Skip to content
All library documents

How Quantitative Investing Signals Map Data to Decisions

Article Quant Q&A · Author: Qwerty
Read the original

Summary

The document explains a signal as a variable derived from market information that represents or triggers an investment decision. It may be binary, indicating whether to act, or take a range of values that can encode degrees or directions of action, such as sell, hold, or buy. The precise form depends on what the execution process expects.

It connects this usage to signal processing, where noisy inputs are transformed into information that can guide an action. In quantitative investing, the inputs are market data and the output is a decision-relevant value. The text also notes digital signal processing as a field with applications in quantitative finance, but offers no specific strategy, construction method, performance evidence, or guidance on validating signals. Its contribution is a broad definition and a simple conceptual link between data processing and trading decisions.

Key ideas

  • A quantitative investing signal maps information to an investment decision.
  • Signals can be binary, continuous, or encode multiple actions such as sell, hold, and buy.
  • The required signal format depends on the execution process.
  • Signal processing offers a general analogy for turning noisy data into decision-relevant information.
  • The document mentions digital signal processing in finance but gives no specific applications or evidence.

Tags

StatisticsExecutionMachine learning
Full text
# What is "signal" in quant investing?


# What is "signal" in quant investing?












Can somebody explain (and give examples) of "signals" in quant investing? What are those? What does this word mean?

## Answer by Mike (score 2)

https://quant.stackexchange.com/a/58378

As people in the comments noted, signal broadly refers to a trigger variable that denotes an investment decision. This is normally a boolean variable (i.e. 0 or 1) but could be continuous (0 to 1) or any other range (e.g. -1/0/1 sell/hold/buy), depending on what your execution algo might dictate.

Just wanted to add that this terminology comes from the general field of signal processing, where, broadly speaking, you take noisy information and translate it into meaningful information. In this case, you are taking market data and transforming it into a buy/sell decision.

More specifically on signal processing though, Digital Signal Processing is a somewhat popular subfield for quantitative finance, and there are some interesting work done on this.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.