Methodology

#Scope and position

• Finaur provides education and analytics, not investment advice and not portfolio management
• Frameworks are models and scenarios for learning and research
• Users make their own decisions, no order routing, no custody, no execution

#Data and sources

• Primary market data vendors include equities, crypto, and indexes, listed in each strategy page
• Adjusted prices for corporate actions where applicable, splits and dividends included in equities backtests
• Timestamps are recorded in UTC and converted in the interface for readability
• Missing data points are handled with clear rules, either exclusion or forward fill, documented per strategy

#Universes and calendars

• Equities universes target survivorship free sets where available, selection rules appear in strategy details
• Crypto uses continuous calendars, exchanges and symbols are listed per strategy
• Trading calendars respect official market holidays for equities, continuous time for crypto

#Costs and slippage

• Backtests apply transaction costs and slippage, with default values stated on each strategy page
• Sensitivity analysis appears where relevant to show the effect of higher or lower costs
• All cost figures are educational estimates, real trading conditions vary across brokers and venues

#Backtesting method

• Walk forward evaluation with train and test splits, no look ahead usage
• Signals use information available at decision time only, entry and exit timing rules are documented
• Order of operations is consistent, for example universe filter then ranking then position sizing
• Randomness is controlled via seed where random selection is used, seed values appear in result metadata
• All parameter sets are recorded, the exact set used for the public chart is visible on the page

#Metrics and definitions

#Strategy page standard

• Hero chart first, equity curve versus benchmark with time range controls
• Stats panel with CAGR, max drawdown, Sharpe, win rate, trades, time in market
• Method and rules section, clear plain language and parameter table
• Risk notes and known failure modes, typical adverse conditions listed in plain terms
• Assumptions block with costs, slippage, universe, calendar, and data vendors
• Download links where available, report files carry integrity hashes

#Simulator behavior

• Paper simulation mirrors the backtest engine rules, fills use the same price formation logic stated on the page
• No live order routing, no execution, the simulator is for education and practice
• Parameter changes are recorded, each run shows the parameter set and timestamp

#Discipline Score

• Weekly score from zero to one hundred that reflects plan adherence, stop integrity, sizing variance, overtrading index, and hygiene completion
• Weights and guards appear in the Discipline Score page and release notes, versioned in the user record
• Top drivers and one practice suggestion are stored with each weekly score for learning and coaching

#Bias detection

• Heuristics track revenge, fear of missing out, hesitation, premature exit, size creep, and stop dragging
• Signals derive from the gap between plan and execution, timing clusters, frequent stop edits, and exits without triggers
• Bias events are recorded with timestamps and meta fields, visible in the journal and reports

#Publication and integrity

• Each report lists author, reviewer, publish time, methodology version, and an integrity hash
• Archive keeps all reports, no deletions for result reasons, corrections are logged with new hashes
• Scoreboard aggregates outcomes by month and compares with public benchmarks where applicable

#Assumptions and limits

• Backtests are historical simulations, real trading conditions differ because of liquidity, slippage, fees, taxes, and human factors
• Data quality and survivorship handling vary by market and vendor, we document known gaps on each strategy page
• Results are educational, not predictive, past performance is not a reliable indicator of future outcomes

#Versioning and change log

• Strategies carry a name and a semantic version and a parameter snapshot
• Discipline Score has a version label stored with each user week
• This Methodology page uses the version at the top of the document and lists material changes in the change log below

#Reproducibility

• Backtest runs store parameters, universe definition, calendar, cost model, data snapshot identifiers, and random seed when used
• Public artifacts include report integrity hashes, the same inputs recreate the same outputs under the same engine version
• Where we correct a dataset, we publish a new run and note the change in the archive

#Contact