Pandas TA (pandas_ta) Strategies for Custom Technical Analysis

Topics

What is a Pandas TA Strategy?
- Builtin Strategies: AllStrategy and CommonStrategy
- Creating Strategies
Watchlist Class
- Strategy Management and Execution
- NOTE: The watchlist module is independent of Pandas TA. To easily use it, copy it from your local pandas_ta installation directory into your project directory.
Indicator Composition/Chaining for more Complex Strategies
- Comprehensive Example: MACD and RSI Momo with BBANDS and SMAs 50 & 200 and Cumulative Log Returns

In [1]:

%matplotlib inline
import datetime as dt

import pandas as pd
import pandas_ta as ta
from alphaVantageAPI.alphavantage import AlphaVantage  # pip install alphaVantage-api

from watchlist import Watchlist # Is this failing? If so, copy it locally. See above.

print(f"\nPandas TA v{ta.version}\nTo install the Latest Version:\n$ pip install -U git+https://github.com/twopirllc/pandas-ta\n")
%pylab inline

Pandas TA v0.2.49b0
To install the Latest Version:
$ pip install -U git+https://github.com/twopirllc/pandas-ta

Populating the interactive namespace from numpy and matplotlib

What is a Pandas TA Strategy?

A Strategy is a simple way to name and group your favorite TA indicators. Technically, a Strategy is a simple Data Class to contain list of indicators and their parameters. Note: Strategy is experimental and subject to change. Pandas TA comes with two basic Strategies: AllStrategy and CommonStrategy.

Strategy Requirements:

name: Some short memorable string. Note: Case-insensitive "All" is reserved.
ta: A list of dicts containing keyword arguments to identify the indicator and the indicator's arguments

Optional Requirements:

description: A more detailed description of what the Strategy tries to capture. Default: None
created: At datetime string of when it was created. Default: Automatically generated.

Things to note:

A Strategy will fail when consumed by Pandas TA if there is no {"kind": "indicator name"} attribute.

Builtin Examples

All

In [2]:

AllStrategy = ta.AllStrategy
print("name =", AllStrategy.name)
print("description =", AllStrategy.description)
print("created =", AllStrategy.created)
print("ta =", AllStrategy.ta)

name = All
description = All the indicators with their default settings. Pandas TA default.
created = Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)
ta = None

Common

In [3]:

CommonStrategy = ta.CommonStrategy
print("name =", CommonStrategy.name)
print("description =", CommonStrategy.description)
print("created =", CommonStrategy.created)
print("ta =", CommonStrategy.ta)

name = Common Price and Volume SMAs
description = Common Price SMAs: 10, 20, 50, 200 and Volume SMA: 20.
created = Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)
ta = [{'kind': 'sma', 'length': 10}, {'kind': 'sma', 'length': 20}, {'kind': 'sma', 'length': 50}, {'kind': 'sma', 'length': 200}, {'kind': 'sma', 'close': 'volume', 'length': 20, 'prefix': 'VOL'}]

In [ ]:

Creating Strategies

Simple Strategy A

In [4]:

custom_a = ta.Strategy(name="A", ta=[{"kind": "sma", "length": 50}, {"kind": "sma", "length": 200}])
custom_a

Out [4]:

Strategy(name='A', ta=[{'kind': 'sma', 'length': 50}, {'kind': 'sma', 'length': 200}], description='TA Description', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

Simple Strategy B

In [5]:

custom_b = ta.Strategy(name="B", ta=[{"kind": "ema", "length": 8}, {"kind": "ema", "length": 21}, {"kind": "log_return", "cumulative": True}, {"kind": "rsi"}, {"kind": "supertrend"}])
custom_b

Out [5]:

Strategy(name='B', ta=[{'kind': 'ema', 'length': 8}, {'kind': 'ema', 'length': 21}, {'kind': 'log_return', 'cumulative': True}, {'kind': 'rsi'}, {'kind': 'supertrend'}], description='TA Description', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

Bad Strategy. (Misspelled Indicator)

In [6]:

# Misspelled indicator, will fail later when ran with Pandas TA
custom_run_failure = ta.Strategy(name="Runtime Failure", ta=[{"kind": "percet_return"}])
custom_run_failure

Out [6]:

Strategy(name='Runtime Failure', ta=[{'kind': 'percet_return'}], description='TA Description', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

In [ ]:

Strategy Management and Execution with Watchlist

Initialize AlphaVantage Data Source

In [7]:

AV = AlphaVantage(
    api_key="YOUR API KEY", premium=False,
    output_size='full', clean=True,
    export_path=".", export=True
)
AV

Out [7]:

AlphaVantage(
  end_point:str = https://www.alphavantage.co/query,
  api_key:str = YOUR API KEY,
  export:bool = True,
  export_path:str = .,
  output_size:str = full,
  output:str = csv,
  datatype:str = json,
  clean:bool = True,
  proxy:dict = {}
)

Create Watchlist and set it's 'ds' to AlphaVantage

In [8]:

data_source = "av" # Default
# data_source = "yahoo"
watch = Watchlist(["SPY", "IWM"], ds_name=data_source, timed=False)

Info about the Watchlist. Note, the default Strategy is "All"

In [9]:

watch

Out [9]:

Watch(name='Watch: SPY, IWM', ds_name='av', tickers[2]='SPY, IWM', tf='D', strategy[5]='Common Price and Volume SMAs')

Help about Watchlist

In [10]:

help(Watchlist)

Help on class Watchlist in module watchlist:

class Watchlist(builtins.object)
 |  Watchlist(tickers: list, tf: str = None, name: str = None, strategy: pandas_ta.core.Strategy = None, ds_name: str = 'av', **kwargs)
 |  
 |  # Watchlist Class (** This is subject to change! **)
 |  A simple Class to load/download financial market data and automatically
 |  apply Technical Analysis indicators with a Pandas TA Strategy.
 |  
 |  Default Strategy: pandas_ta.CommonStrategy
 |  
 |  ## Package Support:
 |  ### Data Source (Default: AlphaVantage)
 |  - AlphaVantage (pip install alphaVantage-api).
 |  - Python Binance (pip install python-binance). # Future Support
 |  - Yahoo Finance (pip install yfinance). # Almost Supported
 |  
 |  # Technical Analysis:
 |  - Pandas TA (pip install pandas_ta)
 |  
 |  ## Required Arguments:
 |  - tickers: A list of strings containing tickers. Example: ["SPY", "AAPL"]
 |  
 |  Methods defined here:
 |  
 |  __init__(self, tickers: list, tf: str = None, name: str = None, strategy: pandas_ta.core.Strategy = None, ds_name: str = 'av', **kwargs)
 |      Initialize self.  See help(type(self)) for accurate signature.
 |  
 |  __repr__(self) -> str
 |      Return repr(self).
 |  
 |  indicators(self, *args, **kwargs) -> <built-in function any>
 |      Returns the list of indicators that are available with Pandas Ta.
 |  
 |  load(self, ticker: str = None, tf: str = None, index: str = 'date', drop: list = [], plot: bool = False, **kwargs) -> pandas.core.frame.DataFrame
 |      Loads or Downloads (if a local csv does not exist) the data from the
 |      Data Source. When successful, it returns a Data Frame for the requested
 |      ticker. If no tickers are given, it loads all the tickers.
 |  
 |  ----------------------------------------------------------------------
 |  Data descriptors defined here:
 |  
 |  __dict__
 |      dictionary for instance variables (if defined)
 |  
 |  __weakref__
 |      list of weak references to the object (if defined)
 |  
 |  data
 |      When not None, it contains a dictionary of DataFrames keyed by ticker. data = {"SPY": pd.DataFrame, ...}
 |  
 |  name
 |      The name of the Watchlist. Default: "Watchlist: {Watchlist.tickers}".
 |  
 |  strategy
 |      Sets a valid Strategy. Default: pandas_ta.CommonStrategy
 |  
 |  tf
 |      Alias for timeframe. Default: 'D'
 |  
 |  tickers
 |      tickers
 |      
 |      If a string, it it converted to a list. Example: "AAPL" -> ["AAPL"]
 |          * Does not accept, comma seperated strings.
 |      If a list, checks if it is a list of strings.
 |  
 |  verbose
 |      Toggle the verbose property. Default: False

Default Strategy is "Common"

In [11]:

# No arguments loads all the tickers and applies the Strategy to each ticker.
# The result can be accessed with Watchlist's 'data' property which returns a 
# dictionary keyed by ticker and DataFrames as values 
watch.load(verbose=True)

[!] Loading All: SPY, IWM
[+] Downloading[av]: SPY[D]
[+] Strategy: Common Price and Volume SMAs
[i] Indicator arguments: {'timed': False, 'append': True}
[i] Multiprocessing 5 indicators with 7 chunks over 8/8 cpus.
[i] Total indicators: 5
[i] Columns added: 5
[i] Last Run: Saturday March 20, 2021, NYSE: 13:13:13, Local: 17:13:13 PDT, Day 79/365 (22.0%)
[+] Downloading[av]: IWM[D]
[+] Strategy: Common Price and Volume SMAs
[i] Indicator arguments: {'timed': False, 'append': True}
[i] Multiprocessing 5 indicators with 7 chunks over 8/8 cpus.
[i] Total indicators: 5
[i] Columns added: 5
[i] Last Run: Saturday March 20, 2021, NYSE: 13:13:30, Local: 17:13:30 PDT, Day 79/365 (22.0%)

In [12]:

watch.data

Out [12]:

{'SPY':                 open      high       low     close       volume   SMA_10  \
 date                                                                       
 1999-11-01  136.5000  137.0000  135.5625  135.5625    4006500.0      NaN   
 1999-11-02  135.9687  137.2500  134.5937  134.5937    6516900.0      NaN   
 1999-11-03  136.0000  136.3750  135.1250  135.5000    7222300.0      NaN   
 1999-11-04  136.7500  137.3593  135.7656  136.5312    7907500.0      NaN   
 1999-11-05  138.6250  139.1093  136.7812  137.8750    7431500.0      NaN   
 ...              ...       ...       ...       ...          ...      ...   
 2021-03-15  394.3300  396.6850  392.0300  396.4100   73592302.0  387.076   
 2021-03-16  397.0700  397.8300  395.0800  395.9100   73722506.0  388.013   
 2021-03-17  394.5300  398.1200  393.3000  397.2600   97959265.0  389.597   
 2021-03-18  394.4750  396.7200  390.7500  391.4800  115349101.0  391.075   
 2021-03-19  389.8800  391.5690  387.1500  389.4800  113624490.0  391.660   
 
               SMA_20    SMA_50    SMA_200    VOL_SMA_20  
 date                                                     
 1999-11-01       NaN       NaN        NaN           NaN  
 1999-11-02       NaN       NaN        NaN           NaN  
 1999-11-03       NaN       NaN        NaN           NaN  
 1999-11-04       NaN       NaN        NaN           NaN  
 1999-11-05       NaN       NaN        NaN           NaN  
 ...              ...       ...        ...           ...  
 2021-03-15  387.7385  383.6352  349.25675  9.988408e+07  
 2021-03-16  387.9190  384.0758  349.71470  1.010216e+08  
 2021-03-17  388.1625  384.6452  350.17325  1.033322e+08  
 2021-03-18  388.2005  385.0482  350.59025  1.061140e+08  
 2021-03-19  388.1730  385.3668  350.97675  1.076332e+08  
 
 [5380 rows x 10 columns],
 'IWM':               open    high     low   close      volume   SMA_10    SMA_20  \
 date                                                                        
 2000-05-26   91.06   91.44   90.63   91.44     37400.0      NaN       NaN   
 2000-05-30   92.75   94.81   92.75   94.81     28800.0      NaN       NaN   
 2000-05-31   95.13   96.38   95.13   95.75     18000.0      NaN       NaN   
 2000-06-01   97.11   97.31   97.11   97.31      3500.0      NaN       NaN   
 2000-06-02  101.70  102.40  101.70  102.40     14700.0      NaN       NaN   
 ...            ...     ...     ...     ...         ...      ...       ...   
 2021-03-15  233.34  234.53  231.91  234.42  21543170.0  224.147  223.6310   
 2021-03-16  234.02  234.09  229.12  230.50  24675008.0  225.025  223.8645   
 2021-03-17  228.97  232.82  227.36  232.31  29422584.0  226.324  224.2770   
 2021-03-18  230.80  232.93  224.61  225.24  35726301.0  227.529  224.5095   
 2021-03-19  224.57  228.60  222.95  226.94  40855626.0  228.452  224.5970   
 
               SMA_50    SMA_200   VOL_SMA_20  
 date                                          
 2000-05-26       NaN        NaN          NaN  
 2000-05-30       NaN        NaN          NaN  
 2000-05-31       NaN        NaN          NaN  
 2000-06-01       NaN        NaN          NaN  
 2000-06-02       NaN        NaN          NaN  
 ...              ...        ...          ...  
 2021-03-15  217.0930  173.82090  33326068.35  
 2021-03-16  217.7818  174.27890  33409843.35  
 2021-03-17  218.5580  174.73930  33633447.20  
 2021-03-18  219.1330  175.15855  34194686.80  
 2021-03-19  219.5812  175.56925  34675560.35  
 
 [5236 rows x 10 columns]}

In [13]:

watch.data["SPY"]

Out [13]:

	open	high	low	close	volume	SMA_10	SMA_20	SMA_50	SMA_200	VOL_SMA_20
date
1999-11-01	136.5000	137.0000	135.5625	135.5625	4006500.0	NaN	NaN	NaN	NaN	NaN
1999-11-02	135.9687	137.2500	134.5937	134.5937	6516900.0	NaN	NaN	NaN	NaN	NaN
1999-11-03	136.0000	136.3750	135.1250	135.5000	7222300.0	NaN	NaN	NaN	NaN	NaN
1999-11-04	136.7500	137.3593	135.7656	136.5312	7907500.0	NaN	NaN	NaN	NaN	NaN
1999-11-05	138.6250	139.1093	136.7812	137.8750	7431500.0	NaN	NaN	NaN	NaN	NaN
...	...	...	...	...	...	...	...	...	...	...
2021-03-15	394.3300	396.6850	392.0300	396.4100	73592302.0	387.076	387.7385	383.6352	349.25675	9.988408e+07
2021-03-16	397.0700	397.8300	395.0800	395.9100	73722506.0	388.013	387.9190	384.0758	349.71470	1.010216e+08
2021-03-17	394.5300	398.1200	393.3000	397.2600	97959265.0	389.597	388.1625	384.6452	350.17325	1.033322e+08
2021-03-18	394.4750	396.7200	390.7500	391.4800	115349101.0	391.075	388.2005	385.0482	350.59025	1.061140e+08
2021-03-19	389.8800	391.5690	387.1500	389.4800	113624490.0	391.660	388.1730	385.3668	350.97675	1.076332e+08

5380 rows × 10 columns

In [ ]:

In [14]:

watch.load("SPY", plot=True, mas=True)

Out [14]:

[i] Loaded SPY[D]: SPY_D.csv

	open	high	low	close	volume	SMA_10	SMA_20	SMA_50	SMA_200	VOL_SMA_20
date
1999-11-01	136.5000	137.0000	135.5625	135.5625	4006500.0	NaN	NaN	NaN	NaN	NaN
1999-11-02	135.9687	137.2500	134.5937	134.5937	6516900.0	NaN	NaN	NaN	NaN	NaN
1999-11-03	136.0000	136.3750	135.1250	135.5000	7222300.0	NaN	NaN	NaN	NaN	NaN
1999-11-04	136.7500	137.3593	135.7656	136.5312	7907500.0	NaN	NaN	NaN	NaN	NaN
1999-11-05	138.6250	139.1093	136.7812	137.8750	7431500.0	NaN	NaN	NaN	NaN	NaN
...	...	...	...	...	...	...	...	...	...	...
2021-03-15	394.3300	396.6850	392.0300	396.4100	73592302.0	387.076	387.7385	383.6352	349.25675	9.988408e+07
2021-03-16	397.0700	397.8300	395.0800	395.9100	73722506.0	388.013	387.9190	384.0758	349.71470	1.010216e+08
2021-03-17	394.5300	398.1200	393.3000	397.2600	97959265.0	389.597	388.1625	384.6452	350.17325	1.033322e+08
2021-03-18	394.4750	396.7200	390.7500	391.4800	115349101.0	391.075	388.2005	385.0482	350.59025	1.061140e+08
2021-03-19	389.8800	391.5690	387.1500	389.4800	113624490.0	391.660	388.1730	385.3668	350.97675	1.076332e+08

5380 rows × 10 columns

In [ ]:

Easy to swap Strategies and run them

Running Simple Strategy A

In [15]:

# Load custom_a into Watchlist and verify
watch.strategy = custom_a
# watch.debug = True
watch.strategy

Out [15]:

Strategy(name='A', ta=[{'kind': 'sma', 'length': 50}, {'kind': 'sma', 'length': 200}], description='TA Description', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

In [16]:

watch.load("IWM")

Out [16]:

[i] Loaded IWM[D]: IWM_D.csv

	open	high	low	close	volume	SMA_50	SMA_200
date
2000-05-26	91.06	91.44	90.63	91.44	37400.0	NaN	NaN
2000-05-30	92.75	94.81	92.75	94.81	28800.0	NaN	NaN
2000-05-31	95.13	96.38	95.13	95.75	18000.0	NaN	NaN
2000-06-01	97.11	97.31	97.11	97.31	3500.0	NaN	NaN
2000-06-02	101.70	102.40	101.70	102.40	14700.0	NaN	NaN
...	...	...	...	...	...	...	...
2021-03-15	233.34	234.53	231.91	234.42	21543170.0	217.0930	173.82090
2021-03-16	234.02	234.09	229.12	230.50	24675008.0	217.7818	174.27890
2021-03-17	228.97	232.82	227.36	232.31	29422584.0	218.5580	174.73930
2021-03-18	230.80	232.93	224.61	225.24	35726301.0	219.1330	175.15855
2021-03-19	224.57	228.60	222.95	226.94	40855626.0	219.5812	175.56925

5236 rows × 7 columns

Running Simple Strategy B

In [17]:

# Load custom_b into Watchlist and verify
watch.strategy = custom_b
watch.strategy

Out [17]:

Strategy(name='B', ta=[{'kind': 'ema', 'length': 8}, {'kind': 'ema', 'length': 21}, {'kind': 'log_return', 'cumulative': True}, {'kind': 'rsi'}, {'kind': 'supertrend'}], description='TA Description', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

In [18]:

watch.load("SPY")

Out [18]:

[i] Loaded SPY[D]: SPY_D.csv

	open	high	low	close	volume	EMA_8	EMA_21	CUMLOGRET_1	RSI_14	SUPERT_7_3.0	SUPERTd_7_3.0	SUPERTl_7_3.0	SUPERTs_7_3.0
date
1999-11-01	136.5000	137.0000	135.5625	135.5625	4006500.0	NaN	NaN	NaN	NaN	0.000000	1	NaN	NaN
1999-11-02	135.9687	137.2500	134.5937	134.5937	6516900.0	NaN	NaN	-0.007172	NaN	NaN	1	NaN	NaN
1999-11-03	136.0000	136.3750	135.1250	135.5000	7222300.0	NaN	NaN	-0.000461	NaN	NaN	1	NaN	NaN
1999-11-04	136.7500	137.3593	135.7656	136.5312	7907500.0	NaN	NaN	0.007120	NaN	NaN	1	NaN	NaN
1999-11-05	138.6250	139.1093	136.7812	137.8750	7431500.0	NaN	NaN	0.016915	NaN	NaN	1	NaN	NaN
...	...	...	...	...	...	...	...	...	...	...	...	...	...
2021-03-15	394.3300	396.6850	392.0300	396.4100	73592302.0	390.405499	387.590872	1.073016	61.444065	398.910421	-1	NaN	398.910421
2021-03-16	397.0700	397.8300	395.0800	395.9100	73722506.0	391.628722	388.347156	1.071754	60.730184	398.910421	-1	NaN	398.910421
2021-03-17	394.5300	398.1200	393.3000	397.2600	97959265.0	392.880117	389.157415	1.075158	62.013471	398.910421	-1	NaN	398.910421
2021-03-18	394.4750	396.7200	390.7500	391.4800	115349101.0	392.568980	389.368559	1.060502	53.893084	398.910421	-1	NaN	398.910421
2021-03-19	389.8800	391.5690	387.1500	389.4800	113624490.0	391.882540	389.378690	1.055380	51.385705	398.910421	-1	NaN	398.910421

5380 rows × 13 columns

Running Bad Strategy. (Misspelled indicator)

In [19]:

# Load custom_run_failure into Watchlist and verify
watch.strategy = custom_run_failure
watch.strategy

Out [19]:

Strategy(name='Runtime Failure', ta=[{'kind': 'percet_return'}], description='TA Description', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

In [20]:

try:
    iwm = watch.load("IWM")
except AttributeError as error:
    print(f"[X] Oops! {error}")

[i] Loaded IWM[D]: IWM_D.csv
[X] Oops! 'AnalysisIndicators' object has no attribute 'percet_return'

In [ ]:

Indicator Composition/Chaining

When you need an indicator to depend on the value of a prior indicator
Utilitze prefix or suffix to help identify unique columns or avoid column name clashes.

Volume MAs and MA chains

In [21]:

# Set EMA's and SMA's 'close' to 'volume' to create Volume MAs, prefix 'volume' MAs with 'VOLUME' so easy to identify the column
# Take a price EMA and apply LINREG from EMA's output
volmas_price_ma_chain = [
    {"kind":"ema", "close": "volume", "length": 10, "prefix": "VOLUME"},
    {"kind":"sma", "close": "volume", "length": 20, "prefix": "VOLUME"},
    {"kind":"ema", "length": 5},
    {"kind":"linreg", "close": "EMA_5", "length": 8, "prefix": "EMA_5"},
]
vp_ma_chain_ta = ta.Strategy("Volume MAs and Price MA chain", volmas_price_ma_chain)
vp_ma_chain_ta

Out [21]:

Strategy(name='Volume MAs and Price MA chain', ta=[{'kind': 'ema', 'close': 'volume', 'length': 10, 'prefix': 'VOLUME'}, {'kind': 'sma', 'close': 'volume', 'length': 20, 'prefix': 'VOLUME'}, {'kind': 'ema', 'length': 5}, {'kind': 'linreg', 'close': 'EMA_5', 'length': 8, 'prefix': 'EMA_5'}], description='TA Description', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

In [22]:

# Update the Watchlist
watch.strategy = vp_ma_chain_ta
watch.strategy.name

Out [22]:

'Volume MAs and Price MA chain'

In [23]:

spy = watch.load("SPY")
spy

Out [23]:

[i] Loaded SPY[D]: SPY_D.csv

	open	high	low	close	volume	VOLUME_EMA_10	VOLUME_SMA_20	EMA_5	EMA_5_LR_8
date
1999-11-01	136.5000	137.0000	135.5625	135.5625	4006500.0	NaN	NaN	NaN	NaN
1999-11-02	135.9687	137.2500	134.5937	134.5937	6516900.0	NaN	NaN	NaN	NaN
1999-11-03	136.0000	136.3750	135.1250	135.5000	7222300.0	NaN	NaN	NaN	NaN
1999-11-04	136.7500	137.3593	135.7656	136.5312	7907500.0	NaN	NaN	NaN	NaN
1999-11-05	138.6250	139.1093	136.7812	137.8750	7431500.0	NaN	NaN	136.012480	NaN
...	...	...	...	...	...	...	...	...	...
2021-03-15	394.3300	396.6850	392.0300	396.4100	73592302.0	9.820634e+07	9.988408e+07	392.322958	389.839335
2021-03-16	397.0700	397.8300	395.0800	395.9100	73722506.0	9.375474e+07	1.010216e+08	393.518638	391.802334
2021-03-17	394.5300	398.1200	393.3000	397.2600	97959265.0	9.451920e+07	1.033322e+08	394.765759	393.574453
2021-03-18	394.4750	396.7200	390.7500	391.4800	115349101.0	9.830645e+07	1.061140e+08	393.670506	394.215707
2021-03-19	389.8800	391.5690	387.1500	389.4800	113624490.0	1.010915e+08	1.076332e+08	392.273671	393.946701

5380 rows × 9 columns

In [ ]:

MACD BBANDS

In [24]:

# MACD is the initial indicator that BBANDS depends on.
# Set BBANDS's 'close' to MACD's main signal, in this case 'MACD_12_26_9' and add a prefix (or suffix) so it's easier to identify
macd_bands_ta = [
    {"kind":"macd"},
    {"kind":"bbands", "close": "MACD_12_26_9", "length": 20, "prefix": "MACD"}
]
macd_bands_ta = ta.Strategy("MACD BBands", macd_bands_ta, f"BBANDS_{macd_bands_ta[1]['length']} applied to MACD")
macd_bands_ta

Out [24]:

Strategy(name='MACD BBands', ta=[{'kind': 'macd'}, {'kind': 'bbands', 'close': 'MACD_12_26_9', 'length': 20, 'prefix': 'MACD'}], description='BBANDS_20 applied to MACD', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

In [25]:

# Update the Watchlist
watch.strategy = macd_bands_ta
watch.strategy.name

Out [25]:

'MACD BBands'

In [26]:

spy = watch.load("SPY")
spy

Out [26]:

[i] Loaded SPY[D]: SPY_D.csv

	open	high	low	close	volume	MACD_12_26_9	MACDh_12_26_9	MACDs_12_26_9	MACD_BBL_20_2.0	MACD_BBM_20_2.0	MACD_BBU_20_2.0	MACD_BBB_20_2.0
date
1999-11-01	136.5000	137.0000	135.5625	135.5625	4006500.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1999-11-02	135.9687	137.2500	134.5937	134.5937	6516900.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1999-11-03	136.0000	136.3750	135.1250	135.5000	7222300.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1999-11-04	136.7500	137.3593	135.7656	136.5312	7907500.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1999-11-05	138.6250	139.1093	136.7812	137.8750	7431500.0	NaN	NaN	NaN	NaN	NaN	NaN	NaN
...	...	...	...	...	...	...	...	...	...	...	...	...
2021-03-15	394.3300	396.6850	392.0300	396.4100	73592302.0	2.268812	0.874720	1.394091	-0.900289	2.241715	5.383720	280.321462
2021-03-16	397.0700	397.8300	395.0800	395.9100	73722506.0	2.643863	0.999818	1.644046	-0.810546	2.142097	5.094740	275.677819
2021-03-17	394.5300	398.1200	393.3000	397.2600	97959265.0	3.015270	1.096979	1.918291	-0.690283	2.059061	4.808406	267.048329
2021-03-18	394.4750	396.7200	390.7500	391.4800	115349101.0	2.810813	0.714018	2.096795	-0.562339	1.973571	4.509482	256.986903
2021-03-19	389.8800	391.5690	387.1500	389.4800	113624490.0	2.459050	0.289804	2.169246	-0.435212	1.881908	4.199027	246.252186

5380 rows × 12 columns

In [ ]:

Comprehensive Strategy

MACD and RSI Momentum with BBANDS and SMAs and Cumulative Log Returns

In [27]:

momo_bands_sma_ta = [
    {"kind":"sma", "length": 50},
    {"kind":"sma", "length": 200},
    {"kind":"bbands", "length": 20},
    {"kind":"macd"},
    {"kind":"rsi"},
    {"kind":"log_return", "cumulative": True},
    {"kind":"sma", "close": "CUMLOGRET_1", "length": 5, "suffix": "CUMLOGRET"},
]
momo_bands_sma_strategy = ta.Strategy(
    "Momo, Bands and SMAs and Cumulative Log Returns", # name
    momo_bands_sma_ta, # ta
    "MACD and RSI Momo with BBANDS and SMAs 50 & 200 and Cumulative Log Returns" # description
)
momo_bands_sma_strategy

Out [27]:

Strategy(name='Momo, Bands and SMAs and Cumulative Log Returns', ta=[{'kind': 'sma', 'length': 50}, {'kind': 'sma', 'length': 200}, {'kind': 'bbands', 'length': 20}, {'kind': 'macd'}, {'kind': 'rsi'}, {'kind': 'log_return', 'cumulative': True}, {'kind': 'sma', 'close': 'CUMLOGRET_1', 'length': 5, 'suffix': 'CUMLOGRET'}], description='MACD and RSI Momo with BBANDS and SMAs 50 & 200 and Cumulative Log Returns', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

In [28]:

# Update the Watchlist
watch.strategy = momo_bands_sma_strategy
watch.strategy.name

Out [28]:

'Momo, Bands and SMAs and Cumulative Log Returns'

In [29]:

spy = watch.load("SPY")
# Apply constants to the DataFrame for indicators
spy.ta.constants(True, [0, 30, 70])
spy.tail()

Out [29]:

[i] Loaded SPY[D]: SPY_D.csv

	open	high	low	close	volume	SMA_50	SMA_200	BBL_20_2.0	BBM_20_2.0	BBU_20_2.0	BBB_20_2.0	MACD_12_26_9	MACDh_12_26_9	MACDs_12_26_9	RSI_14	CUMLOGRET_1	SMA_5_CUMLOGRET	0	30	70
date
2021-03-15	394.330	396.685	392.03	396.41	73592302.0	383.6352	349.25675	377.450608	387.7385	398.026392	5.306614	2.268812	0.874720	1.394091	61.444065	1.073016	1.062176	0	30	70
2021-03-16	397.070	397.830	395.08	395.91	73722506.0	384.0758	349.71470	377.199689	387.9190	398.638311	5.526572	2.643863	0.999818	1.644046	60.730184	1.071754	1.066640	0	30	70
2021-03-17	394.530	398.120	393.30	397.26	97959265.0	384.6452	350.17325	376.843518	388.1625	399.481482	5.832084	3.015270	1.096979	1.918291	62.013471	1.075158	1.070545	0	30	70
2021-03-18	394.475	396.720	390.75	391.48	115349101.0	385.0482	350.59025	376.842394	388.2005	399.558606	5.851670	2.810813	0.714018	2.096795	53.893084	1.060502	1.069500	0	30	70
2021-03-19	389.880	391.569	387.15	389.48	113624490.0	385.3668	350.97675	376.830092	388.1730	399.515908	5.844254	2.459050	0.289804	2.169246	51.385705	1.055380	1.067162	0	30	70

In [ ]:

Additional Strategy Options

The params keyword takes a tuple as a shorthand to the parameter arguments in order.

Note: If the indicator arguments change, so will results. Breaking Changes will always be posted on the README.

The col_numbers keyword takes a tuple specifying which column to return if the result is a DataFrame.

In [30]:

params_ta = [
    {"kind":"ema", "params": (10,)},
    # params sets MACD's keyword arguments: fast=9, slow=19, signal=10
    # and returning the 2nd column: histogram
    {"kind":"macd", "params": (9, 19, 10), "col_numbers": (1,)},
    # Selects the Lower and Upper Bands and renames them LB and UB, ignoring the MB
    {"kind":"bbands", "col_numbers": (0,2), "col_names": ("LB", "UB")},
    {"kind":"log_return", "params": (5, False)},
]
params_ta_strategy = ta.Strategy(
    "EMA, MACD History, Outter BBands, Log Returns", # name
    params_ta, # ta
    "EMA, MACD History, BBands(LB, UB), and Log Returns Strategy" # description
)
params_ta_strategy

Out [30]:

Strategy(name='EMA, MACD History, Outter BBands, Log Returns', ta=[{'kind': 'ema', 'params': (10,)}, {'kind': 'macd', 'params': (9, 19, 10), 'col_numbers': (1,)}, {'kind': 'bbands', 'col_numbers': (0, 2), 'col_names': ('LB', 'UB')}, {'kind': 'log_return', 'params': (5, False)}], description='EMA, MACD History, BBands(LB, UB), and Log Returns Strategy', created='Saturday March 20, 2021, NYSE: 13:13:10, Local: 17:13:10 PDT, Day 79/365 (22.0%)')

In [31]:

# Update the Watchlist
watch.strategy = params_ta_strategy
watch.strategy.name

Out [31]:

'EMA, MACD History, Outter BBands, Log Returns'

In [32]:

spy = watch.load("SPY")
spy.tail()

Out [32]:

[i] Loaded SPY[D]: SPY_D.csv

	open	high	low	close	volume	EMA_10	MACDh_9_19_10	LB	UB	LOGRET_5
date
2021-03-15	394.330	396.685	392.03	396.41	73592302.0	389.640175	1.360361	385.510398	398.789602	0.037762
2021-03-16	397.070	397.830	395.08	395.91	73722506.0	390.780143	1.425897	389.067672	398.728328	0.022323
2021-03-17	394.530	398.120	393.30	397.26	97959265.0	391.958299	1.461750	392.601825	398.266175	0.019522
2021-03-18	394.475	396.720	390.75	391.48	115349101.0	391.871336	0.889025	390.906241	399.141759	-0.005223
2021-03-19	389.880	391.569	387.15	389.48	113624490.0	391.436547	0.302361	387.988766	400.227234	-0.011691

In [ ]:

Disclaimer

All investments involve risk, and the past performance of a security, industry, sector, market, financial product, trading strategy, or individual’s trading does not guarantee future results or returns. Investors are fully responsible for any investment decisions they make. Such decisions should be based solely on an evaluation of their financial circumstances, investment objectives, risk tolerance, and liquidity needs.
Any opinions, news, research, analyses, prices, or other information offered is provided as general market commentary, and does not constitute investment advice. I will not accept liability for any loss or damage, including without limitation any loss of profit, which may arise directly or indirectly from use of or reliance on such information.

208 KiB Raw Blame History Unescape Escape

Pandas TA (pandas_ta) Strategies for Custom Technical Analysis

Topics

What is a Pandas TA Strategy?

Strategy Requirements:

Optional Requirements:

Things to note:

Builtin Examples

All

Common

Creating Strategies

Simple Strategy A

Simple Strategy B

Bad Strategy. (Misspelled Indicator)

Strategy Management and Execution with Watchlist

Initialize AlphaVantage Data Source

Create Watchlist and set it's 'ds' to AlphaVantage

Info about the Watchlist. Note, the default Strategy is "All"

Help about Watchlist

Default Strategy is "Common"

Easy to swap Strategies and run them

Running Simple Strategy A

Running Simple Strategy B

Running Bad Strategy. (Misspelled indicator)

Indicator Composition/Chaining

Volume MAs and MA chains

MACD BBANDS

Comprehensive Strategy

MACD and RSI Momentum with BBANDS and SMAs and Cumulative Log Returns

Additional Strategy Options

Disclaimer

208 KiB

Raw Blame History