Lessons Learned from Failures & Successes

No screener is perfect. Some will give you lists that look great on paper but fail in real trades. Others may surprise you by working better than expected.

The real value lies in learning from both failures and successes.

1. Common Failure: Overfitting

A screener that looked flawless in backtests but collapsed in real markets.

Example:

An AI prompt during 2020 bull run — “Show me mid-cap IT stocks with RSI below 30, they always bounce back.”

Worked brilliantly then. Failed badly in 2022 when markets turned choppy.

Lesson: Backtest across different market phases, not just one.

2. Common Failure: Blindly Trusting Sentiment

Sentiment-based screeners can be misled by hype.

Example:

During IPO season, AI flagged companies with strong “positive sentiment” keywords like “oversubscribed” and “high demand”.

But many of those stocks fell after listing because fundamentals were weak.

Lesson: Sentiment works only when combined with financial strength and price action.

3. Common Success: Combining Fundamentals + Technicals

The strongest results often come when both sides agree.

Example:

A screener asking for “Nifty 100 companies with ROE above 15%, debt-to-equity below 0.5, and trading above 200-DMA.”

Consistently produced stable shortlists of companies like HDFC Bank, Infosys, Tata

Consumer.

Lesson: Cross-checking fundamentals with technicals reduces false positives.

4. Common Success: Keeping It Simple

The best screeners aren’t the most complicated.

Example:

A very basic filter — “Price above 200-DMA and RSI between 40–60” — used weekly. Didn’t try to be perfect, but gave reliable candidates without overfitting.

Lesson: Simple, repeatable rules often beat complex ones.

How to Apply These Lessons

• Don’t trust results from one market phase — always test across bull, bear, and sideways conditions.
• Combine indicators: fundamentals, technicals, and sentiment together.
• Start with simple prompts; add layers only when needed.
• Track performance: keep a sheet logging which screeners worked and which failed.
• Use failures as feedback — refine prompts instead of abandoning them.

Final Takeaway

Failures show you where screeners break. Successes show you what consistently works. Both are equally valuable.

The key is to treat AI as a partner — test, refine, and learn with it, instead of expecting perfect answers every time.

In the end, good investors don’t avoid mistakes. They use them to sharpen their edge for the next trade.