Skip to content

finance.md

Latest commit

 

History

History
159 lines (131 loc) · 4.67 KB

finance.md

File metadata and controls

159 lines (131 loc) · 4.67 KB
Raw

Finance

image image

Telegram bot for tracking a configurable finance watchlist. The watchlist can contain ETFs and single-company stocks such as VWRA.L, CSPX.L, AAPL, MSFT, or TSLA.

❔ How It Works

flowchart LR
    A["Telegram Commands"] --> B["Load Settings"]
    B --> C["Yahoo Finance Chart API"]
    C --> D["Compute Indicators"]
    D --> E["Build Composite Score"]
    E --> F["Format Telegram Message"]
    F --> G["Send Reply"]

    H["Weekday Scout 16:00 SGT"] --> B
    E --> I{"Score >= Threshold?"}
    I -->|No| J["Skip"]
    I -->|Yes| K{"Score Improved vs Last Run?"}
    K -->|No| J
    K -->|Yes| L["Send Channel Alert"]
    L --> M["Persist Latest Score"]
    J --> M
Loading

🤖 Commands

Command Description
/start Show help
/help Show help
/a Analyze the saved watchlist
/s TSLA Analyze one ticker without saving it
/w Show the current watchlist
/w add AAPL MSFT Add one or more tickers
/w rm AAPL Remove one or more tickers
/w set AAPL MSFT NVDA Replace the watchlist
/w reset Reset to default symbols
/t Show the alert threshold
/t set 5 Set the scout alert threshold
/t reset Reset the threshold to default
/vwra Analyze VWRA.L
/cspx Analyze CSPX.L

⏰ Scout Schedule

The finance scout runs once per weekday at 16:00 Singapore time.

During each run, it:

  1. Loads the saved watchlist for TELEGRAM_CHANNEL_FINANCE
  2. Analyzes every symbol in that watchlist
  3. Sends a Telegram channel alert only when:
    • CompositeScore >= alert_threshold
    • the score is higher than the previously saved score for that same ticker
  4. Saves the latest score state back to Supabase

Examples:

  • VWRA.L from 3 to 4 with threshold 4 -> send
  • CSPX.L from 4 to 5 with threshold 4 -> send
  • AAPL from 5 to 5 -> skip
  • MSFT from 6 to 4 -> skip

📊 Indicators

The finance score is built from four indicators in finance/indicators.go.

RSI

Relative Strength Index with a 14-period window.

Signals:

  • RSI < 30 -> Oversold
  • RSI > 70 -> Overbought
  • otherwise -> Neutral

Score contribution:

  • < 30 -> +3
  • < 40 -> +1
  • > 70 -> -3
  • > 60 -> -1

MACD

Moving Average Convergence Divergence using 12, 26, 9.

Signals:

  • bullish crossover -> +2
  • bearish crossover -> -2
  • otherwise -> 0

The bot compares the latest histogram against the previous histogram to detect the crossover.

SMA

Simple moving averages using SMA50 and SMA200.

Signals:

  • price > SMA50 and SMA50 > SMA200 -> Strong Uptrend
  • price < SMA50 and SMA50 < SMA200 -> Strong Downtrend
  • price above both -> Above Both MAs
  • price below both -> Below Both MAs
  • otherwise -> Mixed

Score contribution:

  • Strong Uptrend -> +2
  • Strong Downtrend -> -2
  • other states -> 0

Bollinger Bands

Bollinger Bands using 20 periods and 2 standard deviations.

Signals:

  • near or below lower band -> bullish bias
  • near or above upper band -> bearish bias
  • otherwise -> within bands

Score contribution:

  • Near Lower Band or Below Lower Band -> +2
  • Near Upper Band or Above Upper Band -> -2
  • Within Bands -> 0

🧮 Composite Score

The bot adds all four indicator scores:

CompositeScore = RSIScore + MACDScore + SMAScore + BBScore

Recommendation mapping:

  • >= 5 -> Strong Buy
  • >= 2 -> Buy
  • >= -1 -> Neutral
  • >= -4 -> Sell
  • < -4 -> Strong Sell

🛠️ Setup

Environment Variables

Name Desc
TELEGRAM_BOT Telegram bot API token
TELEGRAM_CHANNEL_FINANCE Finance channel ID
SUPABASE_URL Supabase URL
SUPABASE_KEY Supabase key

Supabase Table Definition

The finance bot stores everything in a single table.

create table public."FinanceWatchlists" (
  chat_id bigint primary key,
  symbols text not null,
  alert_threshold integer not null default 5,
  score_state text not null default '{}'
);

Column meanings:

  • chat_id: Telegram chat or channel ID
  • symbols: comma-separated watchlist, for example AAPL,MSFT,VWRA.L
  • alert_threshold: minimum composite score required for scout alerts
  • score_state: JSON text of the last saved composite score per ticker, for example {"AAPL":4,"VWRA.L":5}

🧷 Notes

  • Manual analysis commands always return the current analysis immediately.
  • Supabase is the only persistent store for finance settings.
  • If a finance settings row does not exist yet, the bot falls back to default symbols and default threshold in code.