Search results

I'm in.

http://jmlr.csail.mit.edu/papers/volume3/kalai02a/kalai02a.pdf

Hi dillshau, you could you use a map from order_id to price and another map from price to another map/object (a container that represents a book_level). So on order arrival you use the price to book_level map to push that order to its corresponding level, and on order cancel you use the...

Andy, please change mine's to Alejandro Canete, Thanks!

Depends on your definition of High Frequency, if your alpha comes from having bounded latency at the micro second level, Java is not a good decision (because of the non deterministic garbage collection). If your latency have to be bounded at the millisecond(s) level, after some tweaking of the...

Not sure if it helps but I have a http://www.system76.com/product_info.php?cPath=28&products_id=106 , it works perfectly and so far the system76 guys have been very responsive.

Ohh, sorry, cant help you with that ;)

http://en.wikipedia.org/wiki/Multivariate_kernel_density_estimation

From an actual disclaimer of a leveraged ETF: "Leveraged ETFs designed to achieve their investment objective on a daily basis are not designed to, and will not necessarily, track the underlying index or benchmarkover a longer period of time." So you dont get the exact 2x,3x leverage if you are...

Hi Aerial, Machine Learning (kernel methods approach, don't get near the ANN/GP stuff), a big data course (map reduce, grid computing, etc), a computer architecture course (to understand the different factors that make up latency), concurrency programming, some course on design patterns...

Hi, I use HPC on a daily basis, mainly for backtesting and model calibration using level 1 and full depth data.

we just trade them on small test accounts.

I took professor's Liew independent workshop, and I also sat at the final presentation that Alex and David are talking about (great job! btw). I also had access to both lecture notes, home works and final presentations, and I must say that they are pretty much the same. Of course, for obvious...

If you prefer shorter and problem specific books you should take a look at the "Use R!" series by springer http://www.springer.com/series/6991?detailsPage=titles these are domain specific books with, usually, no more than 300 pages each. I read "Analysis of integrated and co integrated time...

I second that if you want to minimize developing time you should go with python or R, if you still stick to Java just make sure you use BigDecimal instead of Double.

Baruch (part time) for me.

Hi MaciejT. I used to rent a studio two blocks from Baruch ( app. 3k). Now I live in Weehawken (NJ) it takes me 30 minutes to get to work (Tribeca) and 45 min to Baruch. (app 1.5k). Hope it helps.

When you use heavy prior knowledge (syntactic model + a word ontology) to solve a problem instead of only data. For example if you want to do machine translation between spanish and english you can do it by either A) Training an optimal n-gram bayesian posterior discriminator using a paired...

Hi dddddd, NLP/Comp Ling is computationally intensive so you have to learn to use clusters and work with 100 gigs data files on a daily basis (thats useful in the HFT/stat arb space) and depending on the approach that you follow (i.e. probabilistic vs machine learning vs structural) you might...

You can also use python (reading from a csv or from the stdIO) http://matplotlib.sourceforge.net/

Hi dddddd, You have to set aside between 2k and 20k a month (depending on the asset class, broker and number of instruments you are going to trade) for colo and data, you can access and manage your algo from any place you want (your algo is going to run from the colo and thus is going to be...

Hi Alexei, It would depend on what portfolio and what do you understand by "future" and by "prediction". Exact return prediction of a pure long portfolio 20 years from now is impossible. But a dollar neutral long/short portfolio can have a predictable conditional return. Take a look at...

Hi M Souto Why do you need to create an executable? Do you want more performance? or do you want to hide the code when delivering it? Thanks

I read it. good intro. High level.

Logic programming My two cents... Although not a language per se. Some of te rule specification scripts on the CEP space out there use prolog "like" syntax. Which is not functional programming, but is not imperative either! rule LongBBANDUSDJPY salience -500 no-loop when...

The same exact formula that andy posted. But you can add an extra filtering/smoothing/averaging method to the CLOSE. Moving average - Wikipedia, the free encyclopedia StockCharts.com - ChartSchool - Stochastic Oscillator (Fast, Slow, and Full)

Hi Actually Stochastic oscillator measures momentum only under mean reversion assumption. It is just a rescaled view of past price trajectory. If you add an SMA in to the mix you are going to be adding lag to your signals. If you are building a mean reversion system use Bollinger bands, if...