bbgo/pkg/strategy/harmonic/strategy.go

package harmonic

import (
	"context"
	"fmt"
	"os"
	"sync"

	"git.qtrade.icu/lychiyu/bbgo/pkg/bbgo"
	"git.qtrade.icu/lychiyu/bbgo/pkg/data/tsv"
	"git.qtrade.icu/lychiyu/bbgo/pkg/datatype/floats"
	"git.qtrade.icu/lychiyu/bbgo/pkg/fixedpoint"
	"git.qtrade.icu/lychiyu/bbgo/pkg/indicator"
	"git.qtrade.icu/lychiyu/bbgo/pkg/types"
	"github.com/sirupsen/logrus"
)

const ID = "harmonic"

var log = logrus.WithField("strategy", ID)

func init() {
	bbgo.RegisterStrategy(ID, &Strategy{})
}

type Strategy struct {
	Environment *bbgo.Environment
	Symbol      string `json:"symbol"`
	Market      types.Market

	types.IntervalWindow
	//bbgo.OpenPositionOptions

	// persistence fields
	Position    *types.Position    `persistence:"position"`
	ProfitStats *types.ProfitStats `persistence:"profit_stats"`
	TradeStats  *types.TradeStats  `persistence:"trade_stats"`

	ExitMethods bbgo.ExitMethodSet `json:"exits"`

	session       *bbgo.ExchangeSession
	orderExecutor *bbgo.GeneralOrderExecutor

	bbgo.QuantityOrAmount

	// StrategyController
	bbgo.StrategyController

	shark *SHARK

	AccountValueCalculator *bbgo.AccountValueCalculator

	// whether to draw graph or not by the end of backtest
	DrawGraph       bool   `json:"drawGraph"`
	GraphPNLPath    string `json:"graphPNLPath"`
	GraphCumPNLPath string `json:"graphCumPNLPath"`

	// for position
	buyPrice     float64 `persistence:"buy_price"`
	sellPrice    float64 `persistence:"sell_price"`
	highestPrice float64 `persistence:"highest_price"`
	lowestPrice  float64 `persistence:"lowest_price"`

	// Accumulated profit report
	AccumulatedProfitReport *AccumulatedProfitReport `json:"accumulatedProfitReport"`
}

// AccumulatedProfitReport For accumulated profit report output
type AccumulatedProfitReport struct {
	// AccumulatedProfitMAWindow Accumulated profit SMA window, in number of trades
	AccumulatedProfitMAWindow int `json:"accumulatedProfitMAWindow"`

	// IntervalWindow interval window, in days
	IntervalWindow int `json:"intervalWindow"`

	// NumberOfInterval How many intervals to output to TSV
	NumberOfInterval int `json:"NumberOfInterval"`

	// TsvReportPath The path to output report to
	TsvReportPath string `json:"tsvReportPath"`

	// AccumulatedDailyProfitWindow The window to sum up the daily profit, in days
	AccumulatedDailyProfitWindow int `json:"accumulatedDailyProfitWindow"`

	// Accumulated profit
	accumulatedProfit         fixedpoint.Value
	accumulatedProfitPerDay   floats.Slice
	previousAccumulatedProfit fixedpoint.Value

	// Accumulated profit MA
	accumulatedProfitMA       *indicator.SMA
	accumulatedProfitMAPerDay floats.Slice

	// Daily profit
	dailyProfit floats.Slice

	// Accumulated fee
	accumulatedFee       fixedpoint.Value
	accumulatedFeePerDay floats.Slice

	// Win ratio
	winRatioPerDay floats.Slice

	// Profit factor
	profitFactorPerDay floats.Slice

	// Trade number
	dailyTrades               floats.Slice
	accumulatedTrades         int
	previousAccumulatedTrades int
}

func (r *AccumulatedProfitReport) Initialize() {
	if r.AccumulatedProfitMAWindow <= 0 {
		r.AccumulatedProfitMAWindow = 60
	}
	if r.IntervalWindow <= 0 {
		r.IntervalWindow = 7
	}
	if r.AccumulatedDailyProfitWindow <= 0 {
		r.AccumulatedDailyProfitWindow = 7
	}
	if r.NumberOfInterval <= 0 {
		r.NumberOfInterval = 1
	}
	r.accumulatedProfitMA = &indicator.SMA{IntervalWindow: types.IntervalWindow{Interval: types.Interval1d, Window: r.AccumulatedProfitMAWindow}}
}

func (r *AccumulatedProfitReport) RecordProfit(profit fixedpoint.Value) {
	r.accumulatedProfit = r.accumulatedProfit.Add(profit)
}

func (r *AccumulatedProfitReport) RecordTrade(fee fixedpoint.Value) {
	r.accumulatedFee = r.accumulatedFee.Add(fee)
	r.accumulatedTrades += 1
}

func (r *AccumulatedProfitReport) DailyUpdate(tradeStats *types.TradeStats) {
	// Daily profit
	r.dailyProfit.Update(r.accumulatedProfit.Sub(r.previousAccumulatedProfit).Float64())
	r.previousAccumulatedProfit = r.accumulatedProfit

	// Accumulated profit
	r.accumulatedProfitPerDay.Update(r.accumulatedProfit.Float64())

	// Accumulated profit MA
	r.accumulatedProfitMA.Update(r.accumulatedProfit.Float64())
	r.accumulatedProfitMAPerDay.Update(r.accumulatedProfitMA.Last(0))

	// Accumulated Fee
	r.accumulatedFeePerDay.Update(r.accumulatedFee.Float64())

	// Win ratio
	r.winRatioPerDay.Update(tradeStats.WinningRatio.Float64())

	// Profit factor
	r.profitFactorPerDay.Update(tradeStats.ProfitFactor.Float64())

	// Daily trades
	r.dailyTrades.Update(float64(r.accumulatedTrades - r.previousAccumulatedTrades))
	r.previousAccumulatedTrades = r.accumulatedTrades
}

// Output Accumulated profit report to a TSV file
func (r *AccumulatedProfitReport) Output(symbol string) {
	if r.TsvReportPath != "" {
		tsvwiter, err := tsv.AppendWriterFile(r.TsvReportPath)
		if err != nil {
			panic(err)
		}
		defer tsvwiter.Close()
		// Output symbol, total acc. profit, acc. profit 60MA, interval acc. profit, fee, win rate, profit factor
		_ = tsvwiter.Write([]string{"#", "Symbol", "accumulatedProfit", "accumulatedProfitMA", fmt.Sprintf("%dd profit", r.AccumulatedDailyProfitWindow), "accumulatedFee", "winRatio", "profitFactor", "60D trades"})
		for i := 0; i <= r.NumberOfInterval-1; i++ {
			accumulatedProfit := r.accumulatedProfitPerDay.Index(r.IntervalWindow * i)
			accumulatedProfitStr := fmt.Sprintf("%f", accumulatedProfit)
			accumulatedProfitMA := r.accumulatedProfitMAPerDay.Index(r.IntervalWindow * i)
			accumulatedProfitMAStr := fmt.Sprintf("%f", accumulatedProfitMA)
			intervalAccumulatedProfit := r.dailyProfit.Tail(r.AccumulatedDailyProfitWindow+r.IntervalWindow*i).Sum() - r.dailyProfit.Tail(r.IntervalWindow*i).Sum()
			intervalAccumulatedProfitStr := fmt.Sprintf("%f", intervalAccumulatedProfit)
			accumulatedFee := fmt.Sprintf("%f", r.accumulatedFeePerDay.Index(r.IntervalWindow*i))
			winRatio := fmt.Sprintf("%f", r.winRatioPerDay.Index(r.IntervalWindow*i))
			profitFactor := fmt.Sprintf("%f", r.profitFactorPerDay.Index(r.IntervalWindow*i))
			trades := r.dailyTrades.Tail(60+r.IntervalWindow*i).Sum() - r.dailyTrades.Tail(r.IntervalWindow*i).Sum()
			tradesStr := fmt.Sprintf("%f", trades)

			_ = tsvwiter.Write([]string{fmt.Sprintf("%d", i+1), symbol, accumulatedProfitStr, accumulatedProfitMAStr, intervalAccumulatedProfitStr, accumulatedFee, winRatio, profitFactor, tradesStr})
		}
	}
}

func (s *Strategy) Subscribe(session *bbgo.ExchangeSession) {
	session.Subscribe(types.KLineChannel, s.Symbol, types.SubscribeOptions{Interval: s.Interval})

	if !bbgo.IsBackTesting {
		session.Subscribe(types.MarketTradeChannel, s.Symbol, types.SubscribeOptions{})
	}

	s.ExitMethods.SetAndSubscribe(session, s)
}

func (s *Strategy) ID() string {
	return ID
}

func (s *Strategy) InstanceID() string {
	return fmt.Sprintf("%s:%s", ID, s.Symbol)
}

func (s *Strategy) CalcAssetValue(price fixedpoint.Value) fixedpoint.Value {
	balances := s.session.GetAccount().Balances()
	return balances[s.Market.BaseCurrency].Total().Mul(price).Add(balances[s.Market.QuoteCurrency].Total())
}

func (s *Strategy) Run(ctx context.Context, orderExecutor bbgo.OrderExecutor, session *bbgo.ExchangeSession) error {
	var instanceID = s.InstanceID()

	if s.Position == nil {
		s.Position = types.NewPositionFromMarket(s.Market)
	}

	if s.ProfitStats == nil {
		s.ProfitStats = types.NewProfitStats(s.Market)
	}

	if s.TradeStats == nil {
		s.TradeStats = types.NewTradeStats(s.Symbol)
	}

	// StrategyController
	s.Status = types.StrategyStatusRunning

	s.OnSuspend(func() {
		// Cancel active orders
		_ = s.orderExecutor.GracefulCancel(ctx)
	})

	s.OnEmergencyStop(func() {
		// Cancel active orders
		_ = s.orderExecutor.GracefulCancel(ctx)
		// Close 100% position
		//_ = s.ClosePosition(ctx, fixedpoint.One)
	})

	s.session = session

	// Set fee rate
	if s.session.MakerFeeRate.Sign() > 0 || s.session.TakerFeeRate.Sign() > 0 {
		s.Position.SetExchangeFeeRate(s.session.ExchangeName, types.ExchangeFee{
			MakerFeeRate: s.session.MakerFeeRate,
			TakerFeeRate: s.session.TakerFeeRate,
		})
	}

	s.orderExecutor = bbgo.NewGeneralOrderExecutor(session, s.Symbol, ID, instanceID, s.Position)
	s.orderExecutor.BindEnvironment(s.Environment)
	s.orderExecutor.BindProfitStats(s.ProfitStats)
	s.orderExecutor.BindTradeStats(s.TradeStats)

	// AccountValueCalculator
	s.AccountValueCalculator = bbgo.NewAccountValueCalculator(s.session, s.Market.QuoteCurrency)

	// Accumulated profit report
	if bbgo.IsBackTesting {
		if s.AccumulatedProfitReport == nil {
			s.AccumulatedProfitReport = &AccumulatedProfitReport{}
		}
		s.AccumulatedProfitReport.Initialize()
		s.orderExecutor.TradeCollector().OnProfit(func(trade types.Trade, profit *types.Profit) {
			if profit == nil {
				return
			}

			s.AccumulatedProfitReport.RecordProfit(profit.Profit)
		})
		// s.orderExecutor.TradeCollector().OnTrade(func(trade types.Trade, profit fixedpoint.Value, netProfit fixedpoint.Value) {
		// 	s.AccumulatedProfitReport.RecordTrade(trade.Fee)
		// })
		session.MarketDataStream.OnKLineClosed(types.KLineWith(s.Symbol, types.Interval1d, func(kline types.KLine) {
			s.AccumulatedProfitReport.DailyUpdate(s.TradeStats)
		}))
	}

	// For drawing
	profitSlice := floats.Slice{1., 1.}
	price, _ := session.LastPrice(s.Symbol)
	initAsset := s.CalcAssetValue(price).Float64()
	cumProfitSlice := floats.Slice{initAsset, initAsset}

	s.orderExecutor.TradeCollector().OnTrade(func(trade types.Trade, profit fixedpoint.Value, netProfit fixedpoint.Value) {
		if bbgo.IsBackTesting {
			s.AccumulatedProfitReport.RecordTrade(trade.Fee)
		}

		// For drawing/charting
		price := trade.Price.Float64()
		if s.buyPrice > 0 {
			profitSlice.Update(price / s.buyPrice)
			cumProfitSlice.Update(s.CalcAssetValue(trade.Price).Float64())
		} else if s.sellPrice > 0 {
			profitSlice.Update(s.sellPrice / price)
			cumProfitSlice.Update(s.CalcAssetValue(trade.Price).Float64())
		}
		if s.Position.IsDust(trade.Price) {
			s.buyPrice = 0
			s.sellPrice = 0
			s.highestPrice = 0
			s.lowestPrice = 0
		} else if s.Position.IsLong() {
			s.buyPrice = price
			s.sellPrice = 0
			s.highestPrice = s.buyPrice
			s.lowestPrice = 0
		} else {
			s.sellPrice = price
			s.buyPrice = 0
			s.highestPrice = 0
			s.lowestPrice = s.sellPrice
		}
	})

	s.InitDrawCommands(&profitSlice, &cumProfitSlice)
	s.orderExecutor.TradeCollector().OnPositionUpdate(func(position *types.Position) {
		bbgo.Sync(ctx, s)
	})
	s.orderExecutor.Bind()

	for _, method := range s.ExitMethods {
		method.Bind(session, s.orderExecutor)
	}

	kLineStore, _ := s.session.MarketDataStore(s.Symbol)
	s.shark = &SHARK{IntervalWindow: types.IntervalWindow{Window: s.Window, Interval: s.Interval}}
	s.shark.BindK(s.session.MarketDataStream, s.Symbol, s.shark.Interval)
	if klines, ok := kLineStore.KLinesOfInterval(s.shark.Interval); ok {
		s.shark.LoadK((*klines)[0:])
	}

	states := types.NewQueue(s.Window)

	states.Update(0)
	s.session.MarketDataStream.OnKLineClosed(types.KLineWith(s.Symbol, s.Interval, func(kline types.KLine) {

		log.Infof("shark score: %f, current price: %f", s.shark.Last(0), kline.Close.Float64())

		nextState := hmm(s.shark.Array(s.Window), states.Array(s.Window), s.Window)
		states.Update(nextState)
		log.Infof("Denoised signal via HMM: %f", states.Last(0))

		if states.Length() < s.Window {
			return
		}
		direction := 0.
		if s.Position.IsLong() {
			direction = 1.
		} else if s.Position.IsShort() {
			direction = -1.
		}

		if s.Position.IsOpened(kline.Close) && states.Mean(5) == 0 {
			s.orderExecutor.ClosePosition(ctx, fixedpoint.One)
		}
		if states.Mean(5) == 1 && direction != 1 {
			_, _ = s.orderExecutor.SubmitOrders(ctx, types.SubmitOrder{
				Symbol:   s.Symbol,
				Side:     types.SideTypeBuy,
				Quantity: s.Quantity,
				Type:     types.OrderTypeMarket,
				Tag:      "sharkLong",
			})
		} else if states.Mean(5) == -1 && direction != -1 {
			_, _ = s.orderExecutor.SubmitOrders(ctx, types.SubmitOrder{
				Symbol:   s.Symbol,
				Side:     types.SideTypeSell,
				Quantity: s.Quantity,
				Type:     types.OrderTypeMarket,
				Tag:      "sharkShort",
			})
		}
	}))

	bbgo.OnShutdown(ctx, func(ctx context.Context, wg *sync.WaitGroup) {
		defer wg.Done()

		// Output accumulated profit report
		if bbgo.IsBackTesting {
			defer s.AccumulatedProfitReport.Output(s.Symbol)

			if s.DrawGraph {
				if err := s.Draw(&profitSlice, &cumProfitSlice); err != nil {
					log.WithError(err).Errorf("cannot draw graph")
				}
			}
		}

		_, _ = fmt.Fprintln(os.Stderr, s.TradeStats.String())
		_ = s.orderExecutor.GracefulCancel(ctx)
	})

	return nil
}

// TODO: dirichlet distribution is a too naive solution
func observeDistribution(y_t, x_t float64) float64 {
	if x_t == 0. && y_t == 0 {
		// observed zero value from indicator when in neutral state
		return 1.
	} else if x_t > 0. && y_t > 0. {
		// observed positive value from indicator when in long state
		return 1.
	} else if x_t < 0. && y_t < 0. {
		// observed negative value from indicator when in short state
		return 1.
	} else {
		return 0.
	}
}

func transitProbability(x_t0, x_t1 int) float64 {
	// stick to the same sate
	if x_t0 == x_t1 {
		return 0.99
	}
	// transit to next new state
	return 1 - 0.99
}

// HMM main function, ref: https://tr8dr.github.io/HMMFiltering/
/*
# initialize time step 0 using state priors and observation dist p(y | x = s)
for si in states:
    alpha[t = 0, state = si] = pi[si] * p(y[0] | x = si)

# determine alpha for t = 1 .. n
for t in 1 .. n:
    for sj in states:
        alpha[t,sj] = max([alpha[t-1,si] * M[si,sj] for si in states]) * p(y[t] | x = sj)

# determine current state at time t
return argmax(alpha[t,si] over si)
*/
func hmm(y_t []float64, x_t []float64, l int) float64 {
	al := make([]float64, 0, l)
	an := make([]float64, 0, l)
	as := make([]float64, 0, l)
	long := 0.
	neut := 0.
	short := 0.
	// n is the incremental time steps
	for n := 2; n <= len(x_t); n++ {
		for j := -1; j <= 1; j++ {
			sil := make([]float64, 3)
			sin := make([]float64, 3)
			sis := make([]float64, 3)
			for i := -1; i <= 1; i++ {
				sil = append(sil, 0, x_t[n-1-1]*transitProbability(i, j))
				sin = append(sin, 0, x_t[n-1-1]*transitProbability(i, j))
				sis = append(sis, 0, x_t[n-1-1]*transitProbability(i, j))
			}
			if j > 0 {
				_, longArr := floats.MinMax(sil, 3)
				long = longArr[0] * observeDistribution(y_t[n-1], float64(j))
				al = append(al, long)
			} else if j == 0 {
				_, neutArr := floats.MinMax(sin, 3)
				neut = neutArr[0] * observeDistribution(y_t[n-1], float64(j))
				an = append(an, neut)
			} else if j < 0 {
				_, shortArr := floats.MinMax(sis, 3)
				short = shortArr[0] * observeDistribution(y_t[n-1], float64(j))
				as = append(as, short)
			}
		}
	}
	_, maximum := floats.MinMax([]float64{long, neut, short}, 3)
	if maximum[0] == long {
		return 1
	} else if maximum[0] == short {
		return -1
	}
	return 0
}