Technical analysis (lens)

Definition

The technical lens evaluates price, volume, trend, and chart behavior: what the traded record of an instrument itself says about supply, demand, and the positioning of participants. Evidence under this lens is computed from market data (bars, volume, derived indicators), not from issuer financials or economic releases.

How it works / structure

• Inputs: OHLCV bars at any timeframe, plus derived series.
• Core operations: trend measurement (indicator-sma, indicator-ema, indicator-macd), momentum/oscillators (indicator-rsi, indicator-stochastics, indicator-roc), volatility bands and ranges (indicator-atr, indicator-bollinger-bands), volume analysis (indicator-obv, indicator-vwap, indicator-volume-profile), and level/pattern identification (support, resistance, breakouts).
• Output shape: parameterized, computable conditions — “20-day SMA above 50-day SMA”, “RSI(14) below 30” — which is what makes this lens directly executable by the simulation engine.
• Concept-catalog mapping: technical evidence binds to the technical family of the analysis-concept catalog; every indicator concept reads an existing computed field and is never recomputed at read time.

When it applies

Any instrument with a liquid traded record, on horizons from intraday to months. It is the primary lens when the thesis is about price behavior itself (trend continuation, range reversion, breakout) and a supporting lens for timing entries and exits under other lenses. It is weakest where the traded record is thin or gapped (illiquid names, new listings).

Risk profile & failure modes

• Data snooping: with thousands of rule/parameter combinations, some backtest well by chance; Sullivan, Timmermann and White (1999) showed apparent rule profitability can vanish under data-snooping adjustment.
• Regime dependence: rules tuned in trending regimes fail in choppy ones and vice versa (regime-volatility).
• Self-invalidating crowding: widely watched levels and signals attract orders that change the behavior the signal described.
• Costs: high-turnover technical rules can be profitable before costs and unprofitable after spread and slippage (ms-slippage-friction).

Evidence & limits

Brock, Lakonishok and LeBaron (1992) found that simple moving-average and range-break rules had statistically significant predictive content on Dow data from 1897-1986; later work qualified this. Park and Irwin (2007), surveying ~95 modern studies, report mixed results: many studies find profitability, but a substantial share suffer from data-snooping, ex-post rule selection, or ignored costs. Lo, Mamaysky and Wang (2000) found some chart patterns carry incremental information without establishing net profitability. The honest summary is: some technical regularities have peer-reviewed support in some markets and periods; the claim that technical analysis is reliably profitable after costs is unproven. Pattern folklore without published evidence is labeled folklore in this KB.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Instrument X will close above its 200-day SMA within 20 trading days” — falsified if no such close occurs in the window.
• “Instrument Y’s 14-day RSI will not exceed 70 during the next month” — falsified by any daily close with RSI(14) above 70.

Cross-references

• Pillar 6 indicators: indicator-sma through indicator-options-flow
• Strategies that operationalize this lens: strategy-breakout, strategy-mean-reversion, strategy-momentum, strategy-swing-trading
• Adjacent lenses: lens-quantitative (formal testing of technical rules), lens-market (breadth and regime context)