lotusmarket 0.4.0

Vietnamese Stock Market Toolkit — real-time quotes, technical analysis, sentiment, money flow signals for HOSE/HNX stocks

lotusmarket

Vietnamese Stock Market Toolkit for Go & Python.

Real-time quotes, historical OHLCV, technical analysis, sentiment, money flow signals, and more for HOSE/HNX stocks. All public APIs — no API key required.

Install

Go:

go get github.com/ducnhd/lotusmarket/go

Python:

pip install lotusmarket

# With fetchers (httpx for API calls):
pip install lotusmarket[fetchers]

# Everything:
pip install lotusmarket[all]

Quick Start

Go

package main

import (
    "context"
    "fmt"

    "github.com/ducnhd/lotusmarket/go/fetchers"
    "github.com/ducnhd/lotusmarket/go/technical"
    "github.com/ducnhd/lotusmarket/go/sentiment"
    "github.com/ducnhd/lotusmarket/go/signals"
    "github.com/ducnhd/lotusmarket/go/market"
)

func main() {
    ctx := context.Background()

    // 1. Real-time quote
    stock, _ := fetchers.VPS(ctx, "ACB")
    fmt.Printf("ACB: %.0f VND, vol: %d, foreign net: %d\n",
        stock.Close, stock.Volume, stock.ForeignNetVol)

    // 2. Multiple stocks at once
    stocks, _ := fetchers.VPSMultiple(ctx, []string{"ACB", "VNM", "FPT", "HPG"})

    // 3. Historical OHLCV (200 days)
    history, _ := fetchers.EntradeHistory(ctx, "ACB", 200)
    closes := make([]float64, len(history))
    for i, h := range history {
        closes[i] = h.Close
    }

    // 4. Technical analysis — all-in-one
    d := technical.Dashboard(closes)
    fmt.Printf("RSI: %.1f, MA20: %.0f, Signal: %s, Score: %.0f\n",
        d.RSI, *d.MA20, d.Signal, d.Score)

    // 5. Individual indicators
    rsi := technical.RSI(closes, 14)
    ma20 := technical.MA(closes, 20)
    mom := technical.Momentum(closes, 20)

    // 6. Weekly candles from daily
    weekly := technical.AggregateWeekly(history)

    // 7. Vietnamese sentiment
    r := sentiment.Analyze("Ngân hàng tăng mạnh, khối ngoại mua ròng")
    fmt.Printf("Sentiment: %s (%.2f)\n", r.Label, r.Score)

    // 8. Domestic flow — buy/sell pressure
    flow := signals.ComputeDomesticFlow(stocks)
    fmt.Printf("Domestic: %s (buy pressure: %.1f%%)\n", flow.Signal, flow.BuyPressure)

    // 9. Sector flow ranking (now includes avg_change_pct, total_volume_vnd, advances/total)
    sectors := market.RankSectorsByFlow(stocks)
    for _, s := range sectors {
        fmt.Printf("#%d %s: %+.2f%% · KL %.0f tỷ · %d/%d tăng\n",
            s.Rank, s.Sector, s.AvgChangePct, s.TotalVolumeVND/1e9,
            s.AdvancesCount, s.TotalCount)
    }

    // 9b. Price driver attribution — why did this stock move today?
    attr := market.AttributeDrivers(market.DriverFeatures{
        PriceChangePct:   2.5,
        ForeignNetVND:    15e9,
        MarketDeltaPct:   0.5,
        SectorDeltaPct:   1.2,
        NewsCount:        3,
        NewsSentimentAvg: 0.6,
        RSI:              58,
        MASignal:         "BUY",
        VolumeRatio:      1.8,
    })
    fmt.Printf("Dominant driver: %s (%.1f%%)\n", attr.Dominant,
        map[string]float64{"NEWS": attr.NewsPct, "FOREIGN": attr.ForeignPct,
            "SECTOR": attr.SectorPct, "MARKET": attr.MarketPct,
            "TECHNICAL": attr.TechnicalPct, "VOLUME": attr.VolumePct}[attr.Dominant])

    // 9c. News clustering — gom tin cùng chủ đề từ nhiều nguồn
    titles := []string{
        "VN-Index tăng 1% phiên cuối tuần",
        "VNIndex đóng cửa tăng 1% phiên cuối",
        "Giá dầu WTI vượt 85 USD",
    }
    groups := sentiment.ClusterTitles(titles, 0.3)
    // groups = [[0, 1], [2]] — first two grouped as same story

    // 10. Fundamentals
    fund, _ := fetchers.KBS(ctx, "ACB")
    fmt.Printf("P/E: %.1f, P/B: %.1f, EPS: %.0f\n", fund.PE, fund.PB, fund.EPS)

    // 11. With fallback (VPS -> Entrade)
    stock2, _ := fetchers.StockWithFallback(ctx, "VNM")

    // 12. AI analysis (requires CLAUDE_API_KEY)
    import "github.com/ducnhd/lotusmarket/go/ai"
    aiClient, err := ai.New(ai.Config{APIKey: os.Getenv("CLAUDE_API_KEY")})
    if err != nil {
        // → ErrNoAPIKey with setup instructions
    }
    // Trend analysis with pre-computed indicators
    analysis, _ := aiClient.AnalyzeTrend(ctx, "ACB", history)
    fmt.Println(analysis.Text) // Vietnamese analysis

    // Q&A
    answer, _ := aiClient.AskQuestion(ctx, "ACB có nên mua không?")
    fmt.Println(answer.Text)

    _ = stock2
    _ = rsi
    _ = ma20
    _ = mom
    _ = weekly
}

Python

import os
os.environ["LOTUSMARKET_QUIET"] = "1"  # suppress startup banner

import pandas as pd
import lotusmarket as lm

# === 1. Real-time quote ===
from lotusmarket.fetchers import vps, vps_multiple, entrade_history, kbs, stock_with_fallback

stock = vps("ACB")
print(f"ACB: {stock.close:.0f} VND, vol: {stock.volume}, foreign net: {stock.foreign_net_vol}")

# Multiple stocks
stocks = vps_multiple(["ACB", "VNM", "FPT", "HPG"])

# === 2. Historical OHLCV ===
history = entrade_history("ACB", days=200)  # returns list of StockData
closes = pd.Series([h.close for h in history])

# === 3. Technical analysis — all-in-one ===
d = lm.technical.dashboard(closes)
print(f"RSI: {d.rsi:.1f}, MA20: {d.ma20:.0f}, Signal: {d.signal}, Score: {d.score:.0f}")

# Individual indicators (pandas Series in/out)
df = pd.DataFrame({"close": closes})
df["rsi"] = lm.technical.rsi(df["close"], period=14)
df["ma20"] = lm.technical.ma(df["close"], period=20)
df["ma50"] = lm.technical.ma(df["close"], period=50)

# Score and signal
score = lm.technical.score(closes)
signal = lm.technical.signal(closes)

# === 4. Weekly candles ===
daily_df = pd.DataFrame({
    "date": [h.date for h in history],
    "open": [h.open for h in history],
    "high": [h.high for h in history],
    "low": [h.low for h in history],
    "close": [h.close for h in history],
    "volume": [h.volume for h in history],
})
weekly = lm.technical.aggregate_weekly(daily_df)

# === 5. Vietnamese sentiment ===
r = lm.sentiment.analyze("Ngân hàng tăng mạnh, khối ngoại mua ròng")
print(f"Sentiment: {r.label} ({r.score:.2f})")

# === 6. Volume surge detection ===
surge = lm.signals.classify_volume_surge(
    current_vol=5000000, avg_vol_20=2000000, stddev_vol_20=500000, price_change=2.5
)
print(f"Surge: {surge.signal} / {surge.intensity} — {surge.label}")

# === 7. Domestic flow ===
stocks_df = pd.DataFrame({
    "ticker": ["ACB", "VNM", "FPT"],
    "volume": [1000000, 500000, 800000],
    "bid_vol": [600000, 300000, 500000],
    "ask_vol": [400000, 200000, 300000],
    "foreign_buy_vol": [50000, 30000, 40000],
    "foreign_sell_vol": [30000, 40000, 20000],
})
flow = lm.signals.domestic_flow(stocks_df)
print(f"Domestic: {flow.signal} (buy pressure: {flow.buy_pressure:.1f}%)")

# === 8. Sector flow ranking ===
sector_df = pd.DataFrame({
    "ticker": ["ACB", "VNM", "FPT", "HPG"],
    "close": [25000, 70000, 120000, 28000],
    "volume": [1000000, 500000, 800000, 600000],
    "change_percent": [1.5, -0.5, 2.0, -1.0],
    "foreign_net_vol": [500, -200, 300, -100],
})
sectors = lm.market.sector_flow(sector_df)
# New columns: avg_change_pct, total_volume_vnd, advances_count, total_count
print(sectors[["sector", "rank", "signal", "avg_change_pct", "total_volume_vnd"]])

# === 8b. Price driver attribution ===
attr = lm.market.attribute_drivers(lm.market.DriverFeatures(
    price_change_pct=2.5,
    foreign_net_vnd=15e9,
    market_delta_pct=0.5,
    sector_delta_pct=1.2,
    news_count=3,
    news_sentiment_avg=0.6,
    rsi=58,
    ma_signal="BUY",
    volume_ratio=1.8,
))
print(f"Dominant: {attr.dominant} · News {attr.news_pct}% · Foreign {attr.foreign_pct}%")

# === 8c. News clustering — gom tin cùng chủ đề ===
titles = [
    "VN-Index tăng 1% phiên cuối tuần",
    "VNIndex đóng cửa tăng 1% phiên cuối",
    "Giá dầu WTI vượt 85 USD",
]
groups = lm.sentiment.cluster_titles(titles, threshold=0.3)
# groups = [[0, 1], [2]] — first two grouped as same story

# === 9. Market pulse scoring ===
score_val, signal_str = lm.market.pulse_score(
    breadth_score=70, foreign_flow_score=60, volume_score=55, risk_score=40
)
print(f"Pulse: {score_val} ({signal_str})")

# === 10. Risk indicators ===
print(f"VIX risk: {lm.market.score_vix(25)}")
print(f"Yield curve risk: {lm.market.score_yield_curve(-0.3)}")

# === 11. Fundamentals ===
fund = kbs("ACB")
print(f"P/E: {fund.pe:.1f}, P/B: {fund.pb:.1f}, EPS: {fund.eps:.0f}")

# === 12. Portfolio TWR ===
from lotusmarket.portfolio import twr, ReturnSegment
returns = twr([ReturnSegment(100, 110), ReturnSegment(110, 132)])
print(f"TWR: {returns:.2%}")

# === 13. Vietnamese NLU ===
from lotusmarket.nlu import Parser
parser = Parser()
result = parser.parse("giá ACB hiện tại bao nhiêu")
print(f"Intent: {result.intent}, Tickers: {result.tickers}")

# === 14. Backtest ===
from lotusmarket.backtest import run, BacktestConfig
prices = pd.Series([20000 + i * 50 for i in range(200)], dtype=float)
bt = run(BacktestConfig(strategy="rsi"), prices)
print(f"Return: {bt.total_return:.1f}%, Max DD: {bt.max_drawdown:.1f}%, Trades: {bt.trade_count}")

# === 15. AI analysis (requires CLAUDE_API_KEY) ===
from lotusmarket.ai import AIClient, AIConfig

# Option 1: from env var
# export CLAUDE_API_KEY=sk-ant-...
# client = AIClient()

# Option 2: explicit key
client = AIClient(AIConfig(api_key="sk-ant-..."))

# Trend analysis (pre-computes RSI/MA/momentum before sending to Claude)
analysis = client.analyze_trend("ACB", closes)
print(analysis.text)  # Vietnamese analysis
print(f"Tokens: {analysis.tokens_in} in, {analysis.tokens_out} out")

# Q&A
answer = client.ask_question("ACB có nên mua không?")
print(answer.text)

# Custom model (default: sonnet, can use opus for complex analysis)
client_opus = AIClient(AIConfig(api_key="sk-ant-...", model="claude-opus-4-6", max_tokens=8192))

Modules

Pure computation (no API key, no network)

Module	Go import	Python import	Description
technical	technical	lotusmarket.technical	RSI (Wilder's smoothing), MA20/50/200, Momentum, Signal (BUY/SELL/HOLD), Score (0-100), Dashboard, Weekly aggregation
sentiment	sentiment	lotusmarket.sentiment	Vietnamese financial keyword analysis — 70+ keywords with weights, score -1.0 to +1.0
signals	signals	lotusmarket.signals	Volume surge detection (z-score or ratio), domestic buy/sell pressure (bid/ask ratio)
market	market	lotusmarket.market	Market pulse scoring (0-100, green/yellow/red), sector flow ranking, risk indicators (VIX, yield curve, news sentiment, Fed trend)
portfolio	portfolio	lotusmarket.portfolio	Time-weighted returns (TWR), 6-factor confidence scoring (0-100)
nlu	nlu	lotusmarket.nlu	Vietnamese intent classification (5 intents), ticker extraction, timeframe detection
backtest	backtest	lotusmarket.backtest	Strategy backtesting — RSI, MA cross, combined. Win rate, max drawdown, buy & hold comparison
types	types	lotusmarket.types	StockData, KBSQuote, VN30 list, sector mappings, trading fee constants

AI analysis (requires CLAUDE_API_KEY)

Module	Go import	Python import	Description
ai	ai	lotusmarket.ai	Claude API wrapper — trend analysis (pre-computes indicators), Q&A, custom prompts. Default model: sonnet. Configurable model + max_tokens.

Go: go get github.com/ducnhd/lotusmarket/go/ai (adds anthropic-sdk-go dependency) Python: pip install lotusmarket[ai] (adds anthropic dependency)

Data fetchers (network calls, no API key needed)

Fetcher	Go	Python	Data
VPS	fetchers.VPS(ctx, "ACB")	fetchers.vps("ACB")	Real-time: price, volume, change%, foreign buy/sell, bid/ask
Entrade	fetchers.EntradeHistory(ctx, "ACB", 200)	fetchers.entrade_history("ACB", 200)	Historical daily OHLCV (up to 1000 days)
KBS	fetchers.KBS(ctx, "ACB")	fetchers.kbs("ACB")	Fundamentals: P/E, P/B, EPS, Beta, Dividend Yield, Market Cap
Fallback	fetchers.StockWithFallback(ctx, "ACB")	fetchers.stock_with_fallback("ACB")	VPS primary, Entrade backup (3 retries + backoff)

Python fetchers require httpx: pip install lotusmarket[fetchers]

Data Sources

Source	URL	Data	Auth
VPS	bgapidatafeed.vps.com.vn	Real-time quotes, foreign flow, bid/ask volumes	Public
Entrade	services.entrade.com.vn	Historical daily OHLCV candles	Public
KBS	kbbuddywts.kbsec.com.vn	Fundamental metrics (P/E, P/B, EPS, Beta)	Public
SSI	iboard-query.ssi.com.vn	VN30/HNX30 index component lists	Public

Technical Analysis Details

Indicator	Method	Parameters
RSI	Wilder's exponential smoothing	Default period: 14
MA	Simple Moving Average	Periods: 20, 50, 200
Momentum	Rate of change: (current - past) / past * 100	Default period: 20
Signal	MA crossover + RSI zones	BUY: price > MA20 > MA50, RSI < 70
Score	Composite: signal + RSI zones	Range: 0-100
Weekly	ISO week aggregation from daily OHLCV	O=first, H=max, L=min, C=last, V=sum

Vietnamese Sentiment Keywords

The sentiment analyzer includes 70+ Vietnamese financial keywords with weights:

Positive (sample): tăng mạnh (1.5), bùng nổ (2.0), mua ròng (1.0), vượt đỉnh (1.5), lợi nhuận (1.0)

Negative (sample): giảm sâu (2.0), bán tháo (2.0), lao dốc (2.0), nợ xấu (1.5), rút vốn (1.5)

Score formula: (positive_sum - negative_sum) / total_sum, clamped to [-1.0, 1.0].

Domestic Flow Signal

Based on aggregate bid/ask volume ratios across stocks:

Buy Pressure	Signal
>= 60%	mua manh (strong buy)
>= 53%	nghieng mua (leaning buy)
>= 47%	can bang (balanced)
>= 40%	nghieng ban (leaning sell)
< 40%	ban manh (strong sell)

Volume Surge Detection

Uses z-score (preferred) or ratio fallback:

Z-Score / Ratio	Intensity
>= 3.0 / 3x avg	strong_surge
>= 2.0 / 2x avg	surge
< 2.0	(no surge)

Combined with price direction:

• Price up (>= 0.5%) + surge = accumulation (tien dang do vao)
• Price down (<= -0.5%) + surge = distribution (dang ban thao)
• Flat + surge = high_activity (giao dich dot bien)

Go vs Python API Conventions

Aspect	Go	Python
Input/Output	[]float64, structs	pd.Series, pd.DataFrame, dataclasses
Async	context.Context + goroutines	asyncio (fetchers have _async variants planned)
Error handling	(result, error) return	Exceptions (ValueError, ConnectionError)
Config	Option functions	Keyword arguments
Dependencies	Zero (stdlib only for pure modules)	pandas, numpy (core), httpx (fetchers)

Changelog

v0.4.0

• exposure (go/exposure, python/lotusmarket/exposure.py) — per-ticker mapping of external drivers (commodity / FX / peer) with 2-year backtest-validated Pearson r correlations for HPG, 14 VN30 banks, SSI, FPT, MWG, VHM/VIC/VRE, GAS, PLX, MSN, SAB, VNM, DGC, GVR, VJC; analyze(), format_for_prompt(), regime_change_flags(), bounce_signals() via a HistoryProvider seam.
• market/regime (go/market/regime.go, appended to python/lotusmarket/market.py) — pure deterministic regime classifier: score_regime_signals() / classify_regime() / identify_trigger() covering STABLE / VOLATILE / CRISIS (PANIC vs FUNDAMENTAL) / EUPHORIA from VIX, VN-Index change, foreign flow, news tier, and global contagion inputs.
• fetchers/cafef (go/fetchers/cafef.go, python/lotusmarket/fetchers/cafef.py) — CafeF AJAX JSON insider-trading fetcher; parses /Date(ms)/ timestamps in VN timezone (UTC+7), strips HTML name wrappers, applies 90-day cutoff; InsiderTransaction type added to types.
• earnings (go/earnings, python/lotusmarket/earnings.py) — stateless Vietnamese headline parser: extract_profit(), extract_revenue(), extract_annual_target(), detect_period(), extract_tickers(), bctc_deadlines(year) returning 6 regulatory BCTC submission deadlines.

Support

If you find this useful, consider supporting the project:

• PayPal
• Telegram Stars: @vnlotusmarketbot /donate

License

MIT