Building Synthetic Prices and Factors for Alphalens Analysis
Notebook Alphalens
Summary
This notebook constructs artificial price and factor data to demonstrate the input structure expected by Alphalens and to provide a controlled setting for factor analysis. It creates daily prices for six assets with different deterministic paths, assigns factor values on selected dates with missing observations, and maps assets into groups. It then expands the prices into intraday observations, aligning factor timestamps to the open.
The factor data is cleaned and paired with forward returns over several horizons, after which Alphalens tear sheets examine returns, information, turnover, and event behavior. The notebook repeats the analysis with long-short portfolios and with group neutrality enabled, illustrating how evaluation settings and group treatment affect the reports. Because both prices and factor values are synthetic and intentionally patterned, the output is useful for learning workflow and input conventions, not for inferring real market performance or validating a tradable signal. The example also omits realistic costs and market frictions.
Key ideas
- Synthetic price and factor tables can demonstrate the data format required for Alphalens.
- Intraday timestamps can be constructed from daily prices to study different forward-return periods.
- Missing factor observations and asset group assignments are included in the example dataset.
- Alphalens tear sheets evaluate factor returns, information, turnover, and event behavior.
- Long-short and group-neutral settings provide alternative views of the same synthetic factor data.
Tags
Full text
# Synthetic data examples
# Synthetic data examples
In this Notebook we will build synthetic data suitable to Alphalens analysis. This is useful to understand how Alphalens expects the input to be formatted and also it is a good testing environment to experiment with Alphalens.
```python
import matplotlib.pyplot as plt
import pandas as pd
from numpy import nan
from pandas import (DataFrame, date_range)
from alphalens.tears import (create_returns_tear_sheet,
create_information_tear_sheet,
create_turnover_tear_sheet,
create_summary_tear_sheet,
create_full_tear_sheet,
create_event_returns_tear_sheet,
create_event_study_tear_sheet)
from alphalens.utils import get_clean_factor_and_forward_returns
```
```python
#
# build price
#
price_index = date_range(start='2015-1-10', end='2015-2-28')
price_index.name = 'date'
tickers = ['A', 'B', 'C', 'D', 'E', 'F']
data = [[1.0025**i, 1.005**i, 1.00**i, 0.995**i, 1.005**i, 1.00**i]
for i in range(1, 51)]
base_prices = DataFrame(index=price_index, columns=tickers, data=data)
#
# build factor
#
factor_index = date_range(start='2015-1-15', end='2015-2-13')
factor_index.name = 'date'
factor = DataFrame(index=factor_index, columns=tickers,
data=[[3, 4, 2, 1, nan, nan], [3, nan, nan, 1, 4, 2],
[3, 4, 2, 1, nan, nan], [3, 4, 2, 1, nan, nan],
[3, 4, 2, 1, nan, nan], [3, 4, 2, 1, nan, nan],
[3, nan, nan, 1, 4, 2], [3, nan, nan, 1, 4, 2],
[3, 4, 2, 1, nan, nan], [3, 4, 2, 1, nan, nan],
[3, nan, nan, 1, 4, 2], [3, nan, nan, 1, 4, 2],
[3, nan, nan, 1, 4, 2], [3, nan, nan, 1, 4, 2],
[3, nan, nan, 1, 4, 2], [3, nan, nan, 1, 4, 2],
[3, nan, nan, 1, 4, 2], [3, nan, nan, 1, 4, 2],
[3, nan, nan, 1, 4, 2], [3, nan, nan, 1, 4, 2],
[3, 4, 2, 1, nan, nan], [3, 4, 2, 1, nan, nan],
[3, 4, 2, 1, nan, nan], [3, 4, 2, 1, nan, nan],
[3, 4, 2, 1, nan, nan], [3, 4, 2, 1, nan, nan],
[3, 4, 2, 1, nan, nan], [3, 4, 2, 1, nan, nan],
[3, nan, nan, 1, 4, 2], [3, nan, nan, 1, 4, 2]])
factor_groups = {'A': 'Group1', 'B': 'Group2', 'C': 'Group1', 'D': 'Group2', 'E': 'Group1', 'F': 'Group2'}
```
```python
base_prices.plot()
plt.show()
```
```python
base_prices.head()
```
```python
# create artificial intraday prices
today_open = base_prices.copy()
today_open.index += pd.Timedelta('9h30m')
# every day, after 1 hour from open all stocks increase by 0.1%
today_open_1h = today_open.copy()
today_open_1h.index += pd.Timedelta('1h')
today_open_1h += today_open_1h*0.001
# every day, after 3 hours from open all stocks decrease by 0.2%
today_open_3h = today_open.copy()
today_open_3h.index += pd.Timedelta('3h')
today_open_3h -= today_open_3h*0.002
# prices DataFrame will contain all intraday prices
prices = pd.concat([today_open, today_open_1h, today_open_3h]).sort_index()
```
```python
prices.head(10)
```
```python
prices.plot()
plt.show()
```
```python
# Align factor to open price
factor.index += pd.Timedelta('9h30m')
factor = factor.stack()
factor.index = factor.index.set_names(['date', 'asset'])
```
```python
factor.head(10)
```
```python
# Period 1: today open to open + 1 hour
# Period 2: today open to open + 3 hours
# Period 3: today open to next day open
# Period 6: today open to 2 days open
factor_data = get_clean_factor_and_forward_returns(
factor,
prices,
groupby=factor_groups,
quantiles=4,
periods=(1, 2, 3, 6),
filter_zscore=None)
```
```python
factor_data.head(10)
```
```python
create_full_tear_sheet(factor_data, long_short=False, group_neutral=False, by_group=False)
create_event_returns_tear_sheet(factor_data, prices, avgretplot=(3, 11),
long_short=False, group_neutral=False, by_group=False)
plt.show()
```
```python
create_full_tear_sheet(factor_data, long_short=True, group_neutral=False, by_group=True)
create_event_returns_tear_sheet(factor_data, prices, avgretplot=(3, 11),
long_short=True, group_neutral=False, by_group=True)
plt.show()
```
```python
create_full_tear_sheet(factor_data, long_short=True, group_neutral=True, by_group=True)
create_event_returns_tear_sheet(factor_data, prices, avgretplot=(3, 11),
long_short=True, group_neutral=True, by_group=True)
plt.show()
```
Shown in full with attribution under the source's licence. Licence: Apache-2.0
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.