You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
"description": "Financial institutions regularly estimate possible portfolio losses\narising from (future) changes in the market values of their assets, e.g.\nusing the Value at Risk (VaR) metric. We show a VaR model using Filtered\nHistorical Simulation and illustrate some of the usual processes for\ncalibrating, validating and analysing these models, such as calibration\nand backtesting. Although the model is relatively simple, the challenges\nposed by their different requirements - flexibility, data volume,\nperformance - inevitably lead to many shortcuts and complexity. With the\nultimate goal of managing this complexity, we explore and evaluate, with\nconcrete coded examples, the PyData (e.g. Pandas, Bokeh) and Apache\nSpark frameworks from a practitioner's point of view.\n",
"duration": 2167,
"language": "eng",
"recorded": "2015-06-20",
"speakers": [
"Miguel Vaz"
],
"summary": "Using Bokeh, Pandas and a bit of Apache Spark to address the\nflexibility, performance / data volume requirements of standard\nfinancial risk management processes. Examples in the talk illustrate\nthe model calibration and backtesting processes of a market risk\nmodel (VaR Filtered Historical Simulation), which quantifies possible\nfuture losses arising from market price movements.",