1,213 reputation
719
bio website
location Paris
age
visits member for 3 years, 7 months
seen Aug 26 at 10:39

Feb
8
comment What are the ensemble techniques to forecast returns?
Should I edit the question to make it about consensus forecast?
Feb
8
comment What are the ensemble techniques to forecast returns?
Thanks for the clarification.
Feb
8
asked What are the ensemble techniques to forecast returns?
Feb
7
awarded  Beta
Feb
7
awarded  Nice Question
Feb
7
accepted What are the popular methodologies to minimize data snooping?
Feb
4
awarded  Scholar
Feb
4
asked Is there a way to estimate (predict) the half life of a quantitative trading system?
Feb
4
comment What are the key risks to the quantitative strategy development process?
This is a very good moto I think. Do you know any documentation on this?
Feb
4
answered How useful is Markov chain Monte Carlo for quantitative finance?
Feb
3
comment What are the key risks to the quantitative strategy development process?
The half life is in fact a big problem to me. Is there any way to estimate it? (even roughly)
Feb
3
comment What are the popular methodologies to minimize data snooping?
Yes I maybe wrong but I view these problems as close but yet different. Well if we try to find the best parameters (with maximum likehood, MSE, MAP) and actually fit the predicted time serie to noise instead of real "structure" or "pattern", we have overfitted. In the contest of backtesting I see the data snooping as the bias we introduce when looking for different rules untill we find a working one.
Feb
3
awarded  Teacher
Feb
3
comment What are the popular methodologies to minimize data snooping?
I think that even if one can completly cancel the data snooping bias that wouldn't cancel the variance between backtested and true results. This is due to non stationarity and we cannot do anything about that. Ok for the comments
Feb
3
answered What are the popular methodologies to minimize data snooping?
Feb
3
awarded  Supporter
Feb
3
comment What are the popular methodologies to minimize data snooping?
I'm not sure what you mean by "tweak the parameters". What I do is look at the robustness of the strategy by changing the parameters arround my prefered values. The problem with this approach is that it is difficult to build a proper quatitative test to reject or not the strategies robustness. What is the size of the radius of the hypersphere arround the prefered value one has to look at? What performance variation should we tolerate? At the end it is almost just how you feel about it. I believe that to be dangerous in quant trading.
Feb
3
awarded  Student
Feb
3
asked What are the popular methodologies to minimize data snooping?
Feb
2
answered Is there any theoretical basis for pattern-recognition strategies?