tony lam

A simple ATR strategy

Trading Strategy

In this article, we will implement a simple trading strategy using the ATR indicator.


Trading Idea:

From the discussion in the previous post, ATR itself does not provide indication about price trend. Instead, it is used to measure market volatility. A stock with a high level of volatility usually has a higher ATR, and similarly, a stock with a lower volatility has a lower ATR.

An old trading saying that "Market take the stairs up and the elevator down." This phenomenon is explained by the fear factor from Behavioral Finance, which is the way how the equity market has been functioning for centuries. As seen from the chart below, during downtrends, the ATR indicator tends to post higher volatility than during up trend.

ATR

Thus, we can use ATR as a stop loss indicator for order exit and to protect downside risk (so called Chandelier Exit). The chandelier exit places a trailing stop under the highest high the stock reached since you entered the trade. The distance between the highest high and the stop level is defined as some multiple of the ATR. For example,

  • if you buy a stock at $100 and the ATR value is 3, you may place a stop loss at 2 x ATR below the entry price. i.e. 100-2*3 = $94,
  • if you short-sell a stock at $100 and the ATR is 3, you may place a stop loss at 2 x ATR above the entry price. i.e. 100+2*3 = $106
  • when the market price continues to rise (drop for a sell order), we will revise the stop loss level upward (downward) with the newly calculated ATR multiple


Trading Logic:

Suppose we use RSI(14) for order entry (refer to Technical Indicator - Relative Strength Index (RSI)) and ATR(14) for order exit, and apply this strategy to the daily closing price of Nikkei 225 Index CFD (i.e. JPXJPY) over the year of 2020.

  • Based on a sliding window approach to collect the previous 14 closing price
  • Calculate the latest ATR and RSI values
  • Order open conditions:
    • if we have NO outstanding position,
      • if RSI value < 30, we open a buy order with stoploss level set to current price - 2*ATR
      • if RSI value > 70, we open a sell order with stoploss level set to current price + 2*ATR
  • Order close conditions:
    • if we have outstanding position,
      • if we previously submit a buy order, update the stop-loss level to the maximum of the previous stop-loss and (current price - 2*ATR)
      • if we previously submit a sell order, update the stop-loss level to the minimum of the previous stop-loss and (current price + 2*ATR)
  • Repeat the process until the backtest period end

Now, let's write down our trading algorithm step-by-step.


Step 1. Calculate RSI and ATR value

First of all, we define 'RSI_period', 'ATR_period' and 'instrument' at initialization.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
from AlgoAPI import AlgoAPIUtil, AlgoAPI_Backtest
from datetime import datetime, timedelta
from talib import RSI, ATR
import numpy as np

class AlgoEvent:
    def __init__(self):
        self.timer = datetime(1970,1,1)
        self.RSI_period = 14
        self.ATR_period = 14
        self.instrument = "JPXJPY"

We use the API function getHistoricalBar to collect historical observations. Then, we apply a python library 'talib.RSI', 'talib.ATR' to calculate the latest RSI and ATR value.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
    def on_marketdatafeed(self, md, ab):
        # execute stratey every 24 hours
        if md.timestamp >= self.timer+timedelta(hours=24):
            # get last 14 closing price
            res = self.evt.getHistoricalBar({"instrument":self.instrument}, self.RSI_period+1, "D")
            arrClose = np.array([res[t]['c'] for t in res])
            arrHigh = np.array([res[t]['h'] for t in res])
            arrLow = np.array([res[t]['l'] for t in res])
            
            # calculate the current RSI value
            RSI_cur = RSI(arrClose, self.RSI_period)[-1]

            # calculate the current ATR value
            ATR_cur = ATR(arrHigh, arrLow, arrClose, self.ATR_period)[-1]
            
            # print out RSI and ATR value to console
            self.evt.consoleLog("RSI_cur, ATR_cur = ", RSI_cur, ATR_cur)

            # update timer
            self.timer = md.timestamp


Step 2. Order Entry Conditions

We further define 'rsi_overbought', 'rsi_oversold' and 'ATR_multiple' at initialization.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
from AlgoAPI import AlgoAPIUtil, AlgoAPI_Backtest
from datetime import datetime, timedelta
from talib import RSI, ATR
import numpy as np

class AlgoEvent:
    def __init__(self):
        self.timer = datetime(1970,1,1)
        self.timer = datetime(1970,1,1)
        self.RSI_period = 14
        self.ATR_period = 14
        self.instrument = "JPXJPY"
        self.rsi_overbought = 70
        self.rsi_oversold = 30
        self.ATR_multiple = 2

We also create a function 'open_order' to handle order submissions.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
    def open_order(self, buysell, current_price, ATR):

        # set stop loss level to be 2*ATR below current price for buy order
        if buysell==1:
            sl = current_price - self.ATR_multiple*ATR

        # set stop loss level to be 2*ATR above current price for buy order
        else:
            sl = current_price + self.ATR_multiple*ATR

        # create order object
        order = AlgoAPIUtil.OrderObject(
            instrument = self.instrument,
            openclose = 'open',
            buysell = buysell,    #1=buy, -1=sell
            ordertype = 0,        #0=market, 1=limit
            volume = 0.01,
            stopLossLevel = sl
        )

        # send order to server
        self.evt.sendOrder(order)

Now, we update the system's function 'on_marketdatafeed' for order open logic.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
    def on_marketdatafeed(self, md, ab):
        # execute stratey every 24 hours
        if md.timestamp >= self.timer+timedelta(hours=24):

            # get last 14 closing price
            res = self.evt.getHistoricalBar({"instrument":self.instrument}, self.RSI_period+1, "D")
            arrClose = np.array([res[t]['c'] for t in res])
            arrHigh = np.array([res[t]['h'] for t in res])
            arrLow = np.array([res[t]['l'] for t in res])
            
            # calculate the current RSI value
            RSI_cur = RSI(arrClose, self.RSI_period)[-1]

            # calculate the current ATR value
            ATR_cur = ATR(arrHigh, arrLow, arrClose, self.ATR_period)[-1]
            
            # print out RSI and ATR value to console
            self.evt.consoleLog("RSI_cur, ATR_cur = ", RSI_cur, ATR_cur)


            # get outstanding position
            pos, osOrder, pendOrder = self.evt.getSystemOrders()

            # open an order if we have no outstanding position
            if pos[self.instrument]["netVolume"]==0:

                # open a sell order if it is overbought
                if RSI_cur>self.rsi_overbought:
                    self.open_order(-1, md.lastPrice, ATR_cur)
                    
                # open a buy order if it is oversold
                elif RSI_cur<self.rsi_oversold:
                    self.open_order(1, md.lastPrice, ATR_cur)


            # update timer
            self.timer = md.timestamp

Step 3. Order Close Condition

For order closing logic, we recalculate and update the new stop loss level in 'on_marketdatafeed' function. (refer to line 36 - 49)

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
    def on_marketdatafeed(self, md, ab):
        # execute stratey every 24 hours
        if md.timestamp >= self.timer+timedelta(hours=24):

            # get last 14 closing price
            res = self.evt.getHistoricalBar({"instrument":self.instrument}, self.RSI_period+1, "D")
            arrClose = np.array([res[t]['c'] for t in res])
            arrHigh = np.array([res[t]['h'] for t in res])
            arrLow = np.array([res[t]['l'] for t in res])
            
            # calculate the current RSI value
            RSI_cur = RSI(arrClose, self.RSI_period)[-1]

            # calculate the current ATR value
            ATR_cur = ATR(arrHigh, arrLow, arrClose, self.ATR_period)[-1]
            
            # print out RSI and ATR value to console
            self.evt.consoleLog("RSI_cur, ATR_cur = ", RSI_cur, ATR_cur)


            # get outstanding position
            pos, osOrder, pendOrder = self.evt.getSystemOrders()

            # open an order if we have no outstanding position
            if pos[self.instrument]["netVolume"]==0:

                # open a sell order if it is overbought
                if RSI_cur>self.rsi_overbought:
                    self.open_order(-1, md.lastPrice, ATR_cur)
                    
                # open a buy order if it is oversold
                elif RSI_cur<self.rsi_oversold:
                    self.open_order(1, md.lastPrice, ATR_cur)

            else:
                # calculate the new ATR trailing stop loss level
                for tradeID in list(osOrder):
                    buysell = osOrder[tradeID]["buysell"]
                    sl = osOrder[tradeID]["stopLossLevel"]
                    
                    if buysell==1:
                        sl = max(sl, md.lastPrice-self.ATR_multiple*ATR_cur)


                    elif buysell==-1:
                        sl = min(sl, md.lastPrice+self.ATR_multiple*ATR_cur)

                    # update to new stop loss level
                    self.evt.update_opened_order(tradeID=tradeID, sl=sl)


            # update timer
            self.timer = md.timestamp


Full Source Code

Combining all above, the full script is presented below.

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
from AlgoAPI import AlgoAPIUtil, AlgoAPI_Backtest
from datetime import datetime, timedelta
from talib import RSI, ATR
import numpy as np

class AlgoEvent:
    def __init__(self):
        self.timer = datetime(1970,1,1)
        self.timer = datetime(1970,1,1)
        self.RSI_period = 14
        self.ATR_period = 14
        self.instrument = "JPXJPY"
        self.rsi_overbought = 70
        self.rsi_oversold = 30
        self.ATR_multiple = 2

    def start(self, mEvt):
        self.evt = AlgoAPI_Backtest.AlgoEvtHandler(self, mEvt)
        self.evt.start()


    def on_marketdatafeed(self, md, ab):
        # execute stratey every 24 hours
        if md.timestamp >= self.timer+timedelta(hours=24):

            # get last 14 closing price
            res = self.evt.getHistoricalBar({"instrument":self.instrument}, self.RSI_period+1, "D")
            arrClose = np.array([res[t]['c'] for t in res])
            arrHigh = np.array([res[t]['h'] for t in res])
            arrLow = np.array([res[t]['l'] for t in res])
            
            # calculate the current RSI value
            RSI_cur = RSI(arrClose, self.RSI_period)[-1]

            # calculate the current ATR value
            ATR_cur = ATR(arrHigh, arrLow, arrClose, self.ATR_period)[-1]
            
            # print out RSI and ATR value to console
            self.evt.consoleLog("RSI_cur, ATR_cur = ", RSI_cur, ATR_cur)


            # get outstanding position
            pos, osOrder, pendOrder = self.evt.getSystemOrders()

            # open an order if we have no outstanding position
            if pos[self.instrument]["netVolume"]==0:

                # open a sell order if it is overbought
                if RSI_cur>self.rsi_overbought:
                    self.open_order(-1, md.lastPrice, ATR_cur)
                    
                # open a buy order if it is oversold
                elif RSI_cur<self.rsi_oversold:
                    self.open_order(1, md.lastPrice, ATR_cur)

            else:
                # calculate the new ATR trailing stop loss level
                for tradeID in list(osOrder):
                    buysell = osOrder[tradeID]["buysell"]
                    sl = osOrder[tradeID]["stopLossLevel"]
                    
                    if buysell==1:
                        sl = max(sl, md.lastPrice-self.ATR_multiple*ATR_cur)


                    elif buysell==-1:
                        sl = min(sl, md.lastPrice+self.ATR_multiple*ATR_cur)

                    # update to new stop loss level
                    self.evt.update_opened_order(tradeID=tradeID, sl=sl)


            # update timer
            self.timer = md.timestamp


    def open_order(self, buysell, current_price, ATR):

        # set stop loss level to be 2*ATR below current price for buy order
        if buysell==1:
            sl = current_price - self.ATR_multiple*ATR

        # set stop loss level to be 2*ATR above current price for buy order
        else:
            sl = current_price + self.ATR_multiple*ATR

        # create order object
        order = AlgoAPIUtil.OrderObject(
            instrument = self.instrument,
            openclose = 'open',
            buysell = buysell,    #1=buy, -1=sell
            ordertype = 0,        #0=market, 1=limit
            volume = 0.01,
            stopLossLevel = sl
        )

        # send order to server
        self.evt.sendOrder(order)


    def on_bulkdatafeed(self, isSync, bd, ab):
        pass

    def on_newsdatafeed(self, nd):
        pass

    def on_weatherdatafeed(self, wd):
        pass
    
    def on_econsdatafeed(self, ed):
        pass
        
    def on_orderfeed(self, of):
        pass
            
    def on_dailyPLfeed(self, pl):
        pass

    def on_openPositionfeed(self, op, oo, uo):
        pass

Results

Now, we are prepared to backtest this strategy.

Backtest Settings:

  • Instrument: JPXJPY
  • Period: 2020.01 - 2020.12
  • Initial Capital: US$10,000
  • Data Interval: 1-day bar
  • Leverage: 100
  • Allow Shortsell: True

Backtest Result

result

Final Thoughts

Some highlights for applying ATR in a trading strategy as below:

  • ATR is mainly used to determine order exit condition, but not for order entry
  • Multi-period ATRs (eg. 5 and 14) could be used to compare and have a sense about the recent "market volatility"
  • The rule of thumb for ATR period is 14
    • Is there any better choice?
    • Can this value be dynamically changing?
    • Will it be generally applicable to other financial instruments/ asset classes?


Like and follow me

If you find my articles inspiring, like this post and follow me here to receive my latest updates.

Enter my promote code "AjpQDMOSmzG2" for any purchase on ALGOGENE, you will automatically get 5% discount.


1 0
Posted on : 2021-11-14 11:42:18.539000



Categories