It depends how you hedge/report your risk. If you hedge with a Futures and a zero-coupon, then the zero-coupon absorbs some of the theta...

<t>I don't have a solution yet, but let me rephrase so that we agree on the problem... and have a try.You have a bunch of vector (r_i, theta_i := cos ( 2pi/N sigma(i)), where sigma is a permutation on [[1, N]], and you want to minimise the norm of the sum of the vectors, i.e.min_sigma sqrt{ sum_i ri...

<t>Well the problem does not make too much sense. It is clearly mis-identified as eventually your '2' or '18' distributions collapse in a single 1... the sample distribution. Now you can split, arrange it how you want, it does not make a difference. In fine, without loss of generality and for parsim...

I had this nasty question on phone interview... I tried a few things, direct integration/ polar coordinate, transforms, moment generating function. As far as I remember it wouldn't given anything nice or better that the first line after few transformations.Anyone who has a nice CF plz post

<t>Hi,I suggest you take a look at the boost bind library. Basically, you still write your solver code on some function pointer, with no argument since it's not relevant to solving.Solver: x s.t. g(x) = 0Now, suppose you have a function g(x,y,z,a,b,c) and you want to solve x s.t. g(x,y,z,a,b,c) = 0,...

<t>Let me use the martingale method which I find less error prone.a > 0 and b > 0Note that:exp( - 2m Xt ) is a martingale. Indeed, this is exp(-2m^2 t + 2m B_t ) = exp ( - 1/2 ( (2m)^2 ) + (2m) B_t ) = exp ( -1/2 lambda^2 + lambda B_t) and we recognize a Doleans-Dade martingale.Let M_t := (exp ( -2 ...

<t>xsg, I am not sure your formula is correctU := X-YV := 2X + YYou can write U/ sqrt(V(U)) = rho V / sqrt(V(V)) + (1- rho^2) Wwhere, U/ sqrt(V(U)), V / sqrt(V(V)) and W are N(0,1) and V independent from WSo E(U / V ) = rho sqrt(V(U))/ sqrt(V(V)) V, since E W = 0= Cov(U,V) sqrt(V(V)) / sqrt(V(U)) V ...

<t>This problem belongs to the class of controled markov chain, but as pointed out there is a smart way to solve it.First, the intuition: Since p > 1/2, we are trending up. So we want to minimize the variance of the output by minimizing the size of the bet, thus betting the minimal bet 1.Second, som...

Actually, you can download some lecture notes from Tankov's web page. Only drawback for you maybe, is that they are written in French

<t>You can start with the following asymptotic approximation from Berestycki et al. (2000), which gives a tractable link between the 2 notions.where, sigma are parametrized in y = log(S/K) + rt. (K is the strike for implied vol and the current level of spot for local vol).If you don't go to far out ...

You just need to introduce a forward contract on the average itself, and you still have:Call - Put = ForwardThe value of the Forward contract is easily computed due to linearity.

<t>How much data do you use for your estimation? If the mean-reversion speed parameter is negative, this may come from the inclusion of different regimes in your estimation, that is to say that your model may be suited for given periods of time but is unable to account for regime changes. As shown i...

<t>Let sigma the permutation. ConsiderS = { sigma^k ( x_0), k) where x_0 is the starting string and ^k design the composition.S is a subgroup of G the group of permutation of x_0..Then immediate application of Lagrange theorem gives card(S) divides card of G in particular, card(S) <= card(G) = n! </t>

sevvost, a circle is a 2d-sphere...

At entry level, I'd better work with someone that has already studied a lot than someone slightly smarter but that hasn't 'done its homework'.