Cette stratégie génère des signaux de trading en sélectionnant des indicateurs de tendance rapide et lente et en allant long lorsque la tendance rapide traverse la tendance lente, et en allant court lorsque la tendance rapide traverse en dessous de la tendance lente.
Le noyau de la stratégie est la sélection et la combinaison d'indicateurs de tendance rapide et lente:
FastTrend = User selected fast trend indicator
SlowTrend = User selected slow trend indicator
La tendance rapide comprend les algorithmes de tendance SMA, EMA, KAMA et 20+. La tendance lente peut également être librement sélectionnée.
Les signaux de trading sont générés en jugeant la relation entre les tendances rapides et lentes:
if FastTrend > SlowTrend:
Go long
if FastTrend < SlowTrend:
Close position
Le signal long est déclenché lorsque la tendance rapide traverse la tendance lente. Le signal court est déclenché lorsque la tendance rapide traverse la tendance lente.
La stratégie peut être améliorée dans les domaines suivants:
Ajustez les tendances et les paramètres rapides/lents pour trouver des combinaisons optimales.
Ajoutez des filtres comme le volume pour éviter de faux signaux pendant la volatilité du marché.
Incorporer des stratégies de stop-loss comme le stop-loss de trailing pour contrôler les pertes d'une seule transaction.
Combinez avec d'autres indicateurs comme MACD, KDJ pour améliorer la stabilité.
Optimisez le timing d'entrée, ne comptez pas uniquement sur le croisement des tendances.
La stratégie multi-trend crossover identifie les changements de tendance à travers les délais en combinant des tendances rapides et lentes. Mais elle est sensible aux fluctuations du marché et ne fonctionne bien que sur des marchés à tendance évidente. Nous avons besoin de méthodes comme l'optimisation des paramètres et la gestion des risques pour améliorer la stabilité et la rentabilité de la stratégie.
Je ne sais pas.
/*backtest
start: 2023-08-21 00:00:00
end: 2023-09-20 00:00:00
period: 3h
basePeriod: 15m
exchanges: [{"eid":"Futures_Binance","currency":"BTC_USDT"}]
*/
// This source code is subject to the terms of the Mozilla Public License 2.0 at https://mozilla.org/MPL/2.0/
// @version=5
// Author = TradeAutomation
strategy(title="Multi Trend Cross Strategy Template", shorttitle="Multi Trend Cross Strategy", process_orders_on_close=true, overlay=true, commission_type=strategy.commission.cash_per_contract, commission_value=0.0035, initial_capital = 1000000, default_qty_type=strategy.percent_of_equity, default_qty_value=100)
// Backtest Date Range Inputs //
StartTime = input(defval=timestamp('01 Jan 2000 05:00 +0000'), group="Date Range", title='Start Time')
EndTime = input(defval=timestamp('01 Jan 2099 00:00 +0000'), group="Date Range", title='End Time')
InDateRange = true
// Trend Selector //
TrendSelectorInput = input.string(title="Fast Trend Selector", defval="EMA", group="Core Settings", options=["ALMA", "DEMA", "DSMA", "EMA", "HMA", "JMA", "KAMA", "Linear Regression (LSMA)", "RMA", "SMA", "SMMA", "Price Source", "TEMA", "TMA", "VAMA", "VIDYA", "VMA", "VWMA", "WMA", "WWMA", "ZLEMA"], tooltip="Select your fast trend")
TrendSelectorInput2 = input.string(title="Slow Trend Selector", defval="EMA", group="Core Settings", options=["ALMA", "DEMA", "DSMA", "EMA", "HMA", "JMA", "KAMA", "Linear Regression (LSMA)", "RMA", "SMA", "SMMA", "Price Source", "TEMA", "TMA", "VAMA", "VIDYA", "VMA", "VWMA", "WMA", "WWMA", "ZLEMA"], tooltip="Select your slow trend")
src = input.source(close, "Price Source", group="Core Settings", tooltip="This is the price source being used for the trends to calculate based on")
length = input.int(10, "Fast Trend Length", group="Core Settings", step=5, tooltip="A long is entered when the selected fast trend crosses over the selected slow trend")
length2 = input.int(200, "Slow Trend Length", group="Core Settings", step=5, tooltip="A long is entered when the selected fast trend crosses over the selected slow trend")
LineWidth = input.int(1, "Line Width", group="Core Settings", tooltip="This is the width of the line plotted that represents the selected trend")
// Individual Moving Average / Regression Setting //
AlmaOffset = input.float(0.85, "ALMA Offset", group="Individual Trend Settings", tooltip="This only applies when ALMA is selected")
AlmaSigma = input.float(6, "ALMA Sigma", group="Individual Trend Settings", tooltip="This only applies when ALMA is selected")
ATRFactor = input.float(3, "ATR Multiplier For SuperTrend", group="Individual Trend Settings", tooltip="This only applies when SuperTrend is selected")
ATRLength = input.int(12, "ATR Length For SuperTrend", group="Individual Trend Settings", tooltip="This only applies when SuperTrend is selected")
ssfLength = input.int(20, "DSMA Super Smoother Filter Length", minval=1, tooltip="This only applies when EDSMA is selected", group="Individual Trend Settings")
ssfPoles = input.int(2, "DSMA Super Smoother Filter Poles", options=[2, 3], tooltip="This only applies when EDSMA is selected", group="Individual Trend Settings")
JMApower = input.int(2, "JMA Power Parameter", group="Individual Trend Settings", tooltip="This only applies when JMA is selected")
phase = input.int(-45, title="JMA Phase Parameter", step=10, minval=-110, maxval=110, group="Individual Trend Settings", tooltip="This only applies when JMA is selected")
KamaAlpha = input.float(3, "KAMA's Alpha", minval=1,step=0.5, group="Individual Trend Settings", tooltip="This only applies when KAMA is selected")
LinRegOffset = input.int(0, "Linear Regression Offset", group="Individual Trend Settings", tooltip="This only applies when Linear Regression is selected")
VAMALookback =input.int(12, "VAMA Volatility lookback", group="Individual Trend Settings", tooltip="This only applies when VAMA is selected")
// Trend Indicators With Library Functions //
ALMA = ta.alma(src, length, AlmaOffset, AlmaSigma)
EMA = ta.ema(src, length)
HMA = ta.hma(src, length)
LinReg = ta.linreg(src, length, LinRegOffset)
RMA = ta.rma(src, length)
SMA = ta.sma(src, length)
VWMA = ta.vwma(src, length)
WMA = ta.wma(src, length)
ALMA2 = ta.alma(src, length2, AlmaOffset, AlmaSigma)
EMA2 = ta.ema(src, length2)
HMA2 = ta.hma(src, length2)
LinReg2 = ta.linreg(src, length2, LinRegOffset)
RMA2 = ta.rma(src, length2)
SMA2 = ta.sma(src, length2)
VWMA2 = ta.vwma(src, length2)
WMA2 = ta.wma(src, length2)
// Additional Trend Indicators Built In And/Or Open Sourced //
//DEMA
de1 = ta.ema(src, length)
de2 = ta.ema(de1, length)
DEMA = 2 * de1 - de2
de3 = ta.ema(src, length2)
de4 = ta.ema(de3, length2)
DEMA2 = 2 * de3 - de4
// Ehlers Deviation-Scaled Moving Average - DSMA [Everget]
PI = 2 * math.asin(1)
get2PoleSSF(src, length) =>
arg = math.sqrt(2) * PI / length
a1 = math.exp(-arg)
b1 = 2 * a1 * math.cos(arg)
c2 = b1
c3 = -math.pow(a1, 2)
c1 = 1 - c2 - c3
var ssf = 0.0
ssf := c1 * src + c2 * nz(ssf[1]) + c3 * nz(ssf[2])
get3PoleSSF(src, length) =>
arg = PI / length
a1 = math.exp(-arg)
b1 = 2 * a1 * math.cos(1.738 * arg)
c1 = math.pow(a1, 2)
coef2 = b1 + c1
coef3 = -(c1 + b1 * c1)
coef4 = math.pow(c1, 2)
coef1 = 1 - coef2 - coef3 - coef4
var ssf = 0.0
ssf := coef1 * src + coef2 * nz(ssf[1]) + coef3 * nz(ssf[2]) + coef4 * nz(ssf[3])
zeros = src - nz(src[2])
avgZeros = (zeros + zeros[1]) / 2
// Ehlers Super Smoother Filter
ssf = ssfPoles == 2
? get2PoleSSF(avgZeros, ssfLength)
: get3PoleSSF(avgZeros, ssfLength)
// Rescale filter in terms of Standard Deviations
stdev = ta.stdev(ssf, length)
scaledFilter = stdev != 0
? ssf / stdev
: 0
alpha1 = 5 * math.abs(scaledFilter) / length
EDSMA = 0.0
EDSMA := alpha1 * src + (1 - alpha1) * nz(EDSMA[1])
get2PoleSSF2(src, length2) =>
arg = math.sqrt(2) * PI / length2
a1 = math.exp(-arg)
b1 = 2 * a1 * math.cos(arg)
c2 = b1
c3 = -math.pow(a1, 2)
c1 = 1 - c2 - c3
var ssf2 = 0.0
ssf2 := c1 * src + c2 * nz(ssf2[1]) + c3 * nz(ssf2[2])
get3PoleSSF2(src, length2) =>
arg = PI / length2
a1 = math.exp(-arg)
b1 = 2 * a1 * math.cos(1.738 * arg)
c1 = math.pow(a1, 2)
coef2 = b1 + c1
coef3 = -(c1 + b1 * c1)
coef4 = math.pow(c1, 2)
coef1 = 1 - coef2 - coef3 - coef4
var ssf2 = 0.0
ssf2 := coef1 * src + coef2 * nz(ssf2[1]) + coef3 * nz(ssf2[2]) + coef4 * nz(ssf2[3])
// Ehlers Super Smoother Filter
ssf2 = ssfPoles == 2
? get2PoleSSF2(avgZeros, ssfLength)
: get3PoleSSF2(avgZeros, ssfLength)
// Rescale filter in terms of Standard Deviations
stdev2 = ta.stdev(ssf2, length2)
scaledFilter2 = stdev2 != 0
? ssf2 / stdev2
: 0
alpha12 = 5 * math.abs(scaledFilter2) / length2
EDSMA2 = 0.0
EDSMA2 := alpha12 * src + (1 - alpha12) * nz(EDSMA2[1])
//JMA [Everget]
phaseRatio = phase < -100 ? 0.5 : phase > 100 ? 2.5 : phase / 100 + 1.5
beta = 0.45 * (length - 1) / (0.45 * (length - 1) + 2)
alpha = math.pow(beta, JMApower)
var JMA = 0.0
var e0 = 0.0
e0 := (1 - alpha) * src + alpha * nz(e0[1])
var e1 = 0.0
e1 := (src - e0) * (1 - beta) + beta * nz(e1[1])
var e2 = 0.0
e2 := (e0 + phaseRatio * e1 - nz(JMA[1])) * math.pow(1 - alpha, 2) + math.pow(alpha, 2) * nz(e2[1])
JMA := e2 + nz(JMA[1])
beta2 = 0.45 * (length2 - 1) / (0.45 * (length2 - 1) + 2)
alpha2 = math.pow(beta2, JMApower)
var JMA2 = 0.0
var e02 = 0.0
e02 := (1 - alpha2) * src + alpha2 * nz(e02[1])
var e12 = 0.0
e12 := (src - e02) * (1 - beta2) + beta2 * nz(e12[1])
var e22 = 0.0
e22 := (e02 + phaseRatio * e12 - nz(JMA2[1])) * math.pow(1 - alpha2, 2) + math.pow(alpha2, 2) * nz(e22[1])
JMA2 := e22 + nz(JMA2[1])
//KAMA [Everget]
var KAMA = 0.0
fastAlpha = 2.0 / (KamaAlpha + 1)
slowAlpha = 2.0 / 31
momentum = math.abs(ta.change(src, length))
volatility = math.sum(math.abs(ta.change(src)), length)
efficiencyRatio = volatility != 0 ? momentum / volatility : 0
smoothingConstant = math.pow((efficiencyRatio * (fastAlpha - slowAlpha)) + slowAlpha, 2)
KAMA := nz(KAMA[1], src) + smoothingConstant * (src - nz(KAMA[1], src))
var KAMA2 = 0.0
momentum2 = math.abs(ta.change(src, length2))
volatility2 = math.sum(math.abs(ta.change(src)), length2)
efficiencyRatio2 = volatility2 != 0 ? momentum2 / volatility2 : 0
smoothingConstant2 = math.pow((efficiencyRatio2 * (fastAlpha - slowAlpha)) + slowAlpha, 2)
KAMA2 := nz(KAMA2[1], src) + smoothingConstant2 * (src - nz(KAMA2[1], src))
//SMMA
var SMMA = 0.0
SMMA := na(SMMA[1]) ? ta.sma(src, length) : (SMMA[1] * (length - 1) + src) / length
var SMMA2 = 0.0
SMMA2 := na(SMMA2[1]) ? ta.sma(src, length2) : (SMMA2[1] * (length2 - 1) + src) / length2
//TEMA
t1 = ta.ema(src, length)
t2 = ta.ema(t1, length)
t3 = ta.ema(t2, length)
TEMA = 3 * (t1 - t2) + t3
t12 = ta.ema(src, length2)
t22 = ta.ema(t12, length2)
t32 = ta.ema(t22, length2)
TEMA2 = 3 * (t12 - t22) + t32
//TMA
TMA = ta.sma(ta.sma(src, math.ceil(length / 2)), math.floor(length / 2) + 1)
TMA2 = ta.sma(ta.sma(src, math.ceil(length2 / 2)), math.floor(length2 / 2) + 1)
//VAMA [Duyck]
mid=ta.ema(src,length)
dev=src-mid
vol_up=ta.highest(dev,VAMALookback)
vol_down=ta.lowest(dev,VAMALookback)
VAMA = mid+math.avg(vol_up,vol_down)
mid2=ta.ema(src,length2)
dev2=src-mid2
vol_up2=ta.highest(dev2,VAMALookback)
vol_down2=ta.lowest(dev2,VAMALookback)
VAMA2 = mid2+math.avg(vol_up2,vol_down2)
//VIDYA [KivancOzbilgic]
var VIDYA=0.0
VMAalpha=2/(length+1)
ud1=src>src[1] ? src-src[1] : 0
dd1=src<src[1] ? src[1]-src : 0
UD=math.sum(ud1,9)
DD=math.sum(dd1,9)
CMO=nz((UD-DD)/(UD+DD))
VIDYA := na(VIDYA[1]) ? ta.sma(src, length) : nz(VMAalpha*math.abs(CMO)*src)+(1-VMAalpha*math.abs(CMO))*nz(VIDYA[1])
var VIDYA2=0.0
VMAalpha2=2/(length2+1)
ud12=src>src[1] ? src-src[1] : 0
dd12=src<src[1] ? src[1]-src : 0
UD2=math.sum(ud12,9)
DD2=math.sum(dd12,9)
CMO2=nz((UD2-DD2)/(UD2+DD2))
VIDYA2 := na(VIDYA2[1]) ? ta.sma(src, length2) : nz(VMAalpha2*math.abs(CMO2)*src)+(1-VMAalpha2*math.abs(CMO2))*nz(VIDYA2[1])
//VMA [LazyBear]
sc = 1/length
pdm = math.max((src - src[1]), 0)
mdm = math.max((src[1] - src), 0)
var pdmS = 0.0
var mdmS = 0.0
pdmS := ((1 - sc)*nz(pdmS[1]) + sc*pdm)
mdmS := ((1 - sc)*nz(mdmS[1]) + sc*mdm)
s = pdmS + mdmS
pdi = pdmS/s
mdi = mdmS/s
var pdiS = 0.0
var mdiS = 0.0
pdiS := ((1 - sc)*nz(pdiS[1]) + sc*pdi)
mdiS := ((1 - sc)*nz(mdiS[1]) + sc*mdi)
d = math.abs(pdiS - mdiS)
s1 = pdiS + mdiS
var iS = 0.0
iS := ((1 - sc)*nz(iS[1]) + sc*d/s1)
hhv = ta.highest(iS, length)
llv = ta.lowest(iS, length)
d1 = hhv - llv
vi = (iS - llv)/d1
var VMA=0.0
VMA := na(VMA[1]) ? ta.sma(src, length) : sc*vi*src + (1 - sc*vi)*nz(VMA[1])
sc2 = 1/length2
pdm2 = math.max((src - src[1]), 0)
mdm2 = math.max((src[1] - src), 0)
var pdmS2 = 0.0
var mdmS2 = 0.0
pdmS2 := ((1 - sc2)*nz(pdmS2[1]) + sc2*pdm2)
mdmS2 := ((1 - sc2)*nz(mdmS2[1]) + sc2*mdm2)
s2 = pdmS2 + mdmS2
pdi2 = pdmS2/s2
mdi2 = mdmS2/s2
var pdiS2 = 0.0
var mdiS2 = 0.0
pdiS2 := ((1 - sc2)*nz(pdiS2[1]) + sc2*pdi2)
mdiS2 := ((1 - sc2)*nz(mdiS2[1]) + sc2*mdi2)
d2 = math.abs(pdiS2 - mdiS2)
s12 = pdiS2 + mdiS2
var iS2 = 0.0
iS2 := ((1 - sc2)*nz(iS2[1]) + sc2*d2/s12)
hhv2 = ta.highest(iS2, length)
llv2 = ta.lowest(iS2, length)
d12 = hhv2 - llv2
vi2 = (iS2 - llv2)/d12
var VMA2=0.0
VMA2 := na(VMA2[1]) ? ta.sma(src, length2) : sc2*vi2*src + (1 - sc2*vi2)*nz(VMA2[1])
//WWMA
var WWMA=0.0
WWMA := (1/length)*src + (1-(1/length))*nz(WWMA[1])
var WWMA2=0.0
WWMA2 := (1/length2)*src + (1-(1/length2))*nz(WWMA2[1])
//Zero Lag EMA [KivancOzbilgic]
EMA1a = ta.ema(src,length)
EMA2a = ta.ema(EMA1a,length)
Diff = EMA1a - EMA2a
ZLEMA = EMA1a + Diff
EMA12 = ta.ema(src,length2)
EMA22 = ta.ema(EMA12,length2)
Diff2 = EMA12 - EMA22
ZLEMA2 = EMA12 + Diff2
// Trend Mapping and Plotting //
FastTrend = TrendSelectorInput == "ALMA" ? ALMA : TrendSelectorInput == "DEMA" ? DEMA : TrendSelectorInput == "DSMA" ? EDSMA : TrendSelectorInput == "EMA" ? EMA : TrendSelectorInput == "HMA" ? HMA : TrendSelectorInput == "JMA" ? JMA : TrendSelectorInput == "KAMA" ? KAMA : TrendSelectorInput == "Linear Regression (LSMA)" ? LinReg : TrendSelectorInput == "RMA" ? RMA : TrendSelectorInput == "SMA" ? SMA : TrendSelectorInput == "SMMA" ? SMMA : TrendSelectorInput == "Price Source" ? src : TrendSelectorInput == "TEMA" ? TEMA : TrendSelectorInput == "TMA" ? TMA : TrendSelectorInput == "VAMA" ? VAMA : TrendSelectorInput == "VIDYA" ? VIDYA : TrendSelectorInput == "VMA" ? VMA : TrendSelectorInput == "VWMA" ? VWMA : TrendSelectorInput == "WMA" ? WMA : TrendSelectorInput == "WWMA" ? WWMA : TrendSelectorInput == "ZLEMA" ? ZLEMA : SMA
SlowTrend = TrendSelectorInput2 == "ALMA" ? ALMA2 : TrendSelectorInput2 == "DEMA" ? DEMA2 : TrendSelectorInput2 == "DSMA" ? EDSMA2 : TrendSelectorInput2 == "EMA" ? EMA2 : TrendSelectorInput2 == "HMA" ? HMA2 : TrendSelectorInput2 == "JMA" ? JMA2 : TrendSelectorInput2 == "KAMA" ? KAMA2 : TrendSelectorInput2 == "Linear Regression (LSMA)" ? LinReg2 : TrendSelectorInput2 == "RMA" ? RMA2 : TrendSelectorInput2 == "SMA" ? SMA2 : TrendSelectorInput2 == "SMMA" ? SMMA2 : TrendSelectorInput2 == "Price Source" ? src : TrendSelectorInput2 == "TEMA" ? TEMA2 : TrendSelectorInput2 == "TMA" ? TMA2 : TrendSelectorInput2 == "VAMA" ? VAMA2 : TrendSelectorInput2 == "VIDYA" ? VIDYA2 : TrendSelectorInput2 == "VMA" ? VMA2 : TrendSelectorInput2 == "VWMA" ? VWMA2 : TrendSelectorInput2 == "WMA" ? WMA2 : TrendSelectorInput2 == "WWMA" ? WWMA2 : TrendSelectorInput2 == "ZLEMA" ? ZLEMA2 : SMA2
plot(FastTrend, color=color.green, linewidth=LineWidth)
plot(SlowTrend, color=color.red, linewidth=LineWidth)
//Short & Long Options
Long = input.bool(true, "Model Long Trades", group="Core Settings")
Short = input.bool(false, "Model Short Trades", group="Core Settings")
// Entry & Exit Functions //
if (InDateRange and Long==true and FastTrend>SlowTrend)
strategy.entry("Long", strategy.long, alert_message="Long")
if (InDateRange and Long==true and FastTrend<SlowTrend)
strategy.close("Long", alert_message="Close Long")
if (InDateRange and Short==true and FastTrend<SlowTrend)
strategy.entry("Short", strategy.short, alert_message="Short")
if (InDateRange and Short==true and FastTrend>SlowTrend)
strategy.close("Short", alert_message="Cover Short")
if (not InDateRange)
strategy.close_all(alert_message="End of Date Range")