This could perhaps be labelled BUG, as well.
Change the Sharpe (and algorithm volatiilty) value used to compare
algorithms/backtests so that it is annualized and uses daily returns.
Previously, the Sharpe metric was using the same calculation style
as the fixed size periods, i.e. 3 Month, 6 Month, etc., which can
use the geometric mean when comparing against the risk free.
Change the Sharpe calculation to use the arithmetic mean differenc
against the risk free rate, using daily (non-compounded) values.
Also, use annualized mean returns.
Hold on to the values set for the returns frequency and the cont
index, mainly for debugging purposes mid-process.
It was useful to have these values when debugging why there was
an extra non-midnight time in the index.
The dump and update of curves were both using the entire history.
So instead of having the update use a different code path, always
use dump and overwrite.
Instead of using a raw np.array and keeping track of an index into
that array, use a pd.Series to track the last_sale_price and amounts
in a vector format.
Both unit tests and repeated runs while developing an algorithm
can benefit from having a local copy of the Yahoo data, instead
of doing a network call each time.
Store the web request results as a csv file in a cache directory,
named by symbol and date range.
Most of the functions in date_utils can be done via pandas.
The other functions are no longer used for loading, etc. so remove
the date_utils module to reduce the total surface area of Zipline core.
This utility was referring to functions that had been long since
removed in the loader module.
If the utility is still needed by some, it can be added back in,
but using the pandas read/write instead of msgpack.
Instead of writing our own serialization using msgpack, leverage
the csv serialization provided by pandas.
Also, lessens the need for msgpack and functions in date_utils.
Instead of midnight to midnight for each day, use the trading
environment's market open and close for each day, so that the index
is exactly the trading minutes of each day.
Reduces the amount of memory consumed, but more importantly should
make it easier to inspect the Series that use the index and check
whether the values are correctly being filled.
Remove more use of lists for storing internal risk values to use
pandas structures, for easier matching of time to value.
Accordingy, convert use of -1 for getting last value,
to use current dt.
Continue on path of converting values stored inside of risk metrics
to use a DataFrame instead of storing multiple lists.
Also, the need for latest_dt in getting the current volatility for
the sharpe calculation, shows that we need to set the lastest_dt at
the beginning of the update loop.
Indexes to risk answers were pointing to a previous version.
Also, provide the risk cumulative answers as a pd.Series,
so that it is easier to compare to values produced by risk class.
Merging branch "Minimum price variation"
Ensure limit prices conform to a minimum price variation of a penny, which works for most stocks (not, for instance, BRK). The rounding "midpoint" is custom and depends on buy/sell direction, instead of .5 of a penny.
to account for minimum price variation.
On an order to buy, between .05 below to .95 above a penny, use that penny.
On an order to sell, between .05 above to .95 below a penny, use that penny.
Eventually, all cumulative metrics, (alpha, beta, etc.) will be
stored in the same DataFrame
For easier tracking of dt to values during debugging, but should be
some performance gains as well.
So that it is easier to add new containers, factor out the creation
of the index.
Also, make the returns frequency a parameter, to make the use of
different frequencies more clear from within the risk metrics object,
rather than hot swapping in the new frequency type via the now
removed `initialize_daily_indices`.
The eigen_values, condition_number, algorithm_covariance, and
benchmark variance, which were easy to calculate alongside beta,
since they share the same inputs, but were not passed along to performance.
Remove to trim down the number of risk report members as well as
number of calcluations done.
Can add back in if there is an expressed need for eigen_values etc.,
perhaps in an 'opt-in' type configuration.
Eddie Hebert