Signals
A signal is the moment analysis turns into a decision to act.
This category defines that precisely, across the explainers on what a trade signal is, the difference between buy, sell, and hold, entry versus exit, and how conviction and strength are judged.
It covers the signal lifecycle, single versus multi-factor signals, the lines between quantitative, systematic, discretionary, and algorithmic trading, and the basics of backtesting.
Investing With Purpose frames a signal as a process to be tested and governed, not a tip to be chased.
The payoff is understanding how systematic strategies generate, validate, and ultimately trust the signals they trade on, and why most informal ones never survive contact with data.
A trade signal is a specific trigger that tells you to buy or sell a security. It turns a vague view like "the market…
Buy, sell, and hold are the three basic actions any signal can recommend. Understanding what each one genuinely means,…
An entry signal tells you when to open a position. An exit signal tells you when to close it. They are two halves of…
Signal conviction is how confident a system (or a trader) is in the direction of a given trade signal. It turns a…
A signal lifecycle is the set of states a trade signal passes through from the moment it is generated to the moment the…
You can either subscribe to a service that tells you what to trade, or build the system yourself. Both routes can work.…
A trading setup is a specific pattern of market conditions that triggers an entry. It sits one level below the strategy…
Paper trading is simulated trading with fake capital against real market data. It is the standard way to learn platform…
A **single-factor signal** makes a trading decision from one input. A **multi-factor signal** blends several…
Quantitative trading makes decisions from models and data. Discretionary trading makes decisions from human judgment.…
Systematic trading is a rules-based method where every entry, exit, and sizing decision is defined in advance, tested…
Backtesting is the practice of running a trading strategy against historical market data to see how it would have…
Walk-forward analysis is a backtesting method that fits a strategy's parameters on one slice of history, tests the…
In-sample data is the portion of history you use to fit and tune a trading model. Out-of-sample data is a separate…
Look-ahead bias is the error of using information in a backtest that would not have been available in real time on the…
Signal decay is the tendency for a predictive trading signal to weaken over time as more investors discover and trade…
In investing, the signal-to-noise ratio describes how much of a forecast's variation represents real predictive content…
The risk-reward ratio is the size of the loss you accept on a trade compared with the size of the gain you are…
Position sizing is the step between "the system said BUY" and "how many shares do I buy?" It decides more about your…
An alpha factor is a characteristic of a stock, or any asset, that reliably predicts returns beyond what simple market…
Hit rate, also called win rate, is the percentage of trades that close with a profit. It is one of the most cited…
All high-frequency trading is algorithmic, but almost no algorithmic trading is high-frequency. The two are often…
Overfitting is what happens when a trading rule is tuned so closely to past data that it captures the noise in that…
Survivorship bias is the error of testing a strategy only on assets that still exist today. Failed companies, delisted…
A regime-switching model treats market dynamics as a system that alternates between a small number of hidden states,…
The Kalman filter is a recursive estimator that tracks a hidden state variable whose value you cannot observe directly,…
A hidden Markov model (HMM) treats the market as a system that switches between a small set of unobserved states, each…
Principal component analysis (PCA) rotates a set of correlated asset returns into a new set of uncorrelated factors…
Cointegration pairs trading goes long one asset and short another in a ratio such that the combined spread is…
The augmented Dickey-Fuller (ADF) test asks whether a time series has a unit root, meaning it behaves like a random…
The Johansen test identifies how many independent long-run equilibrium relationships exist among a set of two or more…
Granger causality tests whether past values of one time series help predict another, beyond what the second series…
Dickey-Fuller detrending is the practice of removing a deterministic trend or a stochastic trend from a series so that…
Structural break detection asks whether the parameters of a statistical model, such as a mean, variance, or regression…