I am a PhD in machine learning. I think techniques from datamining and machine learning can be well applied to finance and would like to gain an understanding of the nuts and bolts of finance in general.
Do you think a CFA is a good way to go about it? I don’t want to be a practitioner or something, but definitely would like to learn what goes into investment planning etc, and what kind of tools would be useful for finance people.
The CFA curriculum would certainly give you a good overview of “finance in general”. I don’t think I would bother actually taking the exams though, unless you like to accumulate letters after your name. I would just buy the CFAI books or Schweser books on Ebay. Tons of them become available after the exams.
there are plenty of people with advanced degrees in physics, engineering, Computer Science, etc. that have gone on to get the charter and portfolio managers. It may be worth it to you if you would like to pursue that path and have a true interest.
The CFA seems to offer a general overview finance, especially in the first two levels, however I think you would gain a much better understanding of the ‘nuts and bolts’ by doing a MSc in Finance rather than the CFA itself. From my experience of both to date, the CFA just doesn’t go deep enough and due to the nature of the exams a pass can be manufactured without a true understanding of the underlying principles that I would guess would be of most interest to someone of your thinking.
Good luck with whatever decision you make.
Note: if you are looking to aqcuire the CFA charter you would need to have four years actual finance experience which may or may not be an option for you.
If you want to understand financial markets better, the CFA is a nice way to do this without having to get a university degree.
The CQF program may be a little better suited to you, though, in that the people who apply machine learning to trading systems tend to be quants, and CQF is much more heavily quant-oriented. It’s more expensive than the CFA, takes less time, and is cheaper than a university master’s program in this stuff.
if you want to understand the nuts and bolts of finance then, given your background, it would make more sense to do a msc in financial mathematics where machine learning and your math background/programming knowledge can come to bear. the cfa is not about the nuts and bolts of finance and is not meant to be i think.if it’s just the knowledge you want (without academic/professional certs), just find the reading list used by phd finance peeps and you will be fine given your phd in something like machine learning!
I think the CFA might be useful. It’s not the same as MFE or something like that. MFE teaches you stochastic calculus. CFA teaches you about accounting, different investment types, and a lot of other layman stuff that you might want to know. The broad curriculum means you will get shallow exposure to many finance topics, which I think is what you are looking for since you have zero experience. Also, CFA does not require an application (you just pay money and sign up). Someone like you probably needs 150 hours per exam, so it’s a pretty quick way to study a lot of information quickly.
A lot of people will say “Oh you are a math guy. You should do financial math”. This is good advice if you are looking for a certain type of career. However, keep in mind that quantitative finance is quite a narrow field, and quant programs tend to be quite insular in their view of the world and finance community.
Anyway, for CFA, theoretically, you could just buy some books and study by yourself. However, it is hard to get motivated that way. Plus, you do not get any certificates for self study. If I were you, I would just sign up for L1, which costs like $600 (?). If you decide the program is not for you, you can quit after the first level.
ohai makes some good points. I suggested the quantitative route because the OP explicitly mentioned thinking that his or her machine learning credentials could be put to use in finance, and that’s clearly something that would have him or her operating in the quant space.
As Ohai points out, the CFA is good for learning a lot of the things that ordinary financial analysts have to look at and think about and getting a broader context.
Somtimes it’s kinda strange talking to quants who tend to think that an asset is nothing more than a probability distribution unconnected from the rest of the world, except perhaps by what can be regressed or arbitraged. Those aren’t necessarily wrong ways to think about them, but it is not the way most people think about it.
My main beef with the CFA curriculum, and the “new investment theory” in general.
But to the OP–I think the main reason people decide to do CFA is because they want to “prove” that they are “smart”. You have a PhD in what seems to be a very quantitative, scientific field. (I know nothing about machine learning.) I would say that you don’t need to prove nothing to nobody. And since you don’t want to work in finance, I think taking the test would be overkill.
Actually, I would go in a different direction than most on this board would tell you. I would check out the book “A Random Walk Down Wall Street.” I don’t necessarily agree with everything the author says, but it’s funny, easy to read, and will give you a very small, extremely watered down version of some of the concepts you’ll find in the CFA exams (without all the boring details about how to calculate bond duration and convexity). If you still like it, then maybe you can look at the Level 1 material or the source material.
This is a great book! It is one of the top non-text book finance books out there. I would also include the “Intelligent Investor” by Ben Graham, but “Random Walk” by Malkiel is great. In fact, after reading it you will probably be pretty well versed in general finance theory and practice. Thinkgs like valuation and the like are not covered, but if you believe in efficient markets, as Malkiel does, the market does that work for you.
I’m not sure you need to read a random walk down wall street, but the principle of that book is worth remembering: it’s surprisingly hard to use (non-inside) knowledge to do better than investing in a passive index fund on a risk-adjusted basis.
If you’re new to investing, the concept of risk-adjusted returns reflects the fact that higher risk assets should deliver higher-returns, and that if you want to get a higher expected return that is proportional to the risk of the market, it’s simple enough to borrow some money, invest more than 100% of your capital (your capital + borrowed money) in order to get higher returns in line with the market’s level of risk. In order to really use information to beat the market, you have to do better than a passive market portfolio that has been levered up or down to the amount of risk you took.
There are several different ways of trying to measure risk, and therefore several different ways of trying to measure risk-adjustment, but those are the issues.
The problem with machine learning is that markets seem to change a lot over time - in part due to economic conditions, and in part because the composition of investors changes over time (more algorithms, fewer people means that rules that made sense when there were individuals trading slowly with slow information dissemination may not be relevant for rules that make sense when you are competing with other machine algorithms placed less than 100 yards away from the exchange computers). Unlike a child, you cannot always ask a machine to explain what it learned, and see if it came to some conclusion like “Christmas is about the baby Santa lying under a Christmas tree in a manger with Joe Fresh and Barbie.” Perhaps you can see the rule it uses, but you can’t always pick up the logic that led them to this rule and pick it apart without being in danger of data mining (and attendant risks of overdetermination and parameter fitting).
Just some things to be aware of.
Try to go to your university and read some pieces in the Journal of Portfolio Management. Then maybe try to befriend some of the authors, particularly the practitioners. That is a help.