Let's talk about stochastic calculus first, since it took 75% of my brain processing power in the past year. I was just biking on street today and thinking, the field of derivative pricing in finance is analogous to doing a deadlift exercise. If the main muscle to do a deadlift exercise is gluteus, the main field that one needs for derivative pricing is stochastic calculus. And if the secondary muscle groups that support a deadlift exercise is squat, abs, and lower back, the three fields that are intertwined with stochastic calculus in financial engineering is probability theory, partial differential equations, and numerical methods (monte carlo, solving pde numerically). The beauty of derivative pricing extends beyond that, as much as deadlift is a fullbody workout. One would need to understand measure theory, algebra to derive the build up the framework for stochastic calculus, as much as one need a strong hands and shoulder to lift the bar of weight up. One would also need statistics to estimate the parameters in a pricing models, as much as one need a strong pect (chest muscle) to keep the torso up when he does a deadlift.
So that is the beauty of derivative pricing. A exact mathematical branch that utilises tons of other fields. Where to go next for my future me, learn stochastic volatility model such as Heston, SABR, SJJ...
Now comes time series. Time series (statistics in general) is like chest muscle like I mentioned above. I learned it in the second year of university, just as I do a lot of chest exercise when I started working out. It was my most favourite. I remember reading the note from MIT Opencourseware, and I was amazed with the way time series was presented: just like an analysis course. But the different was time series is wildly used. I learned the Granger causality test for my final paper. It was a beauty, seeing the improvement from a normal regression model to a Granger causality test. Then soon as I knew the field of financial engineering, I forgot time series almost completely. Just as I ignored chest exercise after Quy taught me how to do full body workout. Then the Lent term of my third year came, I talked to a Luc, a dude I knew from the MFE courses at LSE. He has a master degree in econometrics before doing this MFE degree. And he was talking to me about principle component analysis, then DSGE models, then factor models. I had no fucking ideas back then. Then I got some interviews at hedge funds that require time series and statistics. I googled and get amazed another time to learn the beauty of DSGE model. Something that requires Markov chain, Monte Carlo methods, Baynesian inference, and of course it is time series. Something that is being used by Central bankers to forecast macroeconomic variable. Something that Luc told me is of post-graduate level. I was like daaaamn. This is like seeing a hot-as-hell girl in high heel and short skirt walking in Oxford street at night. So I go back to the MIT Opencourseware, where I found all the things such as VAR, principles component analysis, factors models, and DSGE at the end. I told myself: this is my road ahead, learning these each by each.
Too much for one paragraph, let's start a new one. So then I found myself reading the book Quantitative Trading to prepare for an interview at a high frequency trading firm. Once again I found factor model, principle component analysis, cointegration. Hedge funds are indeed using these models. The feeling is like when a Muslim found his god. I know what I need to learn now.
Ok to sum up. Plan for my senior year is:
1. Write a honor thesis for math in stochastic volatility. Apply the Heston model to price a FX Knockout Option. Learn to solve the PDE numerically in C++. And finishing learning Monte Carlo in C++.
2. Master the VAR model, factor model, cointegration, and principle analysis. Apply these to find a market-neutral strategy. Maybe something related to forex, metals, bond, and macro-driven assets. I don't like stocks as much as these.
While I was wasting time writing blog, Obama is skating.
No comments:
Post a Comment