I don't know who else to ask so...

Ok, it is time for Glenn to get a job. I just got a BS in Computer Science and currently work part time at an investment bank while paying my rent with poker. I don't want to be a professional poker player or plan on doing so for very long, but it has been a great way to relax for a while and gives me some time to screw around without being poor. Anyhow, I could easily get a job programming applications, but I find that about as enjoyable as playing 1-5 stud with a bunch of rocks in cowboy hats. I am quite good at math, and although I don't have much classroom experience in finance, I have a strong interest as well as some exposure from my job and personal reading. If all else fails I will probably end up being an actuary, but I just found out that there are people called "quantitative financial analysts". They analyze investments using math and computers. Does anyone know anything about these jobs? If I had a math PhD it would be relatively easy to get a job like this and I would make an ungodly amount of money right away (I heard about one U. of Chicago PhD that was hired as an options analyst out of college at 750K...nice, huh?). Since I don't, and really don't want to go back to school for 4 years it is a little harder. The bankers I work with live in a public finance vacuum, and most of the other smart people I know are medical students or programmers so no one really knows what I am talking about. Can anyone give me an idea of what types of entry level jobs there are in this field and what skills I should have that will help me (e.g. Stochastic Calculus)? I have no idea if any of you will even know what I am talking about, but I figured it was worth a shot... Any other career suggestions are welcome (being an independently weathly guy who doesn't have a real job is my first choice).


Sorry for the long post, thanks if you made it this far...


-Glenn

I made to the end of your post and you're right I don't have a clue about "quantitative financial analysts". However, maybe a corporate Head Hunter could help answer your question. They are in a recession currently and might very well have the time and a clue. Hope this was helpful...


SPM,...play long and prosper...

Try contacting Jonathan Kaplan (you can find him on rec.gambling.poker).

Quantitative financial analyst jobs are realtively few and far between, and with a BS you aren't qualified for one. Any work you get in the financial sector would be indistinguishable from software application development at your level, and it might take you 5-10 years to advance beyond that. Have you thought about a MS? Opens a lot of doors, and only takes 2 years (maybe 1.5). You will use very little of the math that you seem to have an interest in if you take a BS-level job, and you probably don't have enough background in the math you do need (although you may be able to pick up what you need as you go). Here I mean stochastic processes, Kalman filtering, Weiner filtering, and advanced linear algebra such as optimizaztion theory.


Have you thought about software engineering? Tons of jobs in that, many entry level (BS). I am talking about things like real time operating software for anything from cell phone modems to networks to consumer electronics to microcontrollers in car engines. I work with lots of these guys, may be more interesting to you than application development. Helps if you know some digital signal processing.


Good luck,

Lance

I'm certainly not an expert in Kalman filtering but I'm curious as to how it applies to Quantitative Financial Analysis.

I believe that linear predictive filters with input corrupted by Gaussian noise are used to model things like stock performance. Basically, you have a signal (S&P 500 index) corrupted by noise (day to day "random" fluctuations due to things from earnings reports to Greenspan speeches), and you'd like to separate out and predict the value of the signal to the financial benefit of your customers.


I'm an engineer, not a financial guy. I thought about going into (mathematical) financial analysis at one point and did some admittedly light reading on the subject, so that's the basis of my opinion.

Your explanation makes sense to me. You probably know that Kalman filters are used extensively in airplane navigation where noise in the form of air turbulence and such is filtered by using the positions output from two different navigation units to yield a precise position for the aircraft. They work great in airline navigation as I wonder how well they work in the markets? Very interesting and thanks.

I didn't know that, although I knew that they are used in position-locating systems in general. That's essentially what you are trying to do with the stock market, guess its "position". Fascinating that two seemingly completely different fields like this are just two flavors of the same math problem, huh?

Hey Tom,


I just looked at your profile. I'm from Canton, OH, I moved out to San Diego after I graduated in '97. I came close to going to Case-Western but ended up at Purdue.


How do you like studying Econ? What's your plan when you graduate? Like I mentioned above, I thought about going back for a business or econ degree at one point to do analysis, but kinda decided it wasn't for me. Good luck with it though, I think it's a very interesting field.


Lance

Well I'm a long way from finishing my degree in Econ but I'm really enjoying the courses I'm taking. I haven't decided what I'm going to do with it and perhaps it will be something I never do anything with. Definitely one thing I'm considering is financial analysis though as I too find it very interesting. Good to here from a fellow buckeye although there is a good reason that you live in San Diego and not Canton and that I live in Albuquerque and not Clevetown /images/smile.gif.

I went to CWRU too and I still live in Cleveland :-(.

Thanks. I realize that I can't be a true quant without a PhD, but I have seen a few entry level derivatives jobs that might be interesting. I don't like the idea of getting an MS because I probably couldn't get into a decent Math MS program without having been a math major and what I would study for a CS MS would be just more of the same (I could have done BS/MS for 6 extra classes, but the 6 extra classes would have just been a waste of my time--not financially, but I really wouldn't have learned anything useful). Finance MBA is probably where i'll end up...even though I know for a fact that these programs teach you a lot of nothing it will allow me to get into the industry I want to be in because I will have a piece of paper that says I know how to do what I already know how to do (seriously, I do my business school friends' homework sometimes :-) ).

i went just a little south,,college of wooster...when i started classes, there were only two courses regarding computers, and we still used slide rules, wow, lol, giving my long of teeth away,,,gl

If the winters don't bother you Cleveland is a good place to live.

If you guys are really curious about time-series prediction in financial markets, here are a few posts that get into some of the basics:


http://www.twoplustwo.com/cgi-bin/newforums/genpok.1.pl?read=29015


http://www.twoplustwo.com/cgi-bin/newforums/genpok.1.pl?read=29217


http://www.twoplustwo.com/cgi-bin/newforums/stocks.pl?read=1589


http://www.twoplustwo.com/cgi-bin/newforums/stocks.pl?read=1679


In summary, if you filter out a signal and choose to call it noise, then you want to do a "counter-trend" trade. If you accept a signal, then you want to extrapolate it by doing a "trend" trade. The day you figure out what to filter out, and what to extrapolate, you will be richer than G-d.


The latest, most advanced tool I know of is called "interleave" - but I won't get into that here:)


eLROY


P.S. Here's some more brain chewing gum:


You know those sand waves that form on the desert, or on the ocean floor? They slowly move in the direction of the current, so that if you were to fixate on a certain spot, it would rise and fall over time. The heavier the wind, the larger the waves.


I really believe trend following works the same way, where the sweet spot is always moving, and people are always chasing it. It emerges from nothing, and moves in the direction of the trend. People evolve around where it is, not where it will be, and so are perpetually a hair behind.


In the 1980's, a strategy that just traded the current - the trend - worked great. But the more people began to use such strategies, the waves began to form, as if the traders themselves were the sand. I think soon the current will be totally priced in, and all that will be left are the waves.


Oh, and more chewing gum:


There is a great metaphor for what I am about to say in the world of vinyl records. Vintage record collectors get a lot of records that have been played to death over the years. The groove is worn down to where it delivers no signal. But usually it has been worn down by a single owner, using a single needle. If that needle has been square - wearing down the sides of the groove - you can use a pointy needle to probe into the middle of the groove where it never reached yet. Or if their needle was really pointy into the middle, often you can pick up some signal off the sides or walls of the groove using a squarish needle yourself.


In futures, I say stay out of the middle and go for the walls. Or, if there is some texture in the middle, use it as a contrary indicator! And I think volume is your most robust indicator of where the "middle" is. It doesn't require you to make any assumptions about what strategy or indicators your peers are using.


Moreover, I don't think buying volume spikes - to the extent you can call them spikes without a crystal ball - is consistent with the Turtle philosophy anyway. I think buying volume in general is correct, but you can only buy a spike if you assume the volume will be nearly as high again the next day of the trend. At best, you can assume those left behind by the spike will straggle in slowly over the next couple days, waiting for a counter-spike in price.


Looking at the chart, I don't see outsized price moves on volume spikes followed by outsized price moves in the same direction on similar volume. More often, it looks like a single day of high volume is a contrary indicator. One caveat would be, looking at a day bar, if you were somehow getting in a lot of volume at the beginning of the day's range - and obviously somebody is getting a lot of volume done somewhere, so who knows, maybe you can catch the front end intraday.


But if you buy at the top of the volume-spike day, that looks like a fairly consistent contrary indicator. Only in the days following the adjustment/recalibration day, if they also have high volume, they seem to be a positive indicator. You can devise a theory about this. You might say, well, if volume is constant or declines as the price rises, then that is ambient buyers getting left behind. But if the volume gets exacerbated as the price rises, that is people buying because the price is going up - as opposed to not buying, which would mean you should be.


The point is that, on the one hand, you don't want to make a habit of ignoring price moves on huge volume. But on the other hand, you don't want to be getting washed in and washed out at the same time as everyone else. And if all-period rates of change break out, and the daily range breaks out, and the price breaks out of recent ranges, it's going to manage to hit everyone at once - and create a big day even if there wouldn't otherwise have been one. So what are some things you might backtest?


The first thing I can think of is to ignore the first big volume/breakout day, and then use a very fast indicator starting the next day. The logic is, that if you are assuming that a whole lot of trend traders are pushing the price around at the slightest tick in either direction, then you want to get in in front of these waves. This might seem fast, and like you couldn't hope to do it with a lot of size - but - after all, volume is high so maybe you can. You catch them going both ways. At the same time, you are employing wait-and-see, since you missed being fast on the first spike.


At the very least, you might use a very fast counter-trend exit - and then a slow or time-based re-entry in the direction of the trend - so that you're still trying to trade the trend, not the waves of trend traders. If you make the assumption they're all long - so they can only get short - you would have to believe any subsequent up move on volume, but ignore any subsequent down move on volume. On the flip side, you can anticipate exaggerated followthrough on a slight down move, but plenty of time to catch any inching, persistent up move. But remember, I'm always talking about the day after the spike. Any day inside a large trend, or at the extreme, calls up a different set of rules for anticipating the spike - because people are more likely to take an exit signal on a trend they are in then an entry signal in a dull market.


You're basically attempting to use an asymmetric strategy at the elbows, where the trend turns, or where the trend emerges. Maybe you speed up in one direction and slow down in the other direction after the initial spike. Maybe you speed up both sides when you have pulled back inside an existing trend, and slow down when you are at new extremes. There either is an underlying trend or there isn't, and there either are a lot of traders getting long or there aren't. I would say that you can say with a good degree of confidence whether the volume at least includes trend traders entering or reversing. Where there are not, you have a positive ID on a trend.


In summary, where you know you are buying alongside other traders at the front of a trend, you can at least way until you see if they get paid off or whipsawed before you climb on - try to go outside them. And when you are on a trend alongside other traders, you can try to climb off faster, by coming inside them. And an extra clue might be out-sized volume spikes - which you're either trying to get in front of in side an existing trend or behind in an emerging trend. Speed up or slow down, in a way it makes no difference, you're just trying to lock onto something outside the noise, and the volume spike is one of the things that pushes you to one side or the other. Of course, if there's even more volume where you get pushed to, maybe you have a trend!

Yeah but are Kalman filters as good at determining stock prices as they are at determining aircraft positions?

Hey Glenn,

don't know if this is so burried that you won't see it but I'm a quant financial analyst. I'd say one of the more important skills is a good knowledge of statistics. The bad news is that this is probably one of the most difficult time to try and get into the industry. I'd say try to find a derivatives related job or a job working with fixed income. working with a complex financial instrurment that has a few demensions to it is a good way to get your brain wrapped around the ideas involved.


rob

Here are the correlated performances of two "trend" traders.


http://iasg.pertrac2000.com/SnapshotPT.asp?ID=101


http://iasg.pertrac2000.com/SnapshotPT.asp?ID=105


See if you can deduce which one began using a new Kalman-ish filter and when?


eLROY