# There's nothing like the smell of Quant in the morning

Some lame tire industry sales regresstion: Sales(t) = 504.44 + 13.18INDRUB(t) + 9.88CAP(t) + 11.66INDTR(t) + 5.92MILES(t) + 10.45PERS(t) T-values: intercept 1.02 INDRUB 0.67 CAP 1.48 INDTR 0.83 MILES 1.76 PERS 1.13 n = 36 st. error of estimate 1637.38 R2 = 0.955 Variable estimates for 2003: INDRUB 144.1 CAP 90.7 INDTR 143.9 MILES 2359.8 PERS 36.6 1. The 95% confidence interval for PERS is: a. -8.434 to 29.334 b. Insignificant c. -8.416 to 29.316 d. 8.190 to 12.710 2. The goodness of fit for a multiple regression equation is given by: a. Standard error of estimate b. F-statistic c. R2 d. t-values for the regression coefficients 3. What’s the forecast of sales for 2003? a. 19,330.154 mn b. 504.440 mn c. 19.330 mn d. 1,637.38 mn 4. Which is a symption of multicollinearity? a. Low R2 but coefficients statistically significant b. High R2 but coefficients statistically insignificant c. Regression coefficients hardly change when independent variables are dropped d. High standard error of estimate for the regression Open questions: 5. is CAP different from value 6.89 ?

A1. D? 8.14254 - 12.75746 A2. C? R2 is the one A3. A? 19330.138m A4. B? despite of having coefficients statistically insignificant, we are having high R2 - then there is MC A5. Can’t figure this out… don’t we need the STD of the slope for CAP here? t(calc) = [9.88 - 6.98]/(?)

man this is an aggressive question when it’s this late! 1. b 2. a 3. c (doesn’t specify if its in million or thousand though, or did i miss it?) 4. b 5. no?

You can use the t-value to calculate the standard error for Q5.

dinesh.sundrani Wrote: ------------------------------------------------------- > A5. > > Can’t figure this out… don’t we need the STD of > the slope for CAP here? > > t(calc) = [9.88 - 6.98]/(?) Dinesh, I smell victory lol. That was a bit of a weird flick though… anyhow, you dont need std. dev I dont think. You need s.e. which you can get from the t-value and the estimate, which are both given.

I see mcpass beat me to it.

guys… I am far far far light-years away from Quant (I am on PM now… so you could imagine) I don’t remember the exact procedure to get the ans to Q5, but this is what I think. 9.88-0/1.48 = 6.67567568 Ho: b0 = 6.98 Ha: b0 != 6.98 t(calc) = [9.88 - 6.98]/6.67567568 = 0.434413 t(stat) = T(30, 0.025) = 2.042 Decision: Fail to reject null Conclusion: is CAP different from value 6.89 = NO? Let me know if this is correct?

mcpass Wrote: ------------------------------------------------------- > You can use the t-value to calculate the standard > error for Q5. You don’t need standard error here. The t value is 1.48 for CAP_coeff. Given that this is the t statistic for a hypothesis test of CAP_coeff= 0 (and it’s still not significant at any reasonable level), then it’s pretty clear that for 0

Spot on Dinesh. You got the rest right too.

wow mcpass… you made my day… I owe you one nasty question…

wow. i guess i need to brush up on my quant. can you go over how you got #1. i obviously plugged in the wrong numbers then gave up on it. and on 3, how did you know to use mm. or thousands? sorry, i’m getting sleepy over here, don’t know how dinesh does it!

Do it when you wake up tomorrow. One b*tch question to get the day started

lola Wrote: ------------------------------------------------------- > dinesh.sundrani Wrote: > -------------------------------------------------- > ----- > > A5. > > > > Can’t figure this out… don’t we need the STD > of > > the slope for CAP here? > > > > t(calc) = [9.88 - 6.98]/(?) > > Dinesh, I smell victory lol. That was a bit of a > weird flick though… > > anyhow, you dont need std. dev I dont think. You > need s.e. which you can get from the t-value and > the estimate, which are both given. lola good to have you back. You should be well set to score a 120 on 120, right?

thanks Dinesh I wished…

cfasf1 Wrote: ------------------------------------------------------- > wow. i guess i need to brush up on my quant. can > you go over how you got #1. i obviously plugged in > the wrong numbers then gave up on it. and on 3, > how did you know to use mm. or thousands? sorry, > i’m getting sleepy over here, don’t know how > dinesh does it! Hi cfasf1, here is what I did… A1. 10.45 ± 1.13*2.042 A3. 504.44 + 1899.238+896.116+1677.874+13970 + 382.47 Subjectve Ans: Redbull does it to me

you, cpk, delhirockz were ridiculously brutal at L1. I dunno where delhirockz disappeared all of a sudden?

i should go to sleep. i think i was using CAP numbers and on the third question got the correct answer then went a step further to ask myself whether it was in millions or thousands… wtf. i’ll hit it hard tomorrow… I get up at 4:45am to get to work an hour before the mkt opens. you east coasters have it easy… i’ve been up for almost 21 hours with no drugs, except coffee, that is. have fun studying…