Breush-Pagan Test Durbin-Watson Interpretation Multicollinearity
HAHAHAHAHHHHHHHHHHAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Welcome to what we went through this past year when they decided to move it down from level 3, we all bent over backwards learning this shite for a whole weekend and then NOT GET A SINGLE QUESTION TESTED ON IT!
Don’t worry zimzim, just hope you remember something from last year’s studying of Time Series! Have you noticed the LOS on steroids for Time Series this year? My bet is that it will definitely be tested in 2009…
CFA.Rhythym, It took a few times reading it over, but it eventually clicked and wasn’t too tough to review and retain.
Yeah, they bump it up a notch in L2. I thought everything was interesting though. Most of it was just a rehash of some college stats courses for me, with an econ focus. I liked it. However, I had the same issue zimzim had with the LOS imported from L3 NOT getting tested … so effing annoying. I spent extra time on some of that material because I thought it was essentially guaranteed to show up. I wish I had spend some of that time on other areas (like BOP).
Yeah I finished reading the Quant book and I have no idea what’s really going on. All I really understood was the initial topic of Correlation and Regression. I’m waiting for Schweser since they explain things better in my opinion.
CFA.Rhythm Wrote: ------------------------------------------------------- > Breush-Pagan Test > > Durbin-Watson Interpretation > > Multicollinearity Post your questions …
my $$'s still on just a regular old regression/anova table and subsequent questions (some maybe like if the T stats were insignificant and what “problem” might cause this and how to correct or whatever referring to the fun stuff above). Maybe 1 or 2 q’s on time series. Or maybe 6 full q’s? Yikes that would be frightening. I’m doing time series readings this weekend. Good times, good times. Ok quick pop quiz- if you omit an important variable in a regression, will the following be reliable/accurate/not biased in your regression (yes/no): T stats? coefficient estimates? how about if heteroskedasticity is present (t stat and coeff est)? serial correlation? Bonus- if pos serial correlation, T stat bigger or smaller than it should be usually?
bannisja Wrote: ------------------------------------------------------- > my $$'s still on just a regular old > regression/anova table and subsequent questions > (some maybe like if the T stats were insignificant > and what “problem” might cause this and how to > correct or whatever referring to the fun stuff > above). Maybe 1 or 2 q’s on time series. Or > maybe 6 full q’s? Yikes that would be > frightening. I’m doing time series readings this > weekend. Good times, good times. > > Ok quick pop quiz- if you omit an important > variable in a regression, will the following be > reliable/accurate/not biased in your regression > (yes/no): > T stats? > coefficient estimates? > > how about if heteroskedasticity is present (t stat > and coeff est)? > serial correlation? > Bonus- if pos serial correlation, T stat bigger or > smaller than it should be usually? For the life of me, I can’t remember which one leads to a smaller coef est in denominator…I think it’s Hetero, therefore, T-stat would be artificially bigger. Reverse for Correlation, denominator too large, therefore T-stat too small. Or I could be talking completely outta my arse? Banny…where were you when I needed you in the 11th hour last night…aaarrrgggghhh. Needed female perspective on things!
i was sleeping! showered after soccer then passed out. midnight phonecalls when this kid is at work these days around 6am with all of the madness around shorting/stock loan right now? your VM message was hysterical though. glad you signed up.
Yeah, sorry it was so late. I thought for sure you’d be out on the town celebrating your investments’ windfall returns over the past few weeks…or at least the BoSox?
we’ve been busy and it’s been a good busy. we’ll see come bonus time what kind of windfall we’re talkin’. maybe i’ll be your sugar mama. as for the prob above- i don’t have my book so completely off top of head, I want to say hetero you have too small std errors, too big T stats, reject the null when you shouldn’t (type 1 error), but ok coefficient estimates. anyone feel free to correct me b/c while i read it recently, it doesn’t mean it’s sticking! i think omitting a variable f’s with the coefficient estimates and the t stats making your regression junky b/c you forgot something important. pos serial corr/hetero have same sorts of symptoms. multicollinearity the t stats all look insignificant but the F is big. how’d i do coach? you’re way better at quant than i am. it’s my worst by far. maybe this weekend if i really do sit down and study, i’ll throw a few more quant q’s up here, see if anyone’s started or it’s just me and my quant books this early.
bannisja Wrote: ------------------------------------------------------- > we’ve been busy and it’s been a good busy. we’ll > see come bonus time what kind of windfall we’re > talkin’. maybe i’ll be your sugar mama. as for > the prob above- i don’t have my book so completely > off top of head, I want to say hetero you have > too small std errors, too big T stats, reject the > null when you shouldn’t (type 1 error), but ok > coefficient estimates. anyone feel free to > correct me b/c while i read it recently, it > doesn’t mean it’s sticking! i think omitting a > variable f’s with the coefficient estimates and > the t stats making your regression junky b/c you > forgot something important. pos serial > corr/hetero have same sorts of symptoms. > multicollinearity the t stats all look > insignificant but the F is big. > > how’d i do coach? you’re way better at quant than > i am. it’s my worst by far. maybe this weekend > if i really do sit down and study, i’ll throw a > few more quant q’s up here, see if anyone’s > started or it’s just me and my quant books this > early. Hahahah, my bad. It’s not the coef est in the denom, it’s smaller std errors that lead to inflated T-stat when coef est is in the NUMERATOR, right?! That’s purely off the top of my head and not having looked at that shite in four months.
that’s right, you still have the magic zim. your durbin watson stat is right at 2.
bannisja Wrote: ------------------------------------------------------- > that’s right, you still have the magic zim. your > durbin watson stat is right at 2. NICE ONE! I always thought my Durbin-Watson curved a little to the left?
Did anything change between 2008-2009?
not much changed. a few new readings in alternative investments, a sprinkling here and there in econ, fsa, etc… but overall, most is identical. a lot less changed year over year it seems this/last than last/the year before. i think quant, FI, derivs when i glanced fast was 100% the same. so the time series i didn’t learn last time i need to learn now. seasonal lags are my friends!
zimzim78 Wrote: ------------------------------------------------------- > HAHAHAHAHHHHHHHHHHAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA > > Welcome to what we went through this past year > when they decided to move it down from level 3, we > all bent over backwards learning this shite for a > whole weekend and then NOT GET A SINGLE QUESTION > TESTED ON IT! AND THANK GOD FOR THAT 
I was under the impression that Quant for L2 has remained the same from last year. Can someone please clarify if this. Thanks.
dashingdude Wrote: ------------------------------------------------------- > I was under the impression that Quant for L2 has > remained the same from last year. Can someone > please clarify if this. Thanks. Dude, Rhythm was making reference to the increase in difficulty from the perspective of a level 2 virgin (ie. got popped on level 1, but hasn’t seen the Dirk Diggler yet).