It's funny. I trot around these forums looking for ideas and clues to becoming quant-like (my background is in math and computer science, but that was like 15 years ago in college), but all I end up is more confused by talk of stochastics, GARCH, et al.With all of these great minds in one place, what is your advice on where to start? What is step 1 in the process. I've got returns data for 10 years on an index. Great. Now what the hell do I do with it?Any books/websites/journals that can help me out as well?-Dr. J

Three broad strategies come to mind:1. Throw every algorithm calculated with every possible value of the parameters at your data to see if any predict the returns seen in the future.2. Laboriously develop a very clever theory for why you think prices move and then test it on your data.3. Get a job or consulting position with companies/investors that are doing strategy #1 or #2Likely outcomes for each strategy:1. You will go bankrupt because you will readily find something that appears to work from the last 10 years, but fails to work for in the future. (see overfitting, fooled by randomness, etc.)2. You will get old (but maybe lucky) as most of the theories you create fail during backtesting because they either are wrong or they have been exploited by prior market participants. Occassionally, your theory will lead you to outcome #1.3. You'll get a nice salary, perhaps the occasional huge bonus (if your employer's strategy #2 hits pay dirt) and you might collect unemployment occasionally due to blow-ups and bankruptcies (Outcome #1).Good luck! (and I really don't mean that facetiously)

What do I want to do...okay, I want to run a comparison between different indices, looking for hedging opportunities or ways to make alpha on the spread between two (or more indices). For example: if I take returns data for all 10 US sectors (industrials, energy, financials, etc.), and take the top performer over say 10y (call this #1) and then the 2nd worst performer (call it #9), the spread between the two is the money I can make if I'm long 1 and short 9 (1 and 9 are the best pair, 1 and 10 are not, incidentally). So how do I go about manipulating the data to smooth out the garbage and get a proper signal...or do I even need to (according to writers in Stocks & Commodities mag, not detrending and smoothing data leaves you wide open for failure and wrong conclusions).Now, I realize that may be a HUGELY juvenile project quest; there aren't any GARCH or stochastics or differential equations involved...or are there?... I'm not looking at swaps or options or hedging in the traditional sense with bankruptable levels of leverage (we hedge with ETFs), but I am interested in currencies and futures.I guess the bottom line is, if someone is relatively new to the sport, how would you go about acquiring the training and knowledge in order to be a competitive quant (aside from a quality degree program somewhere)? Once I get a solid foundation under me, THEN I can sound like I know what I'm doing. At least kinda.-Dr. J

QuoteOriginally posted by: Traden4AlphaThree broad strategies come to mind:1. Throw every algorithm calculated with every possible value of the parameters at your data to see if any predict the returns seen in the future.2. Laboriously develop a very clever theory for why you think prices move and then test it on your data.3. Get a job or consulting position with companies/investors that are doing strategy #1 or #2Likely outcomes for each strategy:1. You will go bankrupt because you will readily find something that appears to work from the last 10 years, but fails to work for in the future. (see overfitting, fooled by randomness, etc.)2. You will get old (but maybe lucky) as most of the theories you create fail during backtesting because they either are wrong or they have been exploited by prior market participants. Occassionally, your theory will lead you to outcome #1.3. You'll get a nice salary, perhaps the occasional huge bonus (if your employer's strategy #2 hits pay dirt) and you might collect unemployment occasionally due to blow-ups and bankruptcies (Outcome #1).Good luck! (and I really don't mean that facetiously)Yikes. I know #1 is a fool's errand, so I pass. #2, well, I've done some of that and they all turn out useless...besides, are there any original ideas LEFT that people haven't explored (sector comparisons, local currency vs. sector comparisons, foreign exchanges vs. average rainfall in the Sudan, ratio of current astrological sign to number of reality shows vs average GDP of countries the US has NOT invaded in 200 years...I kid, I kid.) and 3, that's where I'm at (the employed part, not the blow up) but don't know how to put the pieces of the puzzle together to create something.There's always going back into porn. -Dr. J

I would look at Turtle Trader and then at Motley Fool. Read and digest then look at your questions again.

QuoteOriginally posted by: Colossus2420 What do I want to do...okay, I want to run a comparison between different indices, looking for hedging opportunities or ways to make alpha on the spread between two (or more indices). For example: if I take returns data for all 10 US sectors (industrials, energy, financials, etc.), and take the top performer over say 10y (call this #1) and then the 2nd worst performer (call it #9), the spread between the two is the money I can make if I'm long 1 and short 9 (1 and 9 are the best pair, 1 and 10 are not, incidentally)...Buy a copy of "Pairs Trading: Quantitative Methods and Analysis" by Ganapathy Vidyamurthy. It's very smart and readable as well.Quote So how do I go about manipulating the data to smooth out the garbage and get a proper signal...or do I even need to (according to writers in Stocks & Commodities mag, not detrending and smoothing data leaves you wide open for failure and wrong conclusions)... The last thing you should be doing is "smoothing" anything.Quote Now, I realize that may be a HUGELY juvenile project quest; there aren't any GARCH or stochastics or differential equations involved...or are there?... I'm not looking at swaps or options or hedging in the traditional sense with bankruptable levels of leverage (we hedge with ETFs), but I am interested in currencies and futures.I guess the bottom line is, if someone is relatively new to the sport, how would you go about acquiring the training and knowledge in order to be a competitive quant (aside from a quality degree program somewhere)? Once I get a solid foundation under me, THEN I can sound like I know what I'm doing. At least kinda.-Dr. JIt's not juvenile. It's just business. Would you think it's juvenile if I pitched an idea for "Hamburger Arbitrage?" We'll buy buns, meat and lettuce for a cost of X, assemble them, and charge customers X+1. Burger King does it well. Lots of people at that company make a really good living.

Colossus2420 - My suggestion would be to learn the relationships between your indexes and the derivatives that are based on them. Understanding how long/short positions in the indexes are hedged with the derivatives will get you a lot closer to understanding the pricing of both the underlying and the derivative. Perhaps then you will be able to detertmine when an arbitrage opportunity exists.Your price data may not be too helpful here, regardless of how you smooth, detrend, clean, it etc. The strategies you seem to be considering usually require nearly instantaneous price data and access to the market, i.e. being in the trading pit and being able to do size. The "long #1 and short #9" idea will probably not cover tradings costs, and as LTCM learned, pricing "aberrations" may persist for a very long time.Mike

QuoteOriginally posted by: Colossus2420What do I want to do...okay, I want to run a comparison between different indices, looking for hedging opportunities or ways to make alpha on the spread between two (or more indices). For example: if I take returns data for all 10 US sectors (industrials, energy, financials, etc.), and take the top performer over say 10y (call this #1) and then the 2nd worst performer (call it #9), the spread between the two is the money I can make if I'm long 1 and short 9 (1 and 9 are the best pair, 1 and 10 are not, incidentally). So how do I go about manipulating the data to smooth out the garbage and get a proper signal...or do I even need to (according to writers in Stocks & Commodities mag, not detrending and smoothing data leaves you wide open for failure and wrong conclusions).Now, I realize that may be a HUGELY juvenile project quest; there aren't any GARCH or stochastics or differential equations involved...or are there?... I'm not looking at swaps or options or hedging in the traditional sense with bankruptable levels of leverage (we hedge with ETFs), but I am interested in currencies and futures.I guess the bottom line is, if someone is relatively new to the sport, how would you go about acquiring the training and knowledge in order to be a competitive quant (aside from a quality degree program somewhere)? Once I get a solid foundation under me, THEN I can sound like I know what I'm doing. At least kinda.-Dr. JThe usual approach to becoming a quant is to study Stochastic Calculus and Option Pricing Theory.You seem to want to do statistical arbitrage. This is completely different stuff. The Pairs trading book recommended by Papercut is quite good -- it will give you ideas to try right away on your data.Maybe you should tell us what your goals are --- are you looking to trade profitably on your own or are you seeking employment ?