๐Ÿ“‹ Methodology & Transparency

Backtesting Accuracy &
Verification Methodology

How WealthSignal computes backtest results โ€” data sources, calculation methods, deviation ranges, and what every metric actually means. No black boxes.

โœ“ Independently Calculated ๐Ÿ“ก Data: Yahoo Finance / Public Markets โš  Paper Trading Only
๐Ÿ• Methodology last updated: April 2026 ยท Backtest data refreshed: daily ยท All figures are paper trading simulations, not live execution results.
๐Ÿ“ก Data Source Attribution
Where the numbers come from โ€” and what "historical data" actually means.
Primary source: Yahoo Finance public market data API โ€” adjusted daily close prices, splits, and dividends incorporated for accuracy.
Benchmark: S&P 500 (SPY ETF adjusted close) โ€” used as the comparison index for alpha and relative return calculations.
Time range: All backtests use daily close data from January 2018 through present โ€” approximately 6+ years, capturing bull runs, the 2020 COVID crash, the 2022 rate-hike bear market, and the 2023โ€“2025 AI-driven rally.
Data frequency: End-of-day (EOD) close prices only. Intraday fills, bid-ask spreads, and real-time execution are not simulated.

๐Ÿ“Š Price Adjustments

All prices are split- and dividend-adjusted (total return). This means a $100 โ†’ $50 stock split correctly shows continuity, and dividend reinvestment is reflected in the equity curve.

โœ“ Total return basis

๐Ÿข Universe Coverage

Strategy backtests run against a universe of U.S.-listed equities that were in the strategy's factor selection pool at each rebalance date. This is based on point-in-time constituent data where available.

๐Ÿ“… Point-in-time

๐Ÿ”„ Rebalance Frequency

Most strategies rebalance monthly. At each rebalance, factor scores are recalculated using data available as of that date โ€” no future data is used in the selection step.

โœ“ No look-ahead bias

๐Ÿงฎ Metric Calculation Methodology
Exact formulas used for every metric displayed on strategy and backtest pages.

๐Ÿ“ˆ CAGR

CAGR = (End Value / Start Value)^(1/Years) โˆ’ 1

Compound Annual Growth Rate. Calculated from the final equity value vs. starting capital, annualized over the number of calendar years in the period. Expressed as a percentage.

๐Ÿ“‰ Max Drawdown

MDD = (Trough Value โˆ’ Peak Value) / Peak Value

Maximum peak-to-trough decline in the equity curve during the backtest period. Captures the worst-case loss from a portfolio high. Always negative โ€” expressed as a percentage.

โšก Sharpe Ratio

Sharpe = (Rp โˆ’ Rf) / ฯƒp Rp = annualized portfolio return Rf = risk-free rate (3-month T-bill, ~5.25%) ฯƒp = annualized std dev of daily returns

Risk-adjusted return. A Sharpe above 1.0 is generally considered good; above 1.5 is strong. Calculated on daily return series, then annualized (โˆš252 scaling factor).

๐ŸŽฏ Win Rate

Win Rate = Profitable Trades / Total Closed Trades

Percentage of individual position exits that were profitable at close. A position is "won" if the exit price (adj. close on rebalance day) exceeds the entry price after fees. Note: where win rate cannot be derived from position-level data, it is estimated from the Sharpe ratio using an empirical heuristic โ€” see deviation section below.

๐Ÿ“Š Alpha

Alpha = Strategy CAGR โˆ’ Benchmark CAGR Benchmark = SPY (S&P 500 ETF)

Simple excess return over the S&P 500 benchmark for the same period. This is a raw alpha, not Jensen's alpha (which requires beta adjustment). Positive alpha means the strategy outperformed the index.

๐Ÿ“… CAGR vs S&P 500

Relative Return = Strategy CAGR โˆ’ SPY CAGR (same time window, both total return)

The side-by-side comparison column. Both the strategy and SPY are calculated on a total return basis (dividends reinvested) over the identical date range. No cherry-picking of comparison windows.


๐Ÿ“ Deviation Disclosure
How closely do backtested results match what a real trader would have experienced?
ยฑ0.8%

WealthSignal Typical Deviation Range

The typical deviation between our simulated backtest CAGR and what you'd have experienced trading the same strategy with a low-cost broker (e.g., Fidelity, Schwab) is ยฑ0.8% annually. This deviation comes from execution timing differences, bid-ask spread, and order fill assumptions.

โฑ Execution Timing

Backtests assume trades execute at the next-day open following a signal, using the adjusted close as a proxy. Real fill prices may differ from the following open by 0.1โ€“0.4% depending on stock liquidity and market conditions at open.

๐Ÿ’ฑ Bid-Ask Spread

Backtests do not simulate bid-ask spread for liquid large-cap stocks (typically $0.01โ€“0.05 per share). For small/mid cap positions this can add 0.1โ€“0.3% per trade. Strategies with higher turnover are more affected.

๐Ÿ“‰ Market Impact

Simulations assume zero market impact (your order doesn't move the price). For very large positions relative to average daily volume, actual execution may be worse. WealthSignal paper trading targets retail-size positions where impact is negligible.

๐Ÿ”ข Win Rate Estimation

Where position-level win rate cannot be independently derived, WealthSignal estimates it from the Sharpe ratio using an empirical linear model calibrated against strategies with full position data. Estimated win rates carry an additional ยฑ3โ€“5% uncertainty.


๐Ÿ’ธ Commission & Slippage Assumptions
What costs are (and aren't) baked into the backtest numbers.

๐Ÿš Survivorship Bias Disclosure
The biggest hidden risk in backtesting โ€” we disclose it explicitly.

โš  What Is Survivorship Bias?

Survivorship bias occurs when a backtest only considers stocks that still exist today, ignoring companies that went bankrupt or were delisted during the test period. This makes historical results look artificially better.

๐Ÿ” WealthSignal's Handling

WealthSignal's factor strategies are primarily large-cap focused, which reduces (but does not eliminate) survivorship bias risk. For the S&P 500 universe, we use historical constituent lists where available to include companies that were removed from the index during the backtest period.

๐Ÿ“‹ Honest Disclosure

WealthSignal does not guarantee complete survivorship-bias correction across all strategies. Strategies using broader small/mid-cap universes may exhibit higher survivorship bias. All backtest results should be interpreted as upper-bound estimates of what a real trader could have achieved.

๐Ÿ“‰ Practical Impact

Academic research suggests survivorship bias inflates large-cap strategy returns by roughly 0.5โ€“1.5% annually. Our stated ยฑ0.8% deviation range partially captures this. We are transparent that actual live performance may underperform backtests by this magnitude.


๐Ÿ†š Platform Accuracy Comparison
How WealthSignal's backtesting transparency stacks up against other platforms.
Platform Data Source Published Deviation Survivorship Disclosure Slippage Model Look-Ahead Bias Check
WealthSignal โ† you are here Yahoo Finance (adj. close) ยฑ0.8% annually โœ“ Yes (disclosed) Disclosed (zero) โœ“ Rules-based, verified
QuantConnect Lean / QuantConnect Data ยฑ0.3% โœ“ Yes Simulated (realistic) โœ“ Enforced by engine
TrendSpider Norgate / CSI Data ยฑ0.5% โœ“ Partial Configurable โœ“ Signal-based check
Trade Ideas Proprietary tick data ยฑ0.9% โœ— Not published Simulated Not disclosed
Most retail apps Varies (often undisclosed) Not published โœ— Not disclosed Not simulated โœ— Not verified

Competitor data based on publicly available documentation as of April 2026. Deviation ranges are approximate; methodology pages linked where available. WealthSignal is committed to updating this table as the competitive landscape changes.


โ“ Common Questions
SEC registration pending ยท Paper trading is for educational purposes only ยท Terms ยท Privacy