From b61a5a31a54bb566ab5ded368543765ca9bbbb49 Mon Sep 17 00:00:00 2001 From: Eddie Hebert Date: Wed, 17 Oct 2012 15:22:58 -0400 Subject: [PATCH] Removes notes about data soures in README. This should live in the docs instead. --- README.md | 13 ------------- 1 file changed, 13 deletions(-) diff --git a/README.md b/README.md index 6a0e0abf..58af8928 100644 --- a/README.md +++ b/README.md @@ -51,19 +51,6 @@ Scientific python on the Mac can be a bit confusing because of the many independ - swig - you need this to build scipy. ```brew install swig``` - hdf5 - you need this to build tables. ```brew install hdf5``` - -Data Sources -============= -The Backtest can handle multiple concurrent data sources. zipline will start a subprocess to run each datasource, and merge all events from all sources into a single serial feed, ordered by date. - -Data sources have events with very different frequencies. For example, liquid stocks will trade many times per minute, while illiquid stocks may trade just once a day. In order to serialize events from all sources into a single feed, qbt loads events from all sources into memory, then sorts. The communication happens like this: - -1. zipline requests the next event from each data source, ignoring date (i.e. just next in sequence for all) -2. Using the earliest date from all the events from all sources, zipline then asks for "next after " from all sources. -3. All datasources send all events in their history from before , moving their internal pointer forward to the next unsent event. -4. zipline merges all events in memory -5. goto 1! - Style Guide ===========