TakeProfit logo CommunityPlatform

Indicator code structure

The general code structure of an indicator 'module' looks like this:

COMMENT_DIRECTIVES
IMPORTS
HELPER_FUNCTION_DEFINITIONS (optional)           ─╮
DECORATED_HELPER_FUNCTION_DEFINITIONS (optional)  ├  can go in any order
CLASS_DEFINITIONS (optional)                      │
MAIN_FUNCTION_OR_CLASS_DEFINITION                ─╯

Syntactically, Indie code is a Python code. But, there are some differences that you may want to be aware about.

Comment directives

Comment directives are special comments that have some effect on how Indie's compiler and/or runtime works. At the moment there is only one comment directive which is:

# indie:lang_version = 4

and it declares the language version of the source code of the indicator.

Imports

The IMPORTS section consists of several import statements. Every import statement tells Indie runtime to import (or connect) one or more symbols from libraries. After that these symbols may be used in the indicator code.

from indie import indicator  # This line imports the `indicator`
                             # decorator from `indie` library package

@indicator('Example')  # Here we use the imported `indicator` decorator
def Main(self):
    # ...

Like in Python, there are several different forms of imports:

import indie.algorithms  # Then you can use any symbols from this package
                         # with their full name, e.g. `indie.algorithms.Sma`

import indie.algorithms as alg  # Then you can use any symbols from this package
                                # with a short package alias, e.g. `alg.Sma`

from indie.algorithms import Sma  # Then you can use only `Sma` symbol from
                                  # this package (without the package prefix)

from indie.algorithms import Sma as my_sma  # Then you can use the symbol
                                            # with an alias `my_sma`

Also in every from form you are allowed to use lists of symbols, for example:

from indie.algorithms import Sma, Wma, Ema
from indie.algorithms import Sma as s, Wma as w, Ema as e  # You can do this, but this is not a
                                                           # good coding style! Because this makes
                                                           # your code harder to read and understand

One note about the aliases. Do not use them if you do not really need them. And you most certainly will need them in two cases:

  • You have a name clash of two symbols from different libraries. Add an alias to one or both of the symbols to resolve the clash. For example:
from some.library import foobar_func
from another.library import foobar_func

Obviously, when you try to call the imported function foobar_func() afterwards, it is unclear which one of the two imported functions should be called. Such code cannot be run and produces an error.

Same example with a name clash resolved:

from some.library import foobar_func
from another.library import foobar_func as another_foobar_func
# Now you may call both `foobar_func()` and `another_foobar_func()` without any ambiguity
  • You import a symbol with a long name, and it is very uncomfortable to use that long name in the code:
from some.library import some_useful_function_with_a_very_long_name
# Call the function later as `some_useful_function_with_a_very_long_name()`

Instead, it is better to use an alias:

from some.library import some_useful_function_with_a_very_long_name as useful_func
# Call the function later as `useful_func()`

Helper function definitions

Indicator algorithms could be quite complex, so it is a good coding style to extract blocks of code into helper functions with some meaningful names. The minimal function that accepts no arguments, does nothing (Python keyword pass is about that) and returns nothing (None) looks like this:

def minimal_helper_func() -> None:
    pass

Obviously there will be not much help of such a helper function like minimal_helper_func, it's just an example. Here is a more realistic example of a function that calculates a maximum of given two integers:

def max(a: int, b: int) -> int:
    if a > b:
        return a
    else:
        return b

At the moment, it is mandatory to declare type hints for every argument as well as for the return value of all the functions (except for the Main function and @sec_context-decorated functions). Read more about typing in the Data types in Indie chapter.

Decorated helper function definitions

There are two decorators that can be applied to helper functions: @algorithm and @sec_context. Both of them are syntactic sugar decorators which help a lot in making indicator code compact and readable.

The @algorithm decorator is a syntactic sugar for making your own series processing algorithm like indie.algorithms.Sma.

@algorithm
def SomeFunc(self) -> SeriesF:
    # Some code...

The @sec_context decorator is a syntactic sugar for making an extra entry point function (like a secondary Main) for additional instrument that indicator may request. This function should be used in combination with a Context.calc_on function. Read more about it in Request additional instruments.

@sec_context
def SomeSecContext(self):
    # Some code, that processes the BINANCEUS:BTC/USD candles data...

@indicator('Example')
class Main(MainContext):
    def __init__(self):
        self._sec_context_res = self.calc_on(SomeSecContext, exchange='BINANCEUS', ticker='BTC/USD')

    def calc(self):
        # Some code that uses self._sec_context_res

Class definitions

From Indie v4 the language partially supports OOP and this support is going to be expanded. At the moment this means that Indie has class keyword that lets to declare classes, just like in Python.

For example the Main entry point could be declared not as a function, but as a class, like this:

# indie:lang_version = 4
from indie import indicator, MainContext

@indicator('Example')
class Main(MainContext):
    def calc(self):
        return self.close[0]

Constructor (the __init__ method) could be optionally added too. It is very useful because it is a good place for some initialization code that must be executed only once. For example:

# indie:lang_version = 4
from indie import indicator, MainContext, IndieError

@indicator('Example')
class Main(MainContext):
    def __init__(self):
        # This code is executed only once at the creation
        # of the indicator before calculations on candle data
        c: str
        if 'EUR' in self.info.ticker:
            c = 'EUR'
        elif 'JPY' in self.info.ticker:
            c = 'JPY'
        elif 'USD' in self.info.ticker:
            c = 'USD'
        else:
            raise IndieError('This indie cannot be added on ' + self.info.ticker + ' chart')
        self._currency = c

    def calc(self):
        # Calculations on candle data are here
        # TODO: Use `self._currency` somehow here...
        return self.close[0]

Secondary contexts can be declared as classes (instead of using @sec_context decorator on a func) in a similar way, for example:

# indie:lang_version = 4
from indie import SecContext

class SecHighLow(SecContext):
    def calc(self):
        return self.high[0], self.low[0]

Finally, algorithms (instead of using @algorithm decorator on a func) can be declared as classes too, for example:

# indie:lang_version = 4
from indie import Algorithm, Context, SeriesF, MutSeriesF

class MyAlgorithm(Algorithm):
    def __init__(self, ctx: Context):
        super().__init__(ctx)

    def calc(self, src1: SeriesF, src2: SeriesF) -> SeriesF:
        return MutSeriesF.new(src1[0] + src2[0])

NOTE: At the moment OOP support in Indie is limited to declaration of main and secondary entry points and algorithms. It is not possible to declare classes for any other purposes yet. But the Indie team is working on it.

Main function or class definition

@indicator('Example 1')
def Main(self):
    # Some code here...

The Main entry point of the whole indicator is mandatory. It is called every time the indicator receives a candle data update and it should return an indicator result for that piece of data. It must have at least one parameter self which is a reference to object of Main class (inherited from MainContext type). Function or class Main must also be decorated with a @indicator decorator. But also other decorators could be added here, for example: