algoTrading

I had made multiple algorithmic trading strategy. Backtesting strategy on NIFTY100 last 5 - 10 year data. Strategies varied from generating return from 200% to 600% in last 27 months. Then i Implemented that on my own capital of 1,00,000 and Achieved the return of 66% in 9 months during the time when NIFTY50 return in negative.

Streak Folder Code Audit Report (March 2026)

Path reviewed: AlgoDeveloper/Streak

1) What each program is doing

• anotherFindSimilarity.py

  • Simulates technical + fundamental data and filters a stock universe by how close each stock is to input-stock feature ranges.
  • Uses synthetic data to avoid API limits (simulate_stock_data).

• findHighVolatile.py

  • Finds high-volatility stocks using 3-month Beta vs NIFTY and ATR, then ranks and exports results.

• findSimilarity.py

  • Pulls technical + fundamental data from yfinance, normalizes using z-score, and builds a Euclidean-distance similarity matrix.

• findWeak3.py

  • Monthly weakness detector: picks stocks with negative monthly return and ADX > 25 over last ~12 months, outputs weakest 20 per month.

• findWeak4.py

  • Short-term weak-stock selector using ATR + ADX + -DI scoring model and ranks lowest-score stocks.

• findWeek.py

  • Backtest-style split (previous month vs current month) for weak-stock continuation, then calculates short-side success rate.

• findWeek2.py

  • Volatility screener using ATR%, Bollinger width, intraday range %, and rolling std deviation threshold filters.

• highVolatileStock.py

  • Similarity/profile matching approach with technical + fundamental ranges (1-month period), then filters a large stock universe.

• withoutFA.py

  • Technical-only profile match (Volatility, ATR%, RSI) and proximity sort to input-stock feature range.

• withoutFundamentalHighLow.py

  • Technical-only filter + “below 52-week midpoint” condition, then ranks by normalized closeness to range midpoints.

• withRSIHighLow.py

  • Technical filter with RSI and Beta + “below 52-week midpoint” condition over a stock universe.

4) Suggested next improvements (high ROI)

1. Build a shared utility module for:
   • download_prices, compute_beta, compute_atr, compute_rsi, and stock-universe loading.
2. Add one common CSV schema for all strategy outputs (date, symbol, score, signal, strategy_name).
3. Add lightweight logging + try/except wrappers with per-symbol error counts.
4. Add simple walk-forward backtest harness so all selection scripts are evaluated with the same metrics.

5) Current status

• Full script-by-script review completed for all programs in AlgoDeveloper/Streak.
• Targeted refactor/improvements completed for 3 high-impact files.
• No strategy logic was radically changed; this pass focused on reliability, performance, and maintainability.