Optimize session close lookups in MinuteResampleSessionBarReader:
- Adds `session_closes_in_range` method (along with
`session_opens_in_range`) to TradingCalendar to allow vectorized
retrieval of all values in a range of sessions.
- Improves code path for resampling a single session's worth of data (as
is the case when calling `get_value`), since we don't actually need to
look up the close minute.
When retrieving the open and close for a given session, we only care
about the scalar values, so using DataFrame.at instead of DataFrame.loc
is significantly faster.
The CFE was closed along with the NYSE in observation of the days of
mourning in honor of the passing of presidents Gerald Ford and Ronald
Reagan. The CFE also observed the closures due to Hurricane Sandy,
along with NYSE. Adding those adhoc holidays to exchange_calendar_cfe
and removing them from cfe.csv in tests. To fit with
USNationalDaysofMourning, also removing the closure in observation of
the day of mourning in honor of the passing of president Nixon in
1994, despite the fact that the exchange did not exist at that time.
Signed-off-by: Maxwell Rounds <maxwell.j.rounds@gmail.com>
Capital base is included in the sim params, so we should define the
value there, or use the default.
This change also unifies the default capital base as 1e5, as was
previously defined in algorithm.py.
Make a new decorator, expect_strictly_bounds, which shares logic
with the existing expect_bounded (which checks that inputs fall
INCLUSIVELY between bounds),
Fixes a bug where we'd fail to raise an error if the start/end of a
history window call don't aren't in the loader's calendar.
We were started dropping this error after a previous change swapped out
calls to `index.get_loc` with calls to `index.searchsorted` to avoid
creating hash tables in pandas.
From pep-0008:
```
Always use a def statement instead of an assignment statement that binds a
lambda expression directly to an identifier.
Yes:
def f(x): return 2*x
No:
f = lambda x: 2*x
The first form means that the name of the resulting function object is
specifically 'f' instead of the generic '<lambda>'. This is more useful for
tracebacks and string representations in general. The use of the assignment
statement eliminates the sole benefit a lambda expression can offer over an
explicit def statement (i.e. that it can be embedded inside a larger expression)
```
Any DataFrame that's had `.loc` or `.iloc `called on it participates in
a cycle, which means they're not immediately garbage collected when they
go out of scope. This matters for pipeline results because they consume
multiple megabytes per column, which means that a pipeline result with
many columns can hold take up over 100MB. By manually breaking
DataFrame cycles, we can ensure that we never hold multiple pipeline
results in memory at once.
Allows ``__funcname`` to be passed to preprocessors like expect_types
and expect_dtypes to override the name displayed in error messages.
This is useful for providing clearer errors for ``__init__`` and
``__new__`` methods in classes.
In days_at_time, use a Timedelta instead of a DateOffset. We were
previously using DateOffset to work around a pandas 16 bug even though
it raises a PerformanceWarning. Now that we're on pandas 18, we can use
the much simpler Timedelta construction.
In favor of a new method `set_restrictions` which takes a Restrictions
object. Calls to `set_do_not_order_list` should raise a deprecation
warning and create an equivalent Restrictions object, with which
`set_restrictions` will be called. For convenience, create a
RestrictionsSet from which the "restrictions" version of a security
list can be accessed
The SecurityList implements a non-exposed method
`current_securities(dt)` which SecurityListRestrictions calls to
determine if an asset is restricted. Deprecate the `__iter__` and
`__contains__` methods of security lists in favor of
`current_securities(dt)`
Pandas 0.18 doesn't like having null-ish values in categoricals. Fixing
this properly requires re-thinking the semantics for missing_value on
pipeline terms, so we're punting on that until after we've upgraded to
0.18.
In pandas 0.18, the behavior of ``nth()`` changed so that Grouper no
longer can be easily used to recover group labels.
Instead of using the built-in grouper behavior, we use a groupby on two
arrays we build ourselves. This recovers the original behavior, and is
about 2x faster as a bonus.
- Fixes a warning on indexing with a float that ultimately came from
pd.Timedelta.total_seconds(). Adds ``timedelta_to_integral_seconds``
and ``timedelta_to_integral_minutes()`` functions and replaces various
usages of ``int(delta.total_seconds())`` with them.
- Fixes a warnings triggered in ``_create_daily_stats`` from
passing tz-aware datetimes to np.datetime64.
One year NYSE test that buys a lot triggers 492,963 calls to
minute_to_session_label. Only 98924 ~(390 * 252) make it past the
cache and trigger the heavier computation.
Jean Bredeche