Since the CME calendar was based off of the NYSE calendar, closings for
Hurricane Sandy were included, but shouldn't have been for CME.
This is the explanation we had included prior to the new exchange
calendars:
http://en.wikipedia.org/wiki/Hurricane_sandy
We *do not* add any non-trading days for Hurricane Sandy. Although
trading on CME Group exchanges was disrupted, trading still occurred at
some point on each day. Importantly, the disruptions do not appear to
have affected the expiration dates stipulated in the futures contracts
expiring during this period.
Trading of CME US equity index futures & NYMEX futures was disrupted due
to the hurricane's impact on New York. CME US equity index futures
markets were disrupted on October 29th 2012, but reopened at 5:00pm CT.
The US equity index markets were closed from 8:15am CT on October 30,
both for the trading floor and CME Globex. The trading floor of the
NYMEX was closed on October 30, but all floor-trading products were
available electronically on CME Globex.
Source:
http://investor.cmegroup.com/investor-relations/releasedetail.cfm?ReleaseID=716923
October 2012
Su Mo Tu We Th Fr Sa
1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30 31
Also see:
-- http://www.chicagobusiness.com/article/20121029/NEWS01/121029818/cme-cboe-other-markets-closed-for-hurricane-sandy-about-500-local-flights-scrapped
-- http://money.cnn.com/2012/10/29/investing/hurricane-sandy-stock-markets
Querying for the price field of an equity actually looks at the close
field, so we should do the same for futures. Otherwise `data.can_trade`
and `data.current` of 'price' fail for futures.
To support testing configurations which need control over the full path
to the asset, adjustment, and equity bcolz directories; which is
required by some of our internal testing which exercises servers which
coordinate these files via a date slug in the full path.
Also, by allowing customization of the full path, it is now possible to
have the AssetFinder and AdjustmentReader sqlite databases be written to
disk, which is also required for our server testing setup.
The tradingcalendar module has been replaced by the new exchange
calendars and trading schedules. Issues a ZiplineDeprecationWarning at
tradingcalendar module scope to be triggered on imports.
next_scheduled_day and previous_scheduled day should raise if
trying a return a date outside the calendar. Previously it just
returns None, but it should be made consistent with the behavior
of add_scheduled_days
- Refactored EventsLoader and BlazeEventsLoader to not require a
subclass per dataset. Instead, you now pass a map from columns to
event fields directly to the EventsLoader constructor.
- Removed a large number of Quantopian-specific datasets and associated
tests.
- Rewrote the core logic of EventsLoader and BlazeEventsLoader to share
index calculations across multiple requested columns.
- Fixed a bug where event fields were incorrectly forward-filled when
null values were present in an event.
Since the first trading day is now passed directly to the DataPortal on
init, there's no need for a method that does this. Moves all the
additional logic/assignments into the init. Also corrects an issue where
we would never create certain attributes if self._first_trading_day was
None.
Adds the ability to specify the first trading day for a data portal in a
test case when using the WithDataPortal fixture.
Fix behavior in minute mode history with frequency `1d`, where on the
day immediately following an adjustment action, the overnight adjustment
would not apply. (However the adjustment would be applied after a 1 day
lag.)
The root cause of the bug was that the history data for minute mode when
using `1d` stitches together a sliding window of the daily data for
previous and the current minute. That daily data sliding window and
corresponding adjustments was being read as if the data was being viewed
from on the last day of the window; however in this case the data is
being viewed from the day after the window has completed. The difference
in view points requires the adjustments to popped and applied by the
adjusted array one index earlier. The fix uses the `extra_slot` value as
signifier on whether the data is being viewed on the following day and
then accordingly adjusts the index of the mulitpy object.
Also, change the split and merger test data ratios to have different values,
to ensure that different adjustment values are applied; as opposed to
doubling up on just one of the values.
ENH: fast stochastic oscillator added.
A fast stochastic oscillator has been added to the technical
factors. This is the simplest of the stochastic oscillators,
and can be used to build the others.
Tests have been added that compare against the values expected
from that of ta-lib STOCHF.
FastStochasticOscillator is marked as window_safe=True to allow taking
moving averages for smoothing.
Andrew Daniels