Practical Regression Examples?

Anybody know of any free online resources I can go to that teach practical regression analysis? Looking for some resources that are pretty beginner-to-intermediate level.

I recently found a pretty cool course through Coursera which has to do with single and multiple variable regression (see link below). I understand that this is pretty similar to what we saw in the curriculum, but I’m looking for some sites I can go to learn more about conducting the regressions myself as opposed to simply being given an ANOVA table.


there are tons of youtube videos out there. I believe the ones made by a user that goes by the name of Nikos Ntoumanis are not bad. I do not know however if that covers your needs.

She uses SPSS, which you can get for free (14 day trial)

Give it a try.

You can look at introduction to regression with various software packages. I’d recommend just picking one and reading through it first to understand what they’re doing and to increase your background knowledge. If you are at a university, you can usually get cheap licensing for a year or so on SAS or SPSS, but SAS University is free online with some reduced, but still sufficient, functionality as far as I know.

You may have to click in a few more links after the first page. Feel free to PM me if you’re having trouble finding it.

Also, Coursera can be good, they have some decent looking courses from JHU for regression modeling-- usually a decent mix of theory and application. Just make sure you look up the background of the instructor. You’d ideally want someone with at least an MS in statistics or biostatistics. I would avoid books or courses written by people who do not have at least an MS in stats/biostats. Six sigma is a joke also-- I can’t tell you how easy it is to find a book written by a person without a formal statistics background where there are numerous critical errors. It’s all too common on Youtube/blogs also-- it’s not hard to fool people into thinking you know what you’re doing because you have an extremely rudimentary understanding of a topic and know how to click buttons in a program to generate output (not knocking the above recommendation, I’ve never checked out that person’s content). In general, you won’t get good content from people who lack the formal background.

For what it’s worth, the curriculum is pretty pathetic for statistics-- it won’t hurt you to see it again from a credible source for statistics. In general, you will learn something new when seeing the same topic from different presenters.

In my opinion, the stronger someone is in Statistics, the more difficulty he will have on learning Statistics from the CFA Curriculum. Statistics in Curriculum is incompatible and not enough rigorous compared to Statistics at a good university.

I exaggerate but for a statistician, the first step is to forget (at least temporally) what he learned before. After this step, he can memorize the Curriculum. Without the first step, he must become crazy.

If you begin with the Statistics in CFA Curriculum and learn Statistics in a more serious course, I think it’s good (good for the exam). But do the opposite, I don’t know if it is a good idea (for the exam). I said " for the exam".

I’m going to respectfully disagree. I had a pretty decent stats background before cracking the CFAI books open. For the sake of completeness, I read the stats portions of the curriculum quickly to make sure I wasn’t missing anything in my skill set. I was able to breeze through those chapters and didn’t do any of the EOCs. I consistently got 6/6 on the online topic tests and even corrected the CFAI on some theoretical/technical points in some answer explanations :grin:.

I guess the only headache I had was seeing how poor of a job they did on many stats topics and things they actually have that are incorrect in the text-- things readily verifiable with and actual statistics book.

Having a decent background before definitely didn’t hurt me in the exam. It gave me a section to consider as “free points” because they were asking questions at a level well below where I had previously studied. This wasn’t necessarily the case for other sections.

In my opinion, there are different schools of thought (sorry, I’m not sure the term I used is correct term, but I think you understand what I mean). In my university, I learn to always ask why (for example, why we have this notion, why this theorem is true,…) before trying to use (for example, how to use this theorem to answer this question). And I know the Anglo Saxon system is more pragmatic.

CFA exams clearly illustrates the pragmatic approach. These exams don’t ask candidates demonstrate anything, they only ask candidates use mathematics notions given in Curriculum. And if you have some questions like “why the t-statistic follows the t-distribution with n-1 degree of freedoms?” or “in the Breusch-Pagan test, why the test statistic is n * R^2?”, 'Why IR = IC * TC * sqrt(N)?" …, you will have real headache and won’t have enough time to study other sessions.

My strategy is to accept everything written in CFA Curriculum, do not ask why (which is not in my nature), to memorize all notions and to apply them to answer the MCQs.

It’s just my point of view, I don’t want to get into any debate :-).

PS: As you all know, CFA certification is normally for people working in investment management, specially for people who want to become portfolio manager. Suppose the Dow Jones index increases 3%, the first question a portfolio manager asks is “what must we do now” rather than “why does this index increase 3%”.

I get what you mean, now. The books are pretty shallow and poorly done, so even if you do ask yourself these important questions, you won’t likely find the answer in the CFAI book.

I agree with your idea to always ask questions. More or less, the CFA exams aren’t requiring a whole lot of that thinking in my opinion. Even many of the conceptual questions are ideas explicitly covered in the curriculum. They showed the same house with a new paint job…

I don’t know why you guys still bother about the completeness of the Quants part of the CFA curriculum (1 and 2).

At CFA 2, in the quants part, you are not taught Statistics, you are taught Econometrics. This discipline relies on mathematics, statistics and economics. Its more than simple statistics.

The goal of the CFA 2 quants topic is to put candidates in context of information assessment and interpretation. A good method is using regression, perhaps the best. We are not even told about maximum likelihood methods of regression…

As I said before, probably 2 or more years ago… If you are looking for a formation in statistics or econometrics, you can’t simply rely on the CFA Curriculum. The same applies for derivatives, equities, fixed income, corporate finance, alternative investments. God, the whole curriculum.

The CFA Program is two inch deep and a mile wide. I will never forget this quote from a fellow candidate in this very forum. Wise words.

I don’t want to be negative, but don’t put all your coins thinking the CFA program will teach you everything you will ever ever need for professional success. It’s just impossible.

The “econometric” techniques in the CFA L2 book aren’t really specific to economics.They’re taking basic statistical techniques and using economic and financial data sets, so I wouldn’t really say that’s enough to classify it as more econometrics than statistics. Many of the topics covered originate from and are general ideas in statistics, not economics. They don’t do a whole lot of teaching you something new in econ but rather they focus on statistical methods.

My main issue with is the introduction of ideas for testing purposes without the proper development of the idea. You can’t claim to provide a world class education when you’re teaching fundamental concepts incorrectly.

Of course, there is no discussion about the lack of deepness of CFA Curriculum. What i meant (I wrote it in the fourth post of this topic) is about another problem.

In my opinion, even if someone (person A) is not strong in Statistics (or Econometrics, or Mathematics), he isn’t disadvantaged compared to those strong in Mathematics. In contrast, maybe he can learn Curriculum better than a person B, who has a good enough background in Maths, Because the person B is used to working with well-posed problems and clear answers, he can’t accept what Curriculum imposes on him. As if someone who used to live in a 400m2 penthouse and now he must accept to live in a 15m2 room.

Take, a European driver, for example. Do you think he will get the british driver license easier than a 15 year old child? The answer is not obvious. As you know, the european driver is used to driving on the right, and in UK, people drive on the left. The first step he must do is to forget what he learned before, if not, he will be always confused between the two driving rule systems and his driving reaction won’t be quick enough to drive in UK. In contrast, the 15 year old child, who hasn’t driven in his life, can learn UK driving rules without any trouble.

Going back to the CFA exams, in my opinion, if someone think he is not good enough in maths but wants to pass the exams, learn maths outside the Curriculum is not always a good idea (but of course, if his objective is to get knowledge, learning from more rigorous course is good).

What I said, you can’t take literally, because it depends on the type of other courses taken. But in general, the more different the courses are compared to the CFA Curriculum, the more perturbation you can have on the exam because the key to pass the CFA exams is to learn and use what given by CFA Curriculum (and not by other courses), and you must answer as quickly as possible (you won’t have much time to think). Just try this course and this examwhich are very different to the CFA exams, if you get used to this course, even you can clear easily the exam, you won’t have enough time to answer 6 Quantitative methods questions in 18 minutes.

I’m pretty sure that we don’t.

:), thank for the remark, I mean in UK.

Are you speaking from personal experience? I’ve taken courses and had exams very similar to the one you posted, and I can assure you the CFA QM questions were a breeze in comparison. I think it took me about 8 minutes, on average, to complete each topic test of 6 questions for QM and about 12 minutes for every other subject. I missed one question in the QM topic tests because they had no correct answer-- eventually something the CFAI admitted. They agreed to remake it or pull the question if they couldn’t remake it (but I’m not sure what came of it). The questions posed on the CFA exams are very light applied questions, and the theoretical questions aren’t really mind-blowing-- more or less, it’s no more than what you’d be introduced to in an introductory undergraduate course in that subject. I’m unsure why someone who learned more of it and learned it correctly the first time will have issues answering the CFAI questions (assuming the CFAI didn’t botch it).

I half agree with you that CFA QM questions are easier than the course. I can give you a metaphor, CFA exam is a sprinter competition, the MIT exam above is a marathon competition. There are correlation between the two competitions, that means a marathoner can do better than an average person in a sprint contest and vice versa. But a marathoner can not win an average sprinter in sprint contest. And we are in the sprint contest here.

As I said in the posts above, CFA exams (at least level 1 and 2) demand speed. The key to win the exams is to know how to memorize the formulas or notions given by CFAI, to able to recall them quickly for answering 3-minute questions. Take, for example, the IR = IC * TC * N^0.5, what we must do is just memorize this formula, and apply it. Suppose the CFAI changes this formula to IR = IC * TC * N^3, it’s indifferent to me, what I must do is just to update the formula, it’s all. For information, I have tried to read the Fundamental Law of Active Management to understand what behind this formula, but I gave up because it takes lot of time and not necessary for the exam. Suppose I know how to get this formula from scratch, won’t I need to memorize the formula? No! When I reconstruct the formula to answer one question of Portfolio Management, other candidates have finish all 6 questions.

I don’t say deep thinking skill isn’t needed on CFA exams (in fact, it seems impossible to memorize 6 books by rote, candidates must at least know how to structure the knowledge), but comparing to a statistics (econometrics, maths,…) course in a good university, deep thinking skill on CFA exams is less demanded.

In my opinion, the best and pragmatic strategy (for CFA exams) is to use CFA materials (including Schweser, Finquiz,…), if you don’t know something, just google it or ask people. If you can’t understand it, just memorize it. CFA Curriculum just give you a big picture (because this is the important thing a portfolio managers needs). Beside, if you think you are not good at maths, it’s not really a disadvantage compared to those strong at maths. And if you want to be good in Statistics (or Econometrics, …), it’s better to take a more serious course after the CFA Exam, but not before.

I don’t think I have a bad score in QM. But if I really have a bad score, I’ll think I just don’t know how to recall enough exactly and quickly some formulas, that’s all, nothing related to my mathematical skill.

I agree that you don’t need to be strong in mathematics to succeed on the CFA exams, and that poor performance on CFA QM doesn’t mean you necessarily have poor quant skills-- they don’t really test that. I’m just not seeing how taking a serious course in stats before the CFA exams could hurt performance. The analogies you gave don’t really convince me of what you’re saying. I don’t know how it is on the average since we don’t have that information, so I’ve no choice but to use anecdotal evidence from myself and a few other people on the forum; prior stats or mathematics backgrounds don’t really seem to be a hindrance to anyone during the QM portion of the exam-- if anything it’s a benefit.

We can agree to disagree though :grin:

I remember someone in this forum said Ethics is difficult for him because in his firm, the Ethics rules are stricter than in CFA Curriculum. He gets use to working with a different code of Ethics and Standard of Product Conduct, the fact that he must learn a second (the CFA CoE & SPC) will make him confused in the exam. In contrast, a student, who has never worked before, doesn’t have this conflict, he can learn CFA Ethics easier.

I said the more difference between CFA Curriculum and the serious course you took, the more difficult you will have in learning CFA curriculum. I learnt some mathematics courses before (not easier than the course in my example), but what I learnt doesn’t completely cover the CFA QM sessions. When I began to learn QM, I thought it is easy. But the more I read, the more I’m confused. I don’t like to use something I don’t know how to get it, so I tried to find out the meaning of notions given by CFAI. And one or two weeks later, I gave up and accepted the fact that I must use math without understand it, if not, I won’t have enough time.

If someone tries to learn other courses beside the Curriculum. First, he won’t have enough time (Ok, if he is not good at math, so he takes a statistics course. But he may be also not good at portfolio management, why doesn’t he take a portfolio management course? And what about Economics, FRA, Derivatives, Fixed Income,…?) Second, if he gets used to having serious courses, he may hate the Curriculum.

PierreCFA: note that we now have superscripts.

Sorry buddy, but we are exactly taught econometrics in the CFA curriculum, even better than an specialized econometrics book (because of the simplified approach the curriculum has). The quantitative part of the L2 curriculum is built on maths and statistics with the goal of putting the candidate in context of information assessment and interpretation [of economic data].

We can look at a common definition of Econometrics: "The branch of Economics concerned with the use of mathematical methods (especially statistics) in describing economic systems"

This is why you won’t find a theorem description or a thorough demonstration of a fundamental formula (despite I found some demonstrations to be honest), because this is not a statistics book, nor mathematics.

The curriculum is a mix of basic knowledge, modern tools and application of those tools to relevant economic data. A candidate, at some point in his career, could be put to read a research conclusion like this:

“…we found that the P/E ratio volatility index wasn’t statistically significant to explain the stock X risk, nor the 92% of stocks of portfolio Gamma”

A passed L2 candidate now knows that he/she must use other variables to predict stock X risk and Gamma Portfolio. He can add a valuable comment at the next Strategic Allocation Committee this Friday. That’s all !!

I think this is the spirit of the CFA Program. Not everyone will be a specialist in statistics, or maths. The common job requires a balance of a lot of disciplines and sciences. Aaaand, if you require a specialized data or input, somebody will provide :wink:

I see your point. Also I see the flaw of your thinking. You are comparing a 5-year college / university statistics or economics career formation to a 300 page section of a Curriculum with a different approach of education.

I tell you that my economics career was based entirely on theorem and formulas demonstration rather than multiple choice questions. This is why CFA exams are way easier than university exams (in my personal opinion).

You must have the mental agility to adapt yourself to a different way of testing, learning and achieving. You just can’t say “My university made me to learn this certain way, now I have to forget that approach and adapt to a test meant for memorizing and multiple choice”. You simply can’t bother over this.

CFA exams are meant for knowledge demonstration and application :grin:

Salute tickersu and Harrogath for keeping up with PierreCFA’s fierce verbosity and elaboration of his/her thoughts which he/she is commonly known for… with proper superscripts and subscripts this time :stuck_out_tongue: