This is my last post (maybe not if it's going to be deleted like my last post), in which case, the post prior to my last post was my last post.I actually did finish the book on algebraic geometry that I started two years ago, but like many math books these days, it will remain unpublished because it contains intellectual property that must remain private. Hey, HF trading pays my bills.It's clearly time to go.Best wishes, Newton

Newton, I once understood you to say that a covariance matrix cannot be symmetric, because a later data point cannot be used to predict an earlier data point. Do you simply mean that a missing data point cannot be used to predict a known data point? Or can a known data point, reflecting the price at which a transaction took place at time B, not be used to predict the unknown price at which an earlier transaction took place at time A? Or is it something only a matrix person can understand?

I've got a great idea. Why don't you submit your book for publication, get it accepted, and then tell them, sorry no, I don't want to publish? And then please scan and post here the letter of acceptance from the publisher. Then we will start to take you seriously. And you get to keep your intellectual property. Make sure that there are chapters on1. non symmetric correlation matrices2. Banach spaces, and why they are not complete3. why it is such a good thing not to define things when you talk about them.

1. non symmetric correlation matrices2. Banach spaces, and why they are not complete3. why it is such a good thing not to define things when you talk about them.Graeme,I was thinking of submitting a couple of manuscripts to AMS. The issues you mention are old news, but using hypergeometric analysis for a proofof Riemann and a second proof of Fermat sounds like fun. The subtitle I picked was "Is 'i' enough for a full complex analysis".kr should be getting the idea.Thanks for all your suggestions,Newton

And G,Ever wonder why correlation matrices created from actual data are always one rotation away from being symmetric??

I think the forum is less active for the last few weeks.Hope you are just gone for holiday.

And G,The study of Metric Spaces is still a very active research area in Set Theory and Analysis. Their existence and properties is no slam dunk as most in measure theory would have you believe.Here's the scoop: All metric spaces are *incomplete* *complex* spaces --- with the exception of Hilbert Space (a metric space with an inner product), which is the only complete, real metric space. It's the only space where a^2 +b^2 = c^2, a very special condition.So Banach Space, as some (re) defined it (a complete metric space), is exactly Hilbert space, inner product and all... no more no less. So is the measure theory worth anything. I'd be very careful.

And G,Doesn't it bother you that stochastic calculus can't explain fat tails or smiles? Or that every trader knows Black-Scholes is wrong? Didn't you ever get that feeling that there is much more to the concept of "stationary" processes than what appears in Hamilton? What if Brownian filtrations can't exist in any realizable system (like inertial navigation systems, the stock market, or any other system you can think of)?Some food for thought. Bye.