So the book mentions roughly 15 people at RenTech. What do the other 2,000 people do? Well, this is more of a rhetorical question. Through a couple of intermediaries, I have heard that only a small, "core" group works on trading strategies. The rest perform IT-ish "sanitation", e...

The following is quite dry but contains many results:

Lipton, A. (2001). Mathematical Methods for Foreign Exchange: A Financial Engineer's Approach.

The following is a classic: Brigo, D., & Mercurio, F. (2006). Interest Rate Models - Theory and Practice. The book is quite technical. So, depending on your background, you may need some preparatory reading. For natural prerequisites of the book as well as suggestions on how to push its ideas f...

Yes, the predictors are observable financials, which jump all the time. However, that allows only the volatility and jump intensity to jump, and only in certain specifications (as stated above). That does not mean that a jump in Y(t) is imminent, just more/less likely.

The entities are business units somewhere. The Y(t) process is the asset portfolio (of the entity), more or less... Some shocks are "scheduled" but not too many.

Actually, that's not too important. The actual decision making is about statistical methodology, not financial meaning. With the estimation procedure, there are so many opportunities to go astray. That's where the real perils are... Having that said, I would share the entities and predictors with yo...

Let X(t) = (X_1(t), …, X_p(t)) be a set of stochastic processes which act as predictors. We are concerned with estimating the following jump-diffusion                   dY(t) = A(X(t)) * dt + B(X(t)) * dW(t) + dC(t),   where                   A() and B() are some deterministic functions,            ...

@Cuchulainn I understand your constraints very well. I am in the same boat, in a way... "Gradient" is just a word in the name of the method (related to original inspirations). The whole method is much broader... More importantly, in the next 10 years the method will become even broader, th...

@Alan, the work was done for a client, so I cannot share it. But I will post a short summary of tried specifications within 24 hours... Still, for the end user my specifications may be too complex or too simple, depending on the data he/she has available. It all runs from the data.

This is fairly general but applies to model validation in particular: apply gradient boosting to model development. That is, use the current model as the starting point. Augment it with various other modeling ideas for those data points (instruments, assets) for which the current modeling iteration ...

None of the above. You should use 20-day returns calculated off the rolling 20-day window. The produced data will be correlated this way because neighboring windows will overlap by 18 days. This is not a problem though. Even for correlated observations barrier frequencies are unbiased estimates of t...

Unfortunately, what you are "concerned with" is less relevant. Painful as is. What matters is the features of the underlying dynamics that you will actually be able to estimate. Jump-diffusions are notoriously difficult to estimate. The estimates are unstable, from one entity to another, f...

The reason we keep going on about the folly of win-loss ratios is that (besides being dangerously wrong) it only works if all prop traders or trading systems have the same disribution of returns and that distribution is symmetric.  If the distribution is assymetric, the break-even win-loss ratio mi...

In an irrational hope to bring this discussion back to the theme of the thread, here's another resource on Spearman's rho. Mann-Kendall and Spearman's rho tests are used for trend detection.

https://rmgsc.cr.usgs.gov/outgoing/thre ... al2002.pdf

Traden4Alpha, trader's know-how is composed of at least 3 segments:  1) trading signals (detection of new prop opportunities),  2) execution strategy (entry / exit levels, maximum duration, stop loss rules if any, etc), 3) risk management. You have mentioned segment 2, where return magnitudes matter...