1/31
Vocabulary flashcards covering quantitative signals, trade execution mechanics, buy types, normalization concepts, and prototype models from the Quant Signals Primer.
Signal
A single numerical output produced every bar by a model evaluating raw market price data, where approximately 0 denotes normal conditions and large absolute magnitudes denote unusual conditions.
Buy Type
The decision layer in a quantitative trading system that evaluates a model's signal and applies rules to set a market position, never inspecting raw price directly.
Position
The exposure state determined by a buy type: long (profiting from price increases), short (profiting from price declines), or flat (holding zero exposure).
Volatility
A stock's characteristic dispersion or typical daily movement around its baseline, colloquially described as its typical wobble.
Normalization
The mathematical conversion of raw dollar differences into units of asset volatility (σ), enabling a single standardized rulebook to trade instruments at any price level.
Rolling Mean (μ)
The arithmetic average of closed prices calculated across a defined rolling window of recent bars.
Rolling Spread (σ)
The sample standard deviation measuring the typical dispersion of prices from their rolling mean across a fixed window of bars.
Z-score (z)
A normalized signal calculated as z=σPrice−μ, defining distance from the mean in units of volatility.
Momentum
The market premise that an asset in motion tends to maintain its trajectory, prompting traders to bet on the continuation of established trends.
Mean Reversion
The market premise that price behaves as if tethered to a central baseline by an elastic band, snapping back toward average levels after extreme deviations.
trend Buy Type
A momentum decision rule that opens a long position when a signal crosses above a positive threshold and a short position when it drops below a negative threshold.
mr Buy Type
A mean-reversion decision rule that opens a short position when a signal climbs above a positive threshold and a long position when it sinks below a negative threshold.
The Peeking Trap
A look-ahead backtesting flaw in which an order on bar t executes based on bar t's closing price before that bar has closed, producing unreplicable backtest performance.
Zero Calibration Point
The shared reference baseline across all models where a reading of 0 signals standard unexceptional price behavior, while non-zero values convey direction and degree of abnormality.
Family A: Everyday Toolkit
The foundational set of models (IDs 0–10) engineered to detect whether an asset's price is unusual under typical, everyday market conditions.
Family B: Survive Wild Days
The advanced set of models (IDs 11–20) specialized in managing severe price outliers, sloping trend channels, and fat-tailed return distributions.
ols
An Ordinary Least Squares regression model that computes price deviation relative to a sloped trendline trajectory rather than a flat horizon.
quantile_reg
A quantile regression model that tracks concurrent 25th and 75th percentile channel bounds to assess price positioning inside a sloped corridor.
skew_zscore
A signal model that adjusts for directional skewness in crash-prone assets exhibiting asymmetric downside tail risk.
t_zscore
A distribution model that compresses z-scores using a Student's t distribution to suppress false alarms caused by frequent fat-tailed market shocks.
vol_regime
A volatility term structure model that measures the ratio between short-term volatility and a long-term baseline to detect expanding market regimes.
mad_zscore
A robust z-score model computed using the rolling median and Median Absolute Deviation, preventing single freak price ticks from distorting the calculated baseline.
winsor_zscore
A robust model that clips the most extreme tails (such as the outer 5%) before calculating the mean and spread.
huber_zscore
A tunable robust center model that smoothly downweights outlier observations without rejecting them entirely.
hampel_z
An outlier-rejection model that acts as a hard filter by assigning exactly zero mathematical weight to extreme price anomalies.
rls_resid
A recursive least squares model measuring residual distance against an adaptive trendline that recalibrates on every bar.
robust_scaler
An outlier-resistant normalization model that scales raw prices using the median and Interquartile Range (IQR).
min_max
A bounded model scaling price positioning within its recent high/low range to a fixed [−1,+1] interval.
sma_diff
A simple trend-following metric that measures the percentage distance between current price and its simple moving average.
rsi
A bounded momentum oscillator that evaluates the relative magnitude of recent upward gains against recent downward losses.
bbands_pct
An indicator quantifying the exact location of price within volatility-adjusted Bollinger Bands.
macd_norm
A moving average convergence/divergence indicator standardized by volatility to allow cross-instrument comparability.