Correlation, Regression, Alpha, Beta and R-Squared
Learn how covariance, correlation, and regression measure the link between two variables, and what alpha, beta, residuals, and R-squared mean.
Category: Statistics
In plain English
Sometimes two things move together. When the whole stock market goes up, lots of individual stocks tend to go up too. These tools are just a tidy way to answer: "When one number changes, how does the other usually change?" We draw a straight line through a cloud of dots to capture that relationship, then describe the line with a few simple numbers.
What it measures
Covariance asks whether two things move in the same direction; correlation rescales that to a clean number from -1 to +1 (+1 = move together perfectly, 0 = no link, -1 = move opposite). A regression then fits the "best-fit line" that comes closest to all the dots. That line is described by a slope, an intercept, the leftover gaps, and a score for how well it fits. - Beta (slope): if the market rises 1% and the stock usually rises 1.5%, beta = 1.5 (it moves 1.5x the market). - Alpha (intercept): the part of the return that is NOT explained by the market move (here, say +0.2% on average). - Residuals (leftover): the gap between a real dot and the line on any single day. - R-squared: how much of the wiggling is explained by the line, from 0% to 100%. R-squared of 0.40 means 40% explained, 60% is other stuff. - Tiny example: 10 days of stock returns vs market returns give correlation 0.7, beta 1.5, alpha +0.2%, R-squared 0.49.
How SurgeFlow uses it
This is the math under several SurgeFlow tabs. Beta versus a market index is computed in the price pipeline and feeds risk views. In the Factors and Backtesting tabs, the same regression idea measures "factor loadings" - how strongly a stock leans on common drivers like value or momentum (the slope on each factor). It also underpins the classic CAPM (Capital Asset Pricing Model) idea that a stock's return relates to the market return through beta. - Beta = the slope of stock return regressed on market return. - Alpha = the intercept, the leftover return not tied to the market. - Factor loadings (Factors tab) = the same slopes, but against factors instead of just the market.
How to read it on the site
On the Price tab, a stock's beta tells you its typical sensitivity to its home market: beta near 1 moves with the market, above 1 is more jumpy, below 1 is calmer. On the Factors tab, the factor loadings show which common styles a name leans on. Treat correlation and R-squared as "how tight is the relationship" gauges - a high R-squared means the line explains a lot; a low one means most movement comes from elsewhere. - Beta around 1.0 - moves roughly with the market. - Beta well above 1.0 - amplifies market moves (more volatile). - Beta below 1.0 or negative - calmer, or moves against the market. - High R-squared - the market (or factor) explains most of the wiggle; low - it does not.
What it cannot tell you
These are descriptive tools: they summarise how two numbers moved in the PAST, not what will happen next. A strong correlation does not prove one thing causes the other - two stocks can rise together simply because the whole market rose, or by pure coincidence. Beta and alpha are estimates from a sample, so they shift as you add days and can be unstable; alpha in particular is noisy and easily mistaken for skill when it is just chance. None of these numbers is a buy or sell signal, and SurgeFlow does not give investment advice. - Past relationship is not a forecast - betas drift over time. - Correlation is not causation - a shared cause (the market) can fake a link. - High R-squared means "well explained," not "good investment." - Alpha can look like skill but often is just leftover noise.