Choosing Between Correlation and Regression for Stock Scores

Summary

The document describes an introductory stock-market assignment: devise a score from information about a stock, then compare that score with returns over several future horizons. The student has calculated a correlation coefficient between score values and realized stock returns and asks whether predictive or multiple regression would be more suitable, whether the variables belong in a particular order, and whether swapping them matters.

The material identifies a useful distinction but does not include an answer. Correlation summarizes the strength and direction of a linear association and is symmetric in its two inputs. Regression instead assigns one variable as the outcome and the other or others as predictors, so the choice of dependent variable matters when building a predictive model. A score intended to forecast future returns would be evaluated with future return as the outcome. The document gives only two example observations and no actual methodological guidance, statistical evidence, or validation results, so it cannot establish that the score predicts returns.

Key ideas

• Correlation describes a symmetric association between two variables.
• Regression distinguishes predictors from an outcome, so variable roles matter.
• For forecasting, future stock return would be the outcome and the score a candidate predictor.
• The two displayed observations do not establish predictive performance.
• The document poses the modeling questions but supplies no answer or validation method.

Tags

# Should I use a correlation coefficient formula or a multiple regression formula?


# Should I use a correlation coefficient formula or a multiple regression formula?












I have an assignment dealing with the stock market and I'm a little lost. My instructions are to come up a method to create a score for a stock then compare the score against what the stock actually earned in 3 time periods (1 day, 3 days, and 7 days). I'm new to coefficients so I have some questions below.

Here is my data that I returned using a regular correlation coefficient formula. I swapped out the stock pct return vs market pct return and used the stock price returns for each day as the stock returns and used the score as the market return. I did this because I can create a score with information that I can get from the stock but I obviously can't get the actual return of the stock.

Score, Return, Correlation Coefficient

```
 26.87, 5.44, .022
 21.34, 3.42, .034
```

- Since I'm plugging in the data that I know and trying to get a return, does that mean I should use a predictive regression formula like a multiple regression formula?

- Am I using them in the right place in the formula or should I swap the values around?

- Will this matter if I swap out the values in the formula?

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.