SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
rmeenaks
Topic Author
Posts: 186
Joined: May 1st, 2006, 2:31 pm

Help with my school project

October 23rd, 2006, 8:58 pm

Hi,In my finance class, we are left to do a simple research project that involves data analysis and regression testing. So, I picked a simple topic:Is the S&P Index negatively or positively correlated with the US Dollar Index, I/R Index, and the Crude Index. The 1st step is to get some responses from the forum as my primary data. I will use the actual settled prices from these indices to run a regression to see how acurate the predictions are. I am planning on using the following linear equation:S&P Index = A * US Dollar Index + B * I/R Index + C * Crude Index where A, B, C are the Beta-Coefficents. I might change the model depending on outliners, etc, etcSo, can anyone care to respond to the question???Thanks,Ram
 
User avatar
Alan
Posts: 10165
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Help with my school project

October 23rd, 2006, 10:26 pm

If you're actually using index levels, it would be amusingto run a regression of the S&P500 level vs. just the price of crude oil over,say, the last 50 years. Of course, they should be strongly positively correlated.This would be a nice example of how to draw exactly the wrong conclusionfrom a regression.regards,
 
User avatar
rmeenaks
Topic Author
Posts: 186
Joined: May 1st, 2006, 2:31 pm

Help with my school project

October 23rd, 2006, 10:32 pm

Wouldnt the regression point to a possible level where the S&P index might be if we predicted the crude prices, dollar index, and the interest rate? I dont see what is wrong. The regression would try to weed out the conditions that are not significant. Can you think of another way to do this or other terms I can add or remove?Thanks,RamPS: We got one guy doing the significance of lite beer vs regular beer for health reasons :-)PPS: We were told to pick ANY topic that interested us as long as we can prove or disprove this statement with regression and secondary data...
 
User avatar
Alan
Posts: 10165
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Help with my school project

October 23rd, 2006, 10:42 pm

QuoteOriginally posted by: rmeenaksWouldnt the regression point to a possible level where the S&P index might be if we predicted the crude prices ..You make my point. Do the regression vs. crude (using the 50 year history) and then post your prediction for where the S&P500 indexwill be if crude reaches $200/barrel. That may be more enlightening than anything I can say.
 
User avatar
CactusMan
Posts: 2990
Joined: October 27th, 2005, 8:26 pm

Help with my school project

October 24th, 2006, 3:37 am

What if the relationships are non-linear?
 
User avatar
rmeenaks
Topic Author
Posts: 186
Joined: May 1st, 2006, 2:31 pm

Help with my school project

October 24th, 2006, 8:03 am

Well, that is where SAS comes in. The analysis of the data and any patterns that are seen will determine if I will need to change the regression model accordingly. I might need to change it to an exponential model, etc, etc. Once the data is plotted, it should somewhat tell me how to set up the model. I know it will not be perfect, but it should point to some direction. After thinking about it some more, it dawned on me that I might have to change it from the S&P index to maybe a single company like IBM, AT&T, or Microsoft to keep the units of the independent variables and dependent variable in sync...Ram
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On