optopsy 2.0.0b1

Python Backtesting library for options trading strategies

Optopsy

Optopsy is a nimble backtesting and statistics library for option strategies, it is designed to answer questions like "How do vertical spreads perform on the SPX?" or "Which strikes and/or expiration dates should I choose to make the most potential profit?"

This library is currently being rewritten and is in beta state, use at your own risk

Supported Option Strategies

• Calls/Puts
• Straddles/Strangles (WIP)
• Butterflies/Iron Condors (WIP)
• Many more to follow

Methodology

Most backtesting tools for option strategies don't reveal how the options are backtested, since this is open sourced, I will try to be as transparent as possible on how the algorithm works. You can decide if it works for your situation. My aim for this libary is not to be able to backtest every possible event driven scenarios out there, but to focus on answering the core questions laid out in the introduction. There are no fancy features such as being 'event driven', but theoretically, it is possible by injecting a list of entry and exit dates created from external sources based on indicators.

There are two main parts of the library:

• Statistics Module - Focused on generating statistics on various options strategies (Independent of chronological events)
• Backtester Module (Comming Soon) - Replay the generated statistics with chronological order to create an approximated backtest.

The algorithm for the statistics module is as follows:

1. Evaluate all option chains provided (filter for desired the entry and exit dates and calculate the profit/loss of each individual option chain)
2. Group each 'evaluated' option chain into buckets (buckets are assigned to 'Days to Expiration' (grouped by days in intervals of 7, by default), and either 'delta' or 'strike distance percent' from current price (grouped in intervals of 5%, by default). Obviously, the smaller the intervals, the more accurate the results should be
3. Aggregate all the option chains into their buckets and calculate the average profit loss for each bucket combination
4. Construct the legs of the option strategy with the previous aggregate amounts and net out the profit/loss
5. The result will contain approximated profit/loss amounts (and other statistics such as min/max, distributions) for all the combinations of inputs (strike dist %/ DTE) of the option strategy

Obviously this is a close approximation of a backtest (due to the bucketing and averaging) and also it does not take into account real world events in a chronological order. For that there, is the backtester module that will replay each backtest ordered by expiration dates with a running balance to simulate what would have happened.

Notes

As the algorithm is heavily based on bucketing and approximations to improve performance, it is not recommeded to make trade decisions based on the results from this library. Please use at your own risk.

Usage

Use Your Data

• Use data from any source, just provide a Pandas dataframe with the required columns when calling optopsy functions.

Dependencies

You will need Python 3.7.x and Pandas 0.24.1 or newer. It is recommended to install Miniconda3. See requirements.txt for full details.

Installation

pip install optopsy

Example

Let's see how long calls perform on the SPX on a small demo dataset on the SPX: Download the following data sample from DeltaNeutral: http://www.deltaneutral.com/files/Sample_SPX_20151001_to_20151030.csv

This dataset is for the month of October in 2015, lets load it into Optopsy. First create a small helper function that returns a file path to our file. We will store it under a folder named 'data', in the same directory as the working python file.

def filepath():
    curr_file = os.path.abspath(os.path.dirname(__file__))
    return os.path.join(curr_file, "./data/SPX_20151001_to_20151030.csv")

Next lets use this function to pass in the file path string into Optopsy's csv_data() function, we will map the column indices using the defined function parameters. We are omitting the start_date and end_date parameters in this call because we want to include the entire dataset. The numeric values represent the column number as found in the sample file, the numbers are 0-indexed:

import optopsy as op

spx_data = op.csv_data(
    filepath(),
    underlying_symbol=0,
    underlying_price=1,
    option_type=5,
    expiration=6,
    quote_date=7,
    strike=8,
    bid=10,
    ask=11,
)

The csv_data() function is a convenience function. Under the hood it uses Panda's read_csv() function to do the import. There are other parameters that can help with loading the csv data, consult the code/future documentation to see how to use them.

Optopsy is a small simple library that offloads the heavy work of backtesting option strategies, the API is designed to be simple and easy to implement into your regular Panda's data analysis workflow. As such, we just need to call the single_calls() function to have Optopsy generate all combinations of a simple long call strategy for the specified time period.

long_single_calls = op.singles_calls(spx_data, strike_dist_pct_interval=0.05, side="long")

At this point, we will be returned a Pandas DataFrame containing the statistics of running long calls on the SPX (The example below is truncated):

	dte_range	strike_otm_pct_range	long_profit_pct_min	long_profit_pct_max	long_profit_pct_mean	long_profit_pct_median	long_profit_pct_std	long_profit_pct_count
0	(0, 7]	(-1.0, -0.95]	-0.01	0.07	0.019	0.02	0.0191669	20
1	(0, 7]	(-0.95, -0.9]	-0.01	0.06	0.0242857	0.02	0.0201424	21
2	(0, 7]	(-0.9, -0.85]	0	0.07	0.0265	0.02	0.0192696	20
3	(0, 7]	(-0.85, -0.8]	-0.01	0.06	0.0211111	0.02	0.0198714	27
4	(0, 7]	(-0.8, -0.75]	0	0.08	0.0288889	0.02	0.0222457	27
5	(0, 7]	(-0.75, -0.7]	-0.01	0.08	0.0251515	0.02	0.0227927	33
6	(0, 7]	(-0.7, -0.65]	-0.01	0.09	0.031	0.025	0.0250947	40
7	(0, 7]	(-0.65, -0.6]	-0.01	0.09	0.0256757	0.02	0.0253356	37
8	(0, 7]	(-0.6, -0.55]	-0.02	0.09	0.0252055	0.02	0.0272901	73
9	(0, 7]	(-0.55, -0.5]	-0.02	0.1	0.0291772	0.03	0.0255877	158
10	(0, 7]	(-0.5, -0.45]	-0.02	0.11	0.032337	0.03	0.0293737	184
11	(0, 7]	(-0.45, -0.4]	-0.02	0.12	0.0357711	0.03	0.0321018	201
12	(0, 7]	(-0.4, -0.35]	-0.02	0.13	0.0392308	0.04	0.0351624	247
13	(0, 7]	(-0.35, -0.3]	-0.02	0.15	0.0455743	0.04	0.0389481	296
14	(0, 7]	(-0.3, -0.25]	-0.03	0.17	0.0539514	0.05	0.045708	329
15	(0, 7]	(-0.25, -0.2]	-0.03	0.2	0.0617045	0.05	0.0539594	352
16	(0, 7]	(-0.2, -0.15]	-0.04	0.26	0.0779634	0.07	0.0673121	383
17	(0, 7]	(-0.15, -0.1]	-0.06	0.37	0.105468	0.09	0.0919648	417
18	(0, 7]	(-0.1, -0.05]	-0.12	0.69	0.176399	0.15	0.158725	461
19	(0, 7]	(-0.05, 0.0]	-1	7.62	0.63998	0.37	1.0276	505
20	(0, 7]	(0.0, 0.05]	-1	68	2.33606	-0.89	8.64909	269
21	(0, 7]	(0.05, 0.1]	-1	-1	-1	-1	0	2
22	(7, 14]	(-1.0, -0.95]	0.01	0.09	0.0483333	0.045	0.0272475	12
23	(7, 14]	(-0.95, -0.9]	0.01	0.1	0.0545455	0.05	0.0230377	22
24	(7, 14]	(-0.9, -0.85]	0.01	0.09	0.048125	0.05	0.0210456	16
25	(7, 14]	(-0.85, -0.8]	0.01	0.11	0.062	0.06	0.0252566	20
26	(7, 14]	(-0.8, -0.75]	0.01	0.11	0.0584	0.06	0.0271846	25
27	(7, 14]	(-0.75, -0.7]	0.01	0.12	0.0642857	0.06	0.029761	21
28	(7, 14]	(-0.7, -0.65]	0.02	0.12	0.0678788	0.06	0.0244639	33
29	(7, 14]	(-0.65, -0.6]	0.02	0.13	0.0677778	0.07	0.0327383	27
30	(7, 14]	(-0.6, -0.55]	0.02	0.14	0.0691111	0.07	0.0287483	45
31	(7, 14]	(-0.55, -0.5]	0.02	0.15	0.069386	0.07	0.0299641	114
32	(7, 14]	(-0.5, -0.45]	0.02	0.16	0.080719	0.08	0.0334645	153
33	(7, 14]	(-0.45, -0.4]	0.02	0.17	0.0868072	0.08	0.0383135	166
34	(7, 14]	(-0.4, -0.35]	0.02	0.19	0.0940102	0.09	0.041301	197
35	(7, 14]	(-0.35, -0.3]	0.02	0.21	0.108085	0.1	0.0436594	235
36	(7, 14]	(-0.3, -0.25]	0.03	0.25	0.125094	0.12	0.0523545	265
37	(7, 14]	(-0.25, -0.2]	0.03	0.3	0.147393	0.14	0.0619861	280
38	(7, 14]	(-0.2, -0.15]	0.04	0.38	0.182899	0.18	0.0776728	307
39	(7, 14]	(-0.15, -0.1]	0.05	0.54	0.245663	0.24	0.105847	332
40	(7, 14]	(-0.1, -0.05]	0.07	0.97	0.401595	0.39	0.18451	370
41	(7, 14]	(-0.05, 0.0]	-0.46	4.4	1.01542	0.865	0.676838	404
42	(7, 14]	(0.0, 0.05]	-1	32	1.5167	-0.73	4.44642	388
43	(7, 14]	(0.05, 0.1]	-1	-0.83	-0.928333	-0.935	0.0629512	36

Since this is just a regular Pandas DataFrame, let's rewrite the previous function with a Panda's round function:

long_single_calls = op.singles_calls(spx_data, strike_dist_pct_interval=0.05, side="long").round(2)

This results in a cleaner output, at this point you may use regular Panda's function to further analyze the results as you need

	dte_range	strike_otm_pct_range	long_profit_pct_min	long_profit_pct_max	long_profit_pct_mean	long_profit_pct_median	long_profit_pct_std	long_profit_pct_count
0	(0, 7]	(-1.0, -0.95]	-0.01	0.07	0.02	0.02	0.02	20
1	(0, 7]	(-0.95, -0.9]	-0.01	0.06	0.02	0.02	0.02	21
2	(0, 7]	(-0.9, -0.85]	0	0.07	0.03	0.02	0.02	20
3	(0, 7]	(-0.85, -0.8]	-0.01	0.06	0.02	0.02	0.02	27
4	(0, 7]	(-0.8, -0.75]	0	0.08	0.03	0.02	0.02	27
5	(0, 7]	(-0.75, -0.7]	-0.01	0.08	0.03	0.02	0.02	33
6	(0, 7]	(-0.7, -0.65]	-0.01	0.09	0.03	0.02	0.03	40
7	(0, 7]	(-0.65, -0.6]	-0.01	0.09	0.03	0.02	0.03	37
8	(0, 7]	(-0.6, -0.55]	-0.02	0.09	0.03	0.02	0.03	73
9	(0, 7]	(-0.55, -0.5]	-0.02	0.1	0.03	0.03	0.03	158
10	(0, 7]	(-0.5, -0.45]	-0.02	0.11	0.03	0.03	0.03	184
11	(0, 7]	(-0.45, -0.4]	-0.02	0.12	0.04	0.03	0.03	201
12	(0, 7]	(-0.4, -0.35]	-0.02	0.13	0.04	0.04	0.04	247
13	(0, 7]	(-0.35, -0.3]	-0.02	0.15	0.05	0.04	0.04	296
14	(0, 7]	(-0.3, -0.25]	-0.03	0.17	0.05	0.05	0.05	329
15	(0, 7]	(-0.25, -0.2]	-0.03	0.2	0.06	0.05	0.05	352
16	(0, 7]	(-0.2, -0.15]	-0.04	0.26	0.08	0.07	0.07	383
17	(0, 7]	(-0.15, -0.1]	-0.06	0.37	0.11	0.09	0.09	417
18	(0, 7]	(-0.1, -0.05]	-0.12	0.69	0.18	0.15	0.16	461
19	(0, 7]	(-0.05, 0.0]	-1	7.62	0.64	0.37	1.03	505
20	(0, 7]	(0.0, 0.05]	-1	68	2.34	-0.89	8.65	269
21	(0, 7]	(0.05, 0.1]	-1	-1	-1	-1	0	2
22	(7, 14]	(-1.0, -0.95]	0.01	0.09	0.05	0.04	0.03	12
23	(7, 14]	(-0.95, -0.9]	0.01	0.1	0.05	0.05	0.02	22
24	(7, 14]	(-0.9, -0.85]	0.01	0.09	0.05	0.05	0.02	16
25	(7, 14]	(-0.85, -0.8]	0.01	0.11	0.06	0.06	0.03	20
26	(7, 14]	(-0.8, -0.75]	0.01	0.11	0.06	0.06	0.03	25
27	(7, 14]	(-0.75, -0.7]	0.01	0.12	0.06	0.06	0.03	21
28	(7, 14]	(-0.7, -0.65]	0.02	0.12	0.07	0.06	0.02	33
29	(7, 14]	(-0.65, -0.6]	0.02	0.13	0.07	0.07	0.03	27
30	(7, 14]	(-0.6, -0.55]	0.02	0.14	0.07	0.07	0.03	45
31	(7, 14]	(-0.55, -0.5]	0.02	0.15	0.07	0.07	0.03	114
32	(7, 14]	(-0.5, -0.45]	0.02	0.16	0.08	0.08	0.03	153
33	(7, 14]	(-0.45, -0.4]	0.02	0.17	0.09	0.08	0.04	166
34	(7, 14]	(-0.4, -0.35]	0.02	0.19	0.09	0.09	0.04	197
35	(7, 14]	(-0.35, -0.3]	0.02	0.21	0.11	0.1	0.04	235
36	(7, 14]	(-0.3, -0.25]	0.03	0.25	0.13	0.12	0.05	265
37	(7, 14]	(-0.25, -0.2]	0.03	0.3	0.15	0.14	0.06	280
38	(7, 14]	(-0.2, -0.15]	0.04	0.38	0.18	0.18	0.08	307
39	(7, 14]	(-0.15, -0.1]	0.05	0.54	0.25	0.24	0.11	332
40	(7, 14]	(-0.1, -0.05]	0.07	0.97	0.4	0.39	0.18	370
41	(7, 14]	(-0.05, 0.0]	-0.46	4.4	1.02	0.86	0.68	404
42	(7, 14]	(0.0, 0.05]	-1	32	1.52	-0.73	4.45	388
43	(7, 14]	(0.05, 0.1]	-1	-0.83	-0.93	-0.94	0.06	36

There are more customization options for Optopsy's strategy functions, consult the codebase/future documentation to see how it can be used to adjust the results, such as increasing/decreasing the intervals and other data to be returned.

The library is currently under development, as such expect changes to the API in the future.