3,193 reputation
1724
bio website
location
age
visits member for 3 years, 8 months
seen Aug 14 at 16:43

finance PhD student


Aug
15
comment How many explanatory variables is too many?
@gsk3 -- You are right that you must test sub-samples, but the holdout sample won't necessarily catch multicollinearity. Using the ridiculous humidity example, because the humidity at Broadway and 34 is practically the same as the humidity at Broadway and Wall, there are a lot of linear combinations that sum to $\epsilon > 0$. In this example the multicollinearity would be obvious because $\beta_{midtown} \approx -1 \times \beta_{downtown}$, but if you're in the habit of having too many regressors, then it may not be easy for you to identify. You need to test adding & removing regressors.
Aug
15
comment Is there a quantitative finance ranking system for universities?
We will see how the community responds to your rephrasing, but this is likely a better question for advisor (you said you were a PhD student, correct?). Even if someone finds a ranking system, I would take it with a large grain of salt. Rankings for UG/MBA programs work because they have huge numbers of applicants and graduates (and graduates only care about USN&WR rankings), but when it comes to PhD and beyond, it is probably best to know the research and meet the researchers, then make a more subjective decision.
Aug
15
comment How many explanatory variables is too many?
@gsk3 -- And this wouldn't set the upper limit on the number of factors/regressors. I could add humidity downtown and humidity midtown as regressors in my model and "improve" its explanatory power, although these almost certainly have no impact on my model. Because these are collinear, I could get economically and statistically significant coefficients on these factors, even though a change in humidity has no impact on the market.
Aug
15
comment How many explanatory variables is too many?
@gsk3 -- Isn't everything in sample? I don't know tomorrow's data and I would use all (relevant) available data to calibrate my model. If my model is correct, then it should work in sub-samples, unless I think there are multiple regimes, but then it should work in sub-samples of the relevant regime.
Aug
13
answered Has high frequency trading (HFT) been a net benefit or cost to society?
Aug
13
answered How many explanatory variables is too many?
Aug
8
answered Indicators and research for stress-based investment strategies
Aug
4
comment How do I backtest a convertible bond arbitrage strategy in R/Matlab?
OK. I guess I consider that more a Greeks question than a backtesting question. Have you considered RQuantLib?
Aug
4
comment How do I backtest a convertible bond arbitrage strategy in R/Matlab?
Does it matter what security generates the price? I have only used the backtest in R, but any of these should let you generate portfolios along with a periodicity and condition for portfolio entry/exit.
Jul
27
awarded  Self-Learner
Jul
15
comment Market Data For Project
Have you tried past questions here? This is a weekly question. quant.stackexchange.com/search?q=data+source
Jul
1
revised Usage of NoSQL storage in Finance
Typos
Jun
25
revised Is Conditional Value-at-Risk (CVaR) coherent?
removed incorrect infimum part
Jun
25
comment Is Conditional Value-at-Risk (CVaR) coherent?
@chang -- Good catch! You're right. I read too quickly and misinterpreted discrete $R$ as $R$ in a finite set, which would require infimum.
Jun
24
answered Is Conditional Value-at-Risk (CVaR) coherent?
Jun
23
answered Control for bid/ask bounce in high-frequency trade data?
Jun
20
comment Vanilla European options: Monte carlo vs BS formula
The number of iterations is just a function of what size standard errors with which you are comfortable.
Jun
20
comment Vanilla European options: Monte carlo vs BS formula
You shouldn't be able to reject that the two solutions are different.
Jun
8
awarded  Organizer
Jun
8
revised How to estimate a multivariate GJR or TARCH model in Eviews?
Added eviews and time-series tags