def get_stats_section(self, offset, count):
"""Creates a multi-line string with several columns of stats about the
universe for a given time period.
Args:
offset: Number of rows (days) back to go when calculating stats.
This must be at least 4 so that there are 2 periods to compare.
count: Number of values to include for volatility and volume.
"""
period_end = self._daily['adj_close'].shape[0]
period_start = period_end - offset
period_midpoint = period_end - int(np.around(offset * .5))
# Prices with the most recent value at a max or min for the period.
price_at_high_cols = self._daily['adj_close'].iloc[
period_start:, :].max(axis=0) == self._daily['adj_close'].iloc[
-1, :]
price_at_high = 'At High\n' + text_utils.get_column(
self._daily['adj_close'].ix[-1, price_at_high_cols], 2)
price_at_low_cols = self._daily['adj_close'].iloc[
period_start:, :].min(axis=0) == self._daily['adj_close'].iloc[
-1, :]
price_at_low = 'At Low\n' + text_utils.get_column(
self._daily['adj_close'].ix[-1, price_at_low_cols], 2)
# Change in price stdev from the first to second half of the period.
# Ranges are inclusive because we care about differences across days.
first_range = range(period_start, period_midpoint)
second_range = range(period_midpoint - 1, period_end)
volatility_change = self._daily['adj_close'].iloc[first_range, :].std(
axis=0) / (self._daily['adj_close'].iloc[second_range, :].std(
axis=0))
volatility_change = 'Volatility Chg\n' + text_utils.get_column(
volatility_change.sort_values(ascending=False)[range(count)], 2, (
True))
# Change in volume from the first to second half of the period.
# Ranges are not inclusive since we are summing volume for each day.
first_range = range(period_start, period_midpoint)
second_range = range(period_midpoint, period_end)
volume_change = self._daily['volume'].iloc[first_range, :].sum(
axis=0) / (self._daily['volume'].iloc[second_range, :].sum(axis=0))
volume_change = 'Volume Chg\n' + text_utils.get_column(
volume_change.sort_values(ascending=False)[range(count)], 2, True)
return text_utils.join_lines([price_at_high, price_at_low, (
volatility_change), volume_change], ' ')
def get_returns_section(self, offset, bins=None):
"""Creates a multi-line string containing a column of top winners and
losers on the left, and a histogram of all returns on the right.
Args:
offset: Number of rows (days) back to go when calculating returns.
bins: List of boundaries between histogram bins in ascending order.
"""
returns = ((self._daily['adj_close'].iloc[-1, :] - (
self._daily['adj_close'].iloc[-(offset + 1), :])) / (
self._daily['adj_close'].iloc[-(offset + 1), :])).sort_values()
# If no bins provided, create default of 20 equally spaced bins.
if bins is None:
bins = np.arange(returns.min() - sys.float_info.epsilon,
returns.max() + sys.float_info.epsilon,
.05 * (returns.max() - returns.min()))
else:
# Possibly add bins at either end to ensure all values are counted.
if returns.min() < bins.min():
bins = np.insert(bins, 0, returns.min() - (
sys.float_info.epsilon))
if returns.max() > bins.max():
bins = np.append(bins, returns.max() + sys.float_info.epsilon)
returns_hist = text_utils.get_histogram(returns, bins, 0, True)
# If there is space for winners and losers and at least one empty line
# separating them, then include them.
extreme_count = int(np.floor((bins.size - 1) * .5)) - 1
if extreme_count > 0:
returns_col = text_utils.get_column(returns[:extreme_count], 1, (
True))
returns_col += ''.join('\n' * (bins.size - 2 * extreme_count - (
1)))
returns_col += text_utils.get_column(returns[-extreme_count:], 1, (
True))
else:
returns_col = ''
return text_utils.join_lines([returns_col, returns_hist], ' ')
