Dear participantsI would like to work for writing an paper in the field of Arfima -Figarch field.If any one of you have any idea in that field then I would like to work.

I am not sure I understand what you are seeking.Are you searching for co-writers? papers on arfima-figarch modeling (seems super restricted to me)? ..?

Actually I am doing PhD from university of Rome italy and I am Modeling exchange rate volatility using Arfima Figarch model.My supervisor told me that the work that i am doing might not be a publication. Therefore, i am trying to write in way that itwould become a publication and I don't have any Idea about how should I proceed. My prof, have idea But its complicated for meits kind of Bayesian stuff, which I am not interested in. Therefore, I am looking for something that is related with arfima-figarch or related field. I could become co writer also.reg

If you mean you want to publish a paper in the finance field (JF, JFE, RFS), rather than some engineering field, I don't feel this is a good direction to go. There are too many GARCH type of models and still they don't even get the correct sign.

Dear Culverin, Which direction should I go as I have background in mathematics and studied mathematical finance courses and some econometrics courses.If you have any Idea we could go on.reg

I think there are some notable failures in the finance theory:1. If you regress return on vol, the sign is frequently zero or negative. In my mind, this is not due to the fact that we need to build more complicated model, but because we need to distinguish between expected return and unexpected return. This is already noticed by some finance people.2. The entire affine class of models don't quite work. For failure for treasure bond, see "Term premia and interest rate forecasts in affine models". It also doesn't work for equity. But even worse, quadratic models still don't work (there are some papers on it).3 People intentionally ignore discounting. Prices are expected discounted cashflow. What about loss? Expected losses are generally computed without discounting. But what is discounting? How do people evaluate? This is also related to the tail risk.4 How to model correlation? A recent paper is "Systemic Sovereign Credit Risk: Lessons from the U.S. and Europe". But I don't feel any macro model have good chance to be successful, including Moody's GCorr.If you have tight budget, just mess around and apply fancy method without touching the core of problem. If you have several years of time, think about finance, not just math. You can read recent papers on AER, JPE, Econometrica, JF, JFE, RFS to get some idea about how academic finance people ask and answer questions. ps.I just notice you are in PhD program. What you want to do is typically called optimistically festival approach or pessimistically kitchen sink work: putting anything you have together. But why should you do this? Is the result going to be surprising? As far as a know, becoming a successful PhD on the job market is hard, the bar is higher than publishing a paper on top journal (ignoring the case that you can be a fourth author). Think about how many papers each year and how many star candidates (or how many top schools recruit).

QuoteOriginally posted by: outrunIt might be good to focus on empirical volatility statistics and distributions. That is observed reality and provides interesting model-free information. Intereting statistics are:* moments (stdev, skew, curtosis)* percentiles* range-scale analysis (autocorrelation, persistence, fractal scaling, ...)* barrier hitting time distributions* state models, eg regime switchingOnce you have good information about the empirical behavior then it's nice to simulate various GARCH and other volatility models and get the same distributions of the simulated paths.Finally compare the fit.The key issue is to not focus on just GARCH and fitting the model parameters and checking that the model features exists in data, but start with empirical behaviour first.