import isNumber from 'lodash-es/isNumber.js'
import size from 'lodash-es/size.js'
import each from 'lodash-es/each.js'
/**
* 計算各週期之Rolling VWAP偏離度(vwapDist)與其以ATR正規化後之偏離度(vwapDistATR)
*
* 各週期len為該期涵蓋之K線根數(以4hr K線為基準，1day=6根)，直接讀取輸入arr內High、Low、Close、Volumn欄位計算，資料筆數不足len+1根時，該期vs為空陣列
* ATR以Wilder平滑計算，種子為前len根TR之簡單平均；VWAP為典型價((High+Low+Close)/3)乘Volumn後之rolling窗口總和除以Volumn總和
* 各期第1筆對應輸入之第len根(與ATR對齊)，其後逐根對齊至最後一根
* vwapDist於Close為0時給0；vwapDistATR於ATR為0時給0
*
* Unit Test: {@link https://github.com/yuda-lyu/w-data-trade/blob/master/test/unit-calcVwap.test.mjs Github}
* @function
* @param {Array} arr 輸入K線陣列，各元素需含time、High、Low、Close、Volumn欄位
* @param {String} key 保留參數(與同系列calc函式介面一致)，目前函式內部未使用
* @param {Object} [opt={}] 輸入設定物件，預設{}，目前函式內部未使用
* @returns {Promise} 回傳Promise，resolve為各期結果陣列，各元素為{period,len,vs}，vs內各元素為{time,vwapDist,vwapDistATR}
* @example
*
* let arr = [
* { time: '2020-01-01T00:00:00', High: 105, Low: 95, Close: 100, Volumn: 10 },
* { time: '2020-01-01T04:00:00', High: 110, Low: 100, Close: 108, Volumn: 12 },
* { time: '2020-01-01T08:00:00', High: 112, Low: 104, Close: 106, Volumn: 8 },
* { time: '2020-01-01T12:00:00', High: 109, Low: 101, Close: 103, Volumn: 15 },
* ]
*
* calcVwap(arr, 'Close')
* .then((rs) => {
* let r = rs.find((v) => v.period === '12hr')
* console.log(r)
* // => {
* // period: '12hr',
* // len: 3,
* // vs: [
* // {
* // time: '2020-01-01T12:00:00',
* // vwapDist: -0.02515025427646778,
* // vwapDistATR: -0.29890109890109784
* // }
* // ]
* // }
* })
*
*/
let calcVwap = (() => {
//Rolling VWAP 各 period 對應 len (4hr K 線)
//len = rolling window 期數 (n × 4hr) — VWAP 與內部 ATR 共用
//經典 N=80 落在 12day 附近, 對應到 7day(42) 與 15day(90) 之間, 由 DE 自行挑選最佳
let kp = {
'12hr': 3,
'16hr': 4,
'20hr': 5,
'1day': 6, // 1 day = 6 * 4 hours
'2day': 12, // 2 days = 12 * 4 hours
'4day': 24, // 4 days = 24 * 4 hours
'7day': 42, // 7 days = 42 * 4 hours
'15day': 90, // 15 days = 90 * 4 hours
'30day': 180, // 30 days = 180 * 4 hours
}
let caVwap = (arr, len, opt = {}) => {
//check
if (!isNumber(len)) {
throw new Error(`len is not a number`)
}
//n
let n = size(arr)
//check, 需 len+1 根 (前 len 根供 VWAP 視窗 + ATR 種子, +1 根為第一筆輸出)
if (n < len + 1) {
return []
}
let kTime = 'time'
let kHigh = 'High'
let kLow = 'Low'
let kClose = 'Close'
let kVolumn = 'Volumn'
//計算各根 TR (從 i=1 開始, 需要前一根 Close), 對齊 calcAtr
let trs = new Array(n).fill(0)
for (let i = 1; i < n; i++) {
let h = arr[i][kHigh]
let l = arr[i][kLow]
let cPrev = arr[i - 1][kClose]
if (!isNumber(h)) {
throw new Error(`invalid h[${h}]`)
}
if (!isNumber(l)) {
throw new Error(`invalid l[${l}]`)
}
if (!isNumber(cPrev)) {
throw new Error(`invalid cPrev[${cPrev}]`)
}
let tr = Math.max(Math.abs(h - l), Math.abs(h - cPrev), Math.abs(l - cPrev))
trs[i] = tr
}
//計算各根 typical price × volume, 與 volume (供 rolling sum)
let tpvs = new Array(n).fill(0)
let vols = new Array(n).fill(0)
for (let i = 0; i < n; i++) {
let h = arr[i][kHigh]
let l = arr[i][kLow]
let c = arr[i][kClose]
let v = arr[i][kVolumn]
if (!isNumber(h)) {
throw new Error(`invalid h[${h}]`)
}
if (!isNumber(l)) {
throw new Error(`invalid l[${l}]`)
}
if (!isNumber(c)) {
throw new Error(`invalid c[${c}]`)
}
if (!isNumber(v)) {
throw new Error(`invalid v[${v}]`)
}
let tp = (h + l + c) / 3
tpvs[i] = tp * v
vols[i] = v
}
//ATR 種子: 前 len 根 TR 的 SMA, 第一筆 ATR 對應 index = len (對齊 calcAtr)
let sumTr = 0
for (let i = 1; i <= len; i++) {
sumTr += trs[i]
}
let atrPrev = sumTr / len
//VWAP 初始 rolling sum: 累加 [0, len-1] 共 len 筆
let sumTpv = 0
let sumVol = 0
for (let i = 0; i < len; i++) {
sumTpv += tpvs[i]
sumVol += vols[i]
}
//此時 sumTpv/sumVol 對應 i=len-1 之 VWAP, 但第一筆輸出對齊 i=len, 故下方需先滑動視窗
//rs
let rs = []
//第一筆: i=len (與 ATR 對齊)
{
//視窗從 [0, len-1] 滑到 [1, len]
sumTpv = sumTpv - tpvs[0] + tpvs[len]
sumVol = sumVol - vols[0] + vols[len]
let c = arr[len][kClose]
let vwap = sumVol !== 0 ? sumTpv / sumVol : c
let atr = atrPrev
let vwapDist = c !== 0 ? (c - vwap) / c : 0
let vwapDistATR = atr !== 0 ? (c - vwap) / atr : 0
rs.push({
time: arr[len][kTime],
vwapDist,
vwapDistATR,
})
}
//後續: i=len+1 起, rolling VWAP + Wilder ATR 同步遞推
for (let i = len + 1; i < n; i++) {
//rolling VWAP 視窗滑動 [i-len+1, i]
sumTpv = sumTpv - tpvs[i - len] + tpvs[i]
sumVol = sumVol - vols[i - len] + vols[i]
//Wilder ATR 遞推 (對齊 calcAtr 公式)
let atr = (atrPrev * (len - 1) + trs[i]) / len
let c = arr[i][kClose]
let vwap = sumVol !== 0 ? sumTpv / sumVol : c
let vwapDist = c !== 0 ? (c - vwap) / c : 0
let vwapDistATR = atr !== 0 ? (c - vwap) / atr : 0
rs.push({
time: arr[i][kTime],
vwapDist,
vwapDistATR,
})
atrPrev = atr
}
// console.log('rs', rs)
return rs
}
let caVwaps = (arr, opt = {}) => {
//rrs
let rrs = []
each(kp, (len, period) => {
//caVwap
let rs = caVwap(arr, len, opt)
//push
rrs.push({
period,
len,
vs: rs,
})
// console.log('rrs', rrs)
})
return rrs
}
let calcVwap = async(arr, key, opt = {}) => {
// arr = [
// {"time":"2020-01-01T00:00:00","Open":7195,"High":7225.62,"Low":7145.01,"Close":7173.32,"Volumn":4657.972543,...},
// ...
// ]
//caVwaps
let rs = caVwaps(arr, opt)
return rs
}
return calcVwap
})()
export default calcVwap