Advances in Financial Machine Learning by Marcos Lopez de Prado

Author:Marcos Lopez de Prado
Language: eng
Format: epub
ISBN: 9781119482109
Publisher: Wiley
Published: 2018-01-18T00:00:00+00:00

Your strategy produces 100,000 forecasts over time. You would like to derive the CPCV distribution of Sharpe ratios by generating 1,000 paths. What are the possible combinations of parameters (N, k) that will allow you to achieve that?

You discover a strategy that achieves a Sharpe ratio of 1.5 in a WF backtest. You write a paper explaining the theory that would justify such result, and submit it to an academic journal. The editor replies that one referee has requested you repeat your backtest using a CPCV method with N = 100 and k = 2, including your code and full datasets. You follow these instructions, and the mean Sharpe ratio is –1 with a standard deviation of 0.5. Furious, you do not reply, but instead withdraw your submission, and resubmit in a different journal of higher impact factor. After 6 months, your paper is accepted. You appease your conscience thinking that, if the discovery is false, it is the journal's fault for not having requested a CPCV test. You think, “It cannot be unethical, since it is permitted, and everybody does it.” What are the arguments, scientific or ethical, to justify your actions?

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.