VI · Feature engineering
Station VI of The Quant Map: one raw variable, many research questions. Transforms, filters, frequency response. Reading order plus the free map PDF.
transforms of observable state, one raw variable, many research questions
One raw series supports many questions, and they are different hypotheses. Carry level asks how big the paycheck is. Carry over volatility asks how big it is per unit of pain. Carry rank asks whether this pair is the best paid in the set. Carry change asks whether the gap is widening. Carry conditioned on trend asks whether price already agrees. Five features, two rates and a spot. RSI, MACD, and a stochastic on the same close are often one band agreeing with itself. Stacking them feels like confirmation. The frequency response says it is one feature in costume.
The filter shelf past this table is the same question asked of one operator: which band it keeps, and how late. Butterworth, sinc, Hilbert, roofing, and the named oscillators are later visits. Start with the primer and the frequency response. A filter output is a feature. The forecast is a later station.
A feature is a question asked of an observable, and one observable supports many questions. The filter shelf, moving averages, oscillators, band-pass constructions, is one operator family: which band it keeps, and how late it tells you.
The Quant Map PDF works this spine through example research: five full descents from observation to live rule, and a library of 835 strategy entries, each with a named payer and a named failure mode. Get the free PDF on the Start Here page.
Reading order
Top to bottom. Stop at the first article you cannot apply to a live trade; that is the one to read slowly.
| Article | Why it is on this row |
|---|---|
| The Indicator Is More Important Than the Model | No transform of a weak input carries more information about the target than the input did. The ceiling is the feature. |
| Why Most Indicators Should Be Transformed Before Modeling | Scale, tails, stationarity, and cross-market units. The transform is part of the feature's name, and its window is not a backtest knob. |
| Feature Engineering Before Machine Learning | On daily bars the sample is small and the signal is thin. The pipeline that stationarizes and checks the histogram does more than the model swap. |
| No Filter Is Predictive: What Traders Misunderstand About Smoothing | A causal filter is a function of the past. A crossover is the moment the less-lagged average caught up, after the turn. |
| DSP and Digital Filters for Traders: The Primer Nobody Wrote First | Trend, deviation, cycle, noise. Lag is the center of mass of the weights. Nyquist is two bars. |
| The Frequency Response of Trading Indicators | RSI, MACD, stochastic, and ADX often sit in the same cycle band. Confluence of those four is one vote. |
| CMMA: A Better Momentum Primitive Than Price-minus-MA Alone | Log price, a causal average, ATR, and a lookback correction, so 10-day and 200-day momentum share an axis. |
| Momentum Is a High-Pass Filter | Momentum is the deviations the low-pass threw away. Name the band or you are holding a smoother with a sign. |
On the map: ← V · Observable world · The grand map · VII · Hypothesis →