulations
Rules, fees, reset policies, daily loss definitions, consistency constraints and payout conditions differ across firms and change. Account EV depends on the modeled probabilities and rules, not a guarantee of passing or receiving payouts. Confirm the current firm contract independently.
Data and conflicts
Market and macro feeds may be delayed, revised, stale or unavailable. Original observation and refresh dates should accompany the data. Catalogue statistics are supplied by the operator and are not independently audited unless explicitly stated. YDLib has a commercial interest in future memberships and software sales. Publishing a report is not independent verification of its claims.
Responsible interpretation
Metric explanations describe calculations and limitations, not a recommendation. Validate code in a safe test environment, use appropriately limited risk and maintain your own monitoring. Past or hypothetical performance does not predict future results.