Loading ...

Teach you to encapsulate a Python strategy into a local file

Author: , Created: 2020-07-09 10:21:31, Updated:

Many developers who write strategies in Python want to put the strategy code files locally, worrying about the safety of the strategy. As a solution proposed in the FMZ API document:

Strategy security

The strategy is developed on the FMZ platform, and the strategy is only visible to the FMZ account holders. And on the FMZ platform, the strategy code can be completely localized, for example, the strategy is encapsulated into a Python package and loaded in the strategy code, so that the strategy localization is realized.

For more details, please go to: https://www.fmz.com/api

In fact, this kind of worry is not necessary, but since there are such needs, we shall provide a complete implementation example.

Encapsulate a strategy

Let’s find a simple Python strategy for demonstration, using the classic Dual Thrust strategy, strategy address: https://www.fmz.com/strategy/21856 We strive not to change any part of the strategy code, encapsulate the strategy into a file that can be called by the strategy code on the FMZ platform, and the execution result is exactly the same as running the strategy directly. The biggest problem with encapsulation is that the global objects, global functions, and constant values called by the strategy code on the FMZ platform cannot be accessed in the files we encapsulate, so we must find a way to pass these objects, functions, variables, and constants to the encapsulated file. let’s do it step by step.

img

Paste into the file testA opened by the local editor.

img

Add some code, and keep the strategy code part copied and pasted intact

# Function, object
exchanges = None
exchange = None
Log = None
Sleep = None
TA = None
Chart = None
LogProfitReset = None
LogStatus = None
_N = None
_C = None 
LogProfit = None  


# Strategy parameters
ContractTypeIdx = None
MarginLevelIdx = None
NPeriod = None
Ks = None
Kx = None
AmountOP = None
Interval = None
LoopInterval = None
PeriodShow = None  

# constant
ORDER_STATE_PENDING = 0
ORDER_STATE_CLOSED = 1
ORDER_STATE_CANCELED = 2
ORDER_STATE_UNKNOWN = 3
ORDER_TYPE_BUY = 0 
ORDER_TYPE_SELL = 1
PD_LONG = 0
PD_SHORT = 1  


def SetExchanges(es):
    global exchanges, exchange
    exchanges = es
    exchange = es[0]  

def SetFunc(pLog, pSleep, pTA, pChart, pLogStatus, pLogProfitReset, p_N, p_C, pLogProfit):
    global Log, Sleep, TA, Chart, LogStatus, LogProfitReset, _N, _C, LogProfit
    Log = pLog
    Sleep = pSleep
    TA = pTA
    Chart = pChart
    LogStatus = pLogStatus
    LogProfitReset = pLogProfitReset
    _N = p_N
    _C = p_C
    LogProfit = pLogProfit  

def SetParams(pContractTypeIdx, pMarginLevelIdx, pNPeriod, pKs, pKx, pAmountOP, pInterval, pLoopInterval, pPeriodShow):
    global ContractTypeIdx, MarginLevelIdx, NPeriod, Ks, Kx, AmountOP, Interval, LoopInterval, PeriodShow
    ContractTypeIdx = pContractTypeIdx
    MarginLevelIdx = pMarginLevelIdx
    NPeriod = pNPeriod
    Ks = pKs
    Kx = pKx
    AmountOP = pAmountOP
    Interval = pInterval
    LoopInterval = pLoopInterval
    PeriodShow = pPeriodShow

The main function of the above code is to declare the global functions and variables used in the current file. Then reserve the interfaces SetExchanges, SetParams, SetFunc to import these functions. Strategies on the FMZ platform call these functions and pass over some used functions and objects.

Startup strategy on FMZ platform

The startup strategy is very simple, as follows:

img

There are only a few lines of code written on the FMZ platform. It should be noted that the parameters of this startup strategy are exactly the same as our packaged strategy Python version of the Dual Thrust OKCoin futures strategy code. In fact, you can directly copy Python version of the Dual Thrust OKCoin futures strategy code Strategy, then just clear the strategy code, paste it.

import sys
# Here I wrote the path where I put the testA file myself. I replaced it with xxx. To put it simply, I set the path of my testA file.
sys.path.append("/Users/xxx/Desktop/pythonPlayground/")
import testA

def main():
    # Passing Exchange Object
    testA.SetExchanges(exchanges)
    # Pass global function SetFunc(pLog, pSleep, pTA, pChart, pLogStatus, pLogProfitReset, p_N, p_C, pLogProfit)
    testA.SetFunc(Log, Sleep, TA, Chart, LogStatus, LogProfitReset, _N, _C, LogProfit)
    # Passing strategy parameters SetParams(pContractTypeIdx, pMarginLevelIdx, pNPeriod, pKs, pKx, pAmountOP, pInterval, pLoopInterval, pPeriodShow)
    testA.SetParams(ContractTypeIdx, MarginLevelIdx, NPeriod, Ks, Kx, AmountOP, Interval, LoopInterval, PeriodShow)
    # Execute the main strategy function in the encapsulated testA file
    testA.main()

In this way, we encapsulate the main body of the strategy logic in the testA file and place it locally on the device where the docker is located. On the FMZ platform, we only need to save a startup strategy. The robot that creates this startup strategy can directly load our local file and run it locally.

Backtesting comparison

  • Load testA file locally for backtest

img

  • Original strategy, backtesting on public server

img

Another simpler way

Load the file directly for execution. This time we prepare a testB file with the code for the Python version of the Dual Thrust OKCoin futures strategy code strategy.

import time
class Error_noSupport(BaseException):
    def __init__(self):
        Log("Only OKCoin futures are supported!#FF0000")

class Error_AtBeginHasPosition(BaseException):
    def __init__(self):
        Log("There is a futures position at startup!#FF0000")

ChartCfg = {
    '__isStock': True,
    'title': {
        'text': 'Dual Thrust Top and bottom rail map'
    },
    'yAxis': {

...

If the strategy is too long, it is omitted and the strategy code does not need to be changed at all.

Then prepare Python version of the Dual Thrust OKCoin futures strategy code (start strategy, directly execute testB file), which is our strategy on the FMZ platform, create a robot, directly load the testB file, and execute it directly. It should be noted that the startup strategy must also have exactly the same strategy parameter settings (strategy interface parameters) as the original version of Python version of the Dual Thrust OKCoin futures strategy code.

img

if __name__ == '__main__':
    Log("run...")
    try:
        # The file path is processed, you can write the actual path of your testB file
        f = open("/Users/xxx/Desktop/pythonPlayground/testB.py", "r")
        code = f.read()
        exec(code)
    except Exception as e:
        Log(e)

Perform a backtest:

img

The backtest result is consistent with the above test.

Obviously the second method above is simpler, it is recommended to use.


More