Gaussian Channel Trend Following Strategy adalah strategi perdagangan yang mengikuti tren berdasarkan indikator Gaussian Channel. Strategi ini bertujuan untuk menangkap tren utama di pasar, membeli dan memegang posisi selama tren naik dan menutup posisi selama tren turun. Ini menggunakan indikator Gaussian Channel untuk mengidentifikasi arah dan kekuatan tren dengan menganalisis hubungan antara harga dan pita atas dan bawah saluran. Tujuan utama strategi adalah untuk memaksimalkan keuntungan selama tren berkelanjutan sambil meminimalkan frekuensi perdagangan selama pasar yang terikat jangkauan.
Inti dari Gaussian Channel Trend Following Strategy adalah Gaussian Channel indicator, yang diusulkan oleh Ehlers. Ini menggabungkan teknik penyaringan Gaussian dengan True Range untuk menganalisis aktivitas tren. Indikator pertama menghitung nilai beta dan alpha berdasarkan periode pengambilan sampel dan jumlah kutub, kemudian menerapkan filter ke data untuk mendapatkan kurva halus (garis tengah). Selanjutnya, strategi mengalikan True Range halus dengan pengganda untuk menghasilkan saluran atas dan bawah. Ketika harga melintasi di atas / di bawah saluran atas / bawah, itu menghasilkan sinyal beli / jual.
Gaussian Channel Trend Following Strategy adalah strategi trading trend-following yang didasarkan pada teknik penyaringan Gaussian, yang bertujuan untuk menangkap tren pasar utama untuk pengembalian stabil jangka panjang. Strategi ini menggunakan indikator Gaussian Channel untuk mengidentifikasi arah dan kekuatan tren sambil menawarkan fitur untuk mengurangi lag dan memberikan respons cepat. Keuntungan dari strategi terletak pada kemampuan trend-following yang kuat dan frekuensi perdagangan yang rendah. Namun, strategi ini juga menghadapi risiko seperti optimasi parameter, pembalikan tren, dan pasar range-bound. Optimasi masa depan dapat mencakup penggabungan indikator teknis lainnya, optimasi parameter dinamis, menambahkan modul kontrol risiko, dan analisis multi-frame waktu untuk lebih meningkatkan ketahanan dan profitabilitas strategi.
/*backtest start: 2023-03-23 00:00:00 end: 2024-03-28 00:00:00 period: 1d basePeriod: 1h exchanges: [{"eid":"Futures_Binance","currency":"BTC_USDT"}] */ //@version=5 strategy(title="Gaussian Channel Strategy v2.0", overlay=true, calc_on_every_tick=false, initial_capital=1000, default_qty_type=strategy.percent_of_equity, default_qty_value=100, commission_type=strategy.commission.percent, commission_value=0.1, slippage=3) //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Gaussian Channel Indicaor - courtesy of @DonovanWall //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Date condition inputs startDate = input(timestamp("1 January 2018 00:00 +0000"), "Date Start", group="Main Algo Settings") endDate = input(timestamp("1 January 2060 00:00 +0000"), "Date Start", group="Main Algo Settings") timeCondition = true // This study is an experiment utilizing the Ehlers Gaussian Filter technique combined with lag reduction techniques and true range to analyze trend activity. // Gaussian filters, as Ehlers explains it, are simply exponential moving averages applied multiple times. // First, beta and alpha are calculated based on the sampling period and number of poles specified. The maximum number of poles available in this script is 9. // Next, the data being analyzed is given a truncation option for reduced lag, which can be enabled with "Reduced Lag Mode". // Then the alpha and source values are used to calculate the filter and filtered true range of the dataset. // Filtered true range with a specified multiplier is then added to and subtracted from the filter, generating a channel. // Lastly, a one pole filter with a N pole alpha is averaged with the filter to generate a faster filter, which can be enabled with "Fast Response Mode". // Custom bar colors are included. // Note: Both the sampling period and number of poles directly affect how much lag the indicator has, and how smooth the output is. // Larger inputs will result in smoother outputs with increased lag, and smaller inputs will have noisier outputs with reduced lag. // For the best results, I recommend not setting the sampling period any lower than the number of poles + 1. Going lower truncates the equation. //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Updates: // Huge shoutout to @e2e4mfck for taking the time to improve the calculation method! // -> migrated to v4 // -> pi is now calculated using trig identities rather than being explicitly defined. // -> The filter calculations are now organized into functions rather than being individually defined. // -> Revamped color scheme. //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Functions - courtesy of @e2e4mfck //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Filter function f_filt9x (_a, _s, _i) => int _m2 = 0, int _m3 = 0, int _m4 = 0, int _m5 = 0, int _m6 = 0, int _m7 = 0, int _m8 = 0, int _m9 = 0, float _f = .0, _x = (1 - _a) // Weights. // Initial weight _m1 is a pole number and equal to _i _m2 := _i == 9 ? 36 : _i == 8 ? 28 : _i == 7 ? 21 : _i == 6 ? 15 : _i == 5 ? 10 : _i == 4 ? 6 : _i == 3 ? 3 : _i == 2 ? 1 : 0 _m3 := _i == 9 ? 84 : _i == 8 ? 56 : _i == 7 ? 35 : _i == 6 ? 20 : _i == 5 ? 10 : _i == 4 ? 4 : _i == 3 ? 1 : 0 _m4 := _i == 9 ? 126 : _i == 8 ? 70 : _i == 7 ? 35 : _i == 6 ? 15 : _i == 5 ? 5 : _i == 4 ? 1 : 0 _m5 := _i == 9 ? 126 : _i == 8 ? 56 : _i == 7 ? 21 : _i == 6 ? 6 : _i == 5 ? 1 : 0 _m6 := _i == 9 ? 84 : _i == 8 ? 28 : _i == 7 ? 7 : _i == 6 ? 1 : 0 _m7 := _i == 9 ? 36 : _i == 8 ? 8 : _i == 7 ? 1 : 0 _m8 := _i == 9 ? 9 : _i == 8 ? 1 : 0 _m9 := _i == 9 ? 1 : 0 // filter _f := math.pow(_a, _i) * nz(_s) + _i * _x * nz(_f[1]) - (_i >= 2 ? _m2 * math.pow(_x, 2) * nz(_f[2]) : 0) + (_i >= 3 ? _m3 * math.pow(_x, 3) * nz(_f[3]) : 0) - (_i >= 4 ? _m4 * math.pow(_x, 4) * nz(_f[4]) : 0) + (_i >= 5 ? _m5 * math.pow(_x, 5) * nz(_f[5]) : 0) - (_i >= 6 ? _m6 * math.pow(_x, 6) * nz(_f[6]) : 0) + (_i >= 7 ? _m7 * math.pow(_x, 7) * nz(_f[7]) : 0) - (_i >= 8 ? _m8 * math.pow(_x, 8) * nz(_f[8]) : 0) + (_i == 9 ? _m9 * math.pow(_x, 9) * nz(_f[9]) : 0) // 9 var declaration fun f_pole (_a, _s, _i) => _f1 = f_filt9x(_a, _s, 1), _f2 = (_i >= 2 ? f_filt9x(_a, _s, 2) : 0), _f3 = (_i >= 3 ? f_filt9x(_a, _s, 3) : 0) _f4 = (_i >= 4 ? f_filt9x(_a, _s, 4) : 0), _f5 = (_i >= 5 ? f_filt9x(_a, _s, 5) : 0), _f6 = (_i >= 6 ? f_filt9x(_a, _s, 6) : 0) _f7 = (_i >= 2 ? f_filt9x(_a, _s, 7) : 0), _f8 = (_i >= 8 ? f_filt9x(_a, _s, 8) : 0), _f9 = (_i == 9 ? f_filt9x(_a, _s, 9) : 0) _fn = _i == 1 ? _f1 : _i == 2 ? _f2 : _i == 3 ? _f3 : _i == 4 ? _f4 : _i == 5 ? _f5 : _i == 6 ? _f6 : _i == 7 ? _f7 : _i == 8 ? _f8 : _i == 9 ? _f9 : na [_fn, _f1] //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Inputs //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Source src = input(defval=hlc3, title="Source") // Poles int N = input.int(defval=4, title="Poles", minval=1, maxval=9) // Period int per = input.int(defval=144, title="Sampling Period", minval=2) // True Range Multiplier float mult = input.float(defval=1.414, title="Filtered True Range Multiplier", minval=0) // Lag Reduction bool modeLag = input.bool(defval=false, title="Reduced Lag Mode") bool modeFast = input.bool(defval=false, title="Fast Response Mode") //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Definitions //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Beta and Alpha Components beta = (1 - math.cos(4*math.asin(1)/per)) / (math.pow(1.414, 2/N) - 1) alpha = - beta + math.sqrt(math.pow(beta, 2) + 2*beta) // Lag lag = (per - 1)/(2*N) // Data srcdata = modeLag ? src + (src - src[lag]) : src trdata = modeLag ? ta.tr(true) + (ta.tr(true) - ta.tr(true)[lag]) : ta.tr(true) // Filtered Values [filtn, filt1] = f_pole(alpha, srcdata, N) [filtntr, filt1tr] = f_pole(alpha, trdata, N) // Lag Reduction filt = modeFast ? (filtn + filt1)/2 : filtn filttr = modeFast ? (filtntr + filt1tr)/2 : filtntr // Bands hband = filt + filttr*mult lband = filt - filttr*mult // Colors color1 = #0aff68 color2 = #00752d color3 = #ff0a5a color4 = #990032 fcolor = filt > filt[1] ? #0aff68 : filt < filt[1] ? #ff0a5a : #cccccc barcolor = (src > src[1]) and (src > filt) and (src < hband) ? #0aff68 : (src > src[1]) and (src >= hband) ? #0aff1b : (src <= src[1]) and (src > filt) ? #00752d : (src < src[1]) and (src < filt) and (src > lband) ? #ff0a5a : (src < src[1]) and (src <= lband) ? #ff0a11 : (src >= src[1]) and (src < filt) ? #990032 : #cccccc //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Outputs //----------------------------------------------------------------------------------------------------------------------------------------------------------------- // Filter Plot filtplot = plot(filt, title="Filter", color=fcolor, linewidth=3) // Band Plots hbandplot = plot(hband, title="Filtered True Range High Band", color=fcolor) lbandplot = plot(lband, title="Filtered True Range Low Band", color=fcolor) // Channel Fill fill(hbandplot, lbandplot, title="Channel Fill", color=color.new(fcolor, 80)) // Bar Color barcolor(barcolor) longCondition = ta.crossover(close, hband) and timeCondition closeAllCondition = ta.crossunder(close, hband) and timeCondition if longCondition strategy.entry("long", strategy.long) if closeAllCondition strategy.close("long")