Donchian Channel Breakout Trading Strategy

Author: ChaoZhang, Date: 2023-11-08 12:31:56
Tags:

Overview

The Donchian channel breakout trading strategy judges current price trends by calculating the channel of highest and lowest prices over a certain period and trades long and short based on channel breakouts. This strategy is suitable for highly volatile stocks and cryptocurrencies.

Strategy Logic

This strategy constructs a channel by calculating the highest price pcmax and lowest price pcmin over the last history periods. The calculation methods for the upper and lower rail of the channel are:

Upper rail yh = pcmax - (pcmax - pcmin) * (100 - percentDev)/100

Lower rail yl = pcmin + (pcmax - pcmin) * percentDev/100

where percentDev defaults to 13.

A long signal is generated when the price breaks through the upper rail. A short signal is generated when the price breaks through the lower rail.

The specific logic to generate trading signals is:

boundup = high > yh to determine if the upper rail is broken
bounddn = low < yl to determine if the lower rail is broken
upsign = sma(bounddn, 2) == 1 uses sma of bounddn to determine persistent breakout of lower rail
dnsign = sma(boundup, 2) == 1 uses sma of boundup to determine persistent breakout of upper rail
exitup = dnsign breakout of upper rail generates exit signal
exitdn = upsign breakout of lower rail generates exit signal
if upsign breakout of lower rail generates long signal
if dnsign breakout of upper rail generates short signal

The strategy also sets start and end trading times to avoid unnecessary overnight positions.

Advantages of the Strategy

Uses Donchian channel to determine trends, good backtest results
Has both long and short signals, allows two-way trading
Uses SMA to filter signals and avoid bad trades
Optional stop loss to control risks
Sets start and end trading times to avoid overnight risks

Risks of the Strategy

Sensitive to history and percentDev parameters, needs optimization for different products
May generate false signals in range-bound markets
Does not consider order management, may impact profitability in live trading
Does not consider position sizing, risks of oversized positions
Does not consider money management, needs reasonable trading capital

Enhancement Ideas

Optimize history and percentDev parameters for different products
Add filters to avoid false signals in ranging markets
Add position sizing module to control single position size
Add money management module to limit total position size
Add order management for optimal order execution

Conclusion

The Donchian channel breakout strategy uses channel breakouts to determine trends and trading signals, with good backtest results and ability to trade both long and short. However, risks exist regarding parameter optimization, filters, position sizing, money management, order management etc. Proper enhancements in these areas are needed before stable live trading. Overall, it is a traditional trend following strategy, and with optimizations can become a reliable quantitative trading strategy.

/*backtest
start: 2023-10-31 00:00:00
end: 2023-11-07 00:00:00
period: 1h
basePeriod: 15m
exchanges: [{"eid":"Futures_Binance","currency":"BTC_USDT"}]
*/

////////////////////////////////////////////////////////////
//  Copyright by AlexInc v1.0 02/07/2018  @aav_1980
// PriceChannel strategy
// If you find this script helpful, you can also help me by sending donation to 
// BTC 16d9vgFvCmXpLf8FiKY6zsy6pauaCyFnzS
// LTC LQ5emyqNRjdRMqHPHEqREgryUJqmvYhffM
////////////////////////////////////////////////////////////
//@version=3
strategy("AlexInc PriceChannel Str", overlay=false)
history = input(20)
percentDev = input(13)
capital = input(100)

needlong = input(true, defval = true, title = "Long")
needshort = input(true, defval = true, title = "Short")
usestoploss = input(true, defval = true, title = "Stop Loss")
stoplossmult = input(3.8, defval = 3.8, minval = 1, maxval = 10, title = "Stop loss multiplicator")


fromyear = input(2018, defval = 2018, minval = 1900, maxval = 2100, title = "From Year")
toyear = input(2100, defval = 2100, minval = 1900, maxval = 2100, title = "To Year")
frommonth = input(01, defval = 01, minval = 01, maxval = 12, title = "From Month")
tomonth = input(12, defval = 12, minval = 01, maxval = 12, title = "To Month")
fromday = input(01, defval = 01, minval = 01, maxval = 31, title = "From day")
today = input(31, defval = 31, minval = 01, maxval = 31, title = "To day")

bodymin = min( open, close)
bodymax = max(open, close)

pcmax = highest(bodymax, history)
pcmin = lowest(bodymin, history)

yh = ((pcmax - pcmin) / 100 * (100 - percentDev)) + pcmin
yl = ((pcmax - pcmin) / 100 * percentDev) + pcmin

plot(pcmax)
plot(pcmin)
plot(yh)
plot(yl)

//1
bounddn = low < yl ? 1 : 0
boundup = high > yh ? 1 : 0
upsign = sma(bounddn, 2) == 1
dnsign = sma(boundup, 2) == 1
//2
//upsign = crossover(bodymin, yl)
//dnsign = crossunder(bodymax , yh)


exitup = dnsign
exitdn = upsign

lot = strategy.equity / close * capital / 100


xATR = atr(history)
nLoss = usestoploss ? stoplossmult * xATR : na

stop_level_long = 0.0
stop_level_long := nz(stop_level_long[1])

stop_level_short = 0.0
stop_level_short := nz(stop_level_short[1])

pos = strategy.position_size
if pos >0 and pos[1] <= 0 //crossover(pos, 0.5)
    stop_level_long = strategy.position_avg_price - nLoss
if pos < 0 and pos[1] >= 0 //crossunder(pos, -0.5)
    stop_level_short = strategy.position_avg_price + nLoss
if pos == 0    
    stop_level_long = bodymin - nLoss
    stop_level_short = bodymax + nLoss

//plot(bodymax + nLoss, color=red)
//plot(bodymin - nLoss, color=red)
plot(stop_level_long, color=red)
plot(stop_level_short, color=red)

if upsign
    strategy.entry("Long", strategy.long, needlong == false ? 0 : lot)

if dnsign
    strategy.entry("Short", strategy.short, needshort == false ? 0 : na)

if true
    strategy.close_all()


//if strategy.position_size != 0
//    strategy.exit("Exit Long", from_entry = "Long", stop = stop_level_long)
//    strategy.exit("Exit Short", from_entry = "Short", stop = stop_level_short)

More

Forums
|
API
|
Github
|
About
|
SiteMap