<t>Exotiq, Thanks for the reply. What I meant by compound return is as follows:For each month, you get a maximum return of 5% and a minimum return of -10% on the capitalappreciation of a stock. The window where that return is measured is reset every month so basically it's like taking for P&L ev...

<t>Hello, Does anyone have any references or insights on how to price a structured product that has a payoff as follows:A) $0.2 quarterly +B) the compound monthly return on a stock where the return has a floor of -10% and a cap of 5%. The monthlyreturn is calculated as of the 20th of every month for...

<t>sm345:I will be using this book in an upcoming course. Frankly, I'm a little pissed about it becausethe prof is making us use it because the authors are his buddies. Does anyoneknow if either Cvitanic or Zapatero have much of a reputation? I would have ratherused something a little more establish...

<t>I thought you wanted a spreadsheet all coded up for you that's all. Ialso made a minor error in the equation. See below.Just write the equation down and have a look at it.On the LHS is a growth rate of yOn the RHS is the growth rate of x4, the lagged x1 to x3'sand the error correction term (y(t-1...

<t>That's getting to be a bit much. You should be able to code up the formula I gave you. Just square the residuals,sum them up, and then feed that to the solver. As for t-stats,you could use bootstrapping or I think that the t-stats fromthe 2 step procedure are asymptotically, normal but you'dhave ...

<t>Okay, now that I see what you're trying to do, I think I have some more usefulinput. I would estimate an error correction model as follows (to anyone else reading,if I make a mistake please let me know as it has been awhile since I've actually done this!): Assuming that x1 to x3 are all stationar...

<t>Aaron, I think that you're confusing in-sample fits with out of sample forecasts.In an efficient market, it is not the case that a forecast regression shouldnot produce any statistically significant results. It may simply be that youpicked the right sample for that variable and if you implemented...

<t>I'm pretty sure that a high R-squared and a low DW stat arethe classic sign that your estimates are inconsistent. Haveyou looked at your residuals over time? That's probably agood starting point. I would test them using the AugmentedDickey Fuller test or maybe Phillips Peron to see if they aresta...

See the following for a pretty good explanation:http://oll.temple.edu/economics/notes/c ... ration.HTM

<t>Are you familiar with stationarity? If not, you should do some readingon this before proceeding or just estimate your regression with thelog difference of all your data. Your independant variables are likelynon-stationary so I would use log differences across the board. If you regress:y=X*B+eif y...

<t>If you don't know cointegration, then I don't think you should berunning the regression you are. I would take the log differences orthe percent change in each price (it's the same thing) and run thatin your regression. If you type cointegration into google, you'll get back a ton of info.In a nuts...

<t>You also might want to check the residuals in your regression for stationarity. The regression you are running is assuming that the stock price and the spotprice of the commodity are cointegrated (Prices are usually I(1)). If they aren't, your estimates are inconsistent. ie R-squared isn't a vali...

<t>Would anyone have some code for matlab for constrainted optimization. I know I can fmincon but I was hoping to try out a new algorithm as analternative to mathworks' ones. I've heard Differential Evolution brought up a few times on this forum.If anyone has some functions that their willing to off...

I don't know the answer but I wanted to move the question back to thetop of the forum so it's not forgotten!! I find that more often than not,a question is ignored right at the point where some input would bemost insightful.r

<t>If I read your message right, you want to estimate the zero coupon term structure and use that as in input to your model? If so, I would say that it wouldn't be a good idea to use corporates in your sample because theyield curve you're going to estimate should hold credit constant throughoutthe t...