Past Meetup

PyDataNYC May 2015 Meetup

This Meetup is past

2 people went

Location visible to members

Details

Registration is closed for this event. We will be posting videos of the event afterwards.

AGENDA:

• 6pm: Doors open, pizza/drinks served, courtesy of Bloomberg

• 6:30pm: Speakers

• Aron Ahmadia: Numba: A simple tool for achieving performance in Python.

• Luis Sanchez: DATA Investing: Using Python and Machine Learning to gain insights into the dynamics of stock price returns around earnings announcements.

• 5-minute lightning talks by attendees (email Jason Grout, [masked], with a title and abstract proposal if you'd like to give one)

• 8pm: Reception - a great time to meet fellow PyDataNYC members!

• 9pm: End of meetup

Abstracts:

Luis Sanchez: DATA Investing: Using Python and Machine Learning to gain insights into the dynamics of stock price returns around earnings announcements. With the relatively recent announcement of hedge funds such as Bridgewater Associates, Point 72, and others building teams of Quants/Data Scientists, there is a sudden interest in artificial intelligence for trading applications. Earnings divergence from analysts expectations presents interesting trading opportunities in the days around announcements. These opportunities cover time frames from weeks, days and even minutes post announcement, but there are also some interesting signals pre announcement. These opportunities are generated by insider-information 'leakage' pre announcement, market overreaction & persistence of returns post announcement, expectation management by companies, etc. We will show how using python and public data we have been able to identify these patterns and how they differ by industry, size of earnings miss, and market capitalization. We will also examine how analyst coverage impacts both earnings forecast accuracy and return reaction following announcements, how these patterns have evolved over time, and a proprietary ranking of accuracy of earning estimates by financial institutions. Finally, we will show the speaker's real-life experience in applying hybrid DS/Quant model in option trading, and preview a system that data mines many sources of public data in real-time to try to beat Wall Street revenue estimates on some stocks. Luis M Sanchez, Founder, CEO and Data Scientist at Ttwick. Luis has over 20 years of experience in capital markets, insurance, consulting, and engineering, with emphasis on financial quantitative analysis. Luis has held multiple senior executive positions and quantitative analyst roles for Barclays Capital, Lehman Brothers, Deutsche Bank, AIG, and a couple of macro hedge funds, where he structured over 10Bn USD in exotic ABS deals. He started his career in finance in 1993 as Director of Quantitative Analysis for a hedge fund, where he designed and programmed several machine learning systems to trade equities and arbitrage OTC exotic options. Luis obtained his MBA in 1992 on a Fulbright LASPAU scholarship, and a BSc in Civil Engineering in 1987. Luis started coding Fortran and Basic in 1982, and now he is a python scientific stack programmer. Eric Giambattista, Data Scientist at Ttwick: Eric is in his final year in the Economics Ph.D. program at New York University, where he is researching macroeconomics and international economics. He graduated summa cum laude and Phi Beta Kappa from the University of California, San Diego with a BA in Economics and History. Before joining Ttwick he worked in the Research Department of the Federal Reserve Bank of Boston, where he focused on forecasting and finance. At Ttwick, he is building a proprietary library of prediction models that is being deployed on multiple datasets collected by Ttwick's systems. Eric is a python scientific stack programmer.

Aron Ahmadia: Numba: A simple tool for achieving performance in Python. Numba is an Open Source NumPy-aware optimizing compiler for Python. It uses the LLVM compiler infrastructure to compile Python syntax to machine code for both x86 architectures and CUDA GPUs. In this talk, I'll provide some motivation for why tools like Numba exist, show how to use Numba to optimize Python code "just-in-time", and compare Numba with other options in the space such as PyPy and Cython. I'll also provide guidance on when Numba is the most helpful, expectations on its limits, and some other general performance tips when working with Python and NumPy.