7,663 reputation
826
bio website bachelierfinance.org
location United States
age 45
visits member for 3 years, 2 months
seen 21 hours ago

Worked in a lot of quant areas


Sep
3
answered Risk-neutral models for rights issues
Sep
3
comment What is the difference between the methods (listed in content) in pricing convertible bond?
If you separate the bond and option components then you cannot deal properly with the fact that default affects both. When I need to explain it to traders who don't understand the math, I generally express it in terms of discount rates for the option component, which in theory would need to vary with the probability of both exercise and default as the underlying moves. As a matter of fact, primitive approaches to converts pricing in the 1980s and early 1990s did just that.
Aug
20
answered how to extend lognormal model so that $\sigma$ is correlated to $\mu$?
Aug
20
answered What is the difference between the methods (listed in content) in pricing convertible bond?
Aug
14
answered relation between asset's and equity volatilities - merton model
Aug
12
comment R: Fast and efficient way of running a multivariate regression across a (really) large panel (First pass of Fama MacBeth)
Ah, by "appending datasets" you mean including both old and new data in the regression. Updating using new data items without discarding or recomputing old data items is the point of biglm. As I mention, the other requirement you have -- discarding of older, windowed-out data -- is a feature biglm lacks so you will have to add it. The math for that is in Huber's post, and the programming for it should be a fairly trivial extension to the existing source code for biglm.
Aug
12
comment R: Fast and efficient way of running a multivariate regression across a (really) large panel (First pass of Fama MacBeth)
I'm pretty sure biglm is not appending datasets. It sounds like you have a bug.
Aug
9
answered R: Fast and efficient way of running a multivariate regression across a (really) large panel (First pass of Fama MacBeth)
Aug
7
comment Is there a strong solution to $\frac{dS}{S}=\sigma(S)dw$?
One view would be that this the same as $\frac{dS}{S}=\lim_{n \to \infty}(\sigma_1+\sigma_2^{1/n})dW$ which would be hard to solve. However, it feels like some reflection principle solution may be possible. Also, I would review the literature on excursion times.
Aug
7
comment How to design back-testing (validation) for such modified Vasicek model?
I understand the question better and I think I fixed the answer
Aug
7
revised How to design back-testing (validation) for such modified Vasicek model?
added 180 characters in body
Aug
6
answered How to design back-testing (validation) for such modified Vasicek model?
Aug
5
answered Estimating early exercise boundary for American put
Aug
5
comment Basket option density in BS model
Doesn't look correct to me....
Aug
1
comment Backtesting - can you buy/sell at open and closing prices?
@MattWolf I ran the other direction :-) they had done well for awhile on the "better lucky than skillful" plan. I had certain...concerns...about its stability.
Jul
30
answered Backtesting - can you buy/sell at open and closing prices?
Jul
30
revised Why does Black-Scholes equation hold on continuation region of American Option?
Fix mistake in operator spec
Jul
30
comment options pricing using vwap
I've never personally perceived it as a problem in modern markets, but it was considered an important point in the aftermath of the '87 crash. Some papers argued the liquidity only came to be priced in properly after that crash. These days the only concerns I have seen have been on the part of the maniacs at ZeroHedge.
Jul
29
comment options pricing using vwap
It is also worth noting that cash-settled futures options make some people nervous about financial system stability, since they worry that these liquidity considerations are not properly "priced in". It was a topic of much discussion back in the 70s and 80s.
Jul
10
comment How can you convert the CUSIP of a bond issue to the CUSIP of the company's stock/
FactSet and Bloomberg both have this data