Understand stock screening for trades

Why Screening Matters (The Economics)

Time efficiency: Manual screening takes ~2 min/stock × 5,000 stocks = 166 hours. Automated screening takes 10 seconds.

Consistency: Humans get tired, forget criteria, apply rules inconsistently. Screners apply identical logic to every stock, every time.

Opportunity cost: Every hour spent on non-qualifying stocks is an hour NOT spent analyzing high-probability setups or managing open positions.

Start with financial statement data → Apply ratio-based filters → Rank by valuation/quality Common fundamental filters: $\text{P/E Ratio} = \frac{\text{Price per Share}}{\text{Earnings per Share}} < 15 \text{ (value screen)}$ $\text{ROE} = \frac{\text{Net Income}}{\text{Shareholder Equity}} > 15\% \text{ (quality screen)}$

$\text{Debt-to-Equity} = \frac{\text{Total Debt}}{\text{Total Equity}} < 0.5 \text{ (safety screen)}$

Why these ratios?

P/E < 15: Identifies "cheap" stocks relative to earnings (value hypothesis: mean reversion)
ROE > 15%: Finds companies efficiently generating profit from equity (quality = sustainable growth)
D/E < 0.5: Avoids over-leveraged companies (reduces bankruptcy risk in downturns)

Example fundamental screen (Value + Quality):

Filter 1: Market Cap > ₹500 crore (liquidity)
Filter 2: P/E < 12 (undervalued)
Filter 3: ROE > 18% (high quality)
Filter 4: Debt/Equity < 0.3 (safe balance sheet)
Filter 5: Revenue growth > 10% YoY (growth momentum)

Why this works: You're buying quality businesses (ROE > 18%) at discount prices (P/E < 12) with low risk (D/E < 0.3). The value-quality combo has historically outperformed.

2. Technical Screening (Swing/Momentum Trading)

Formula Technical Screen Logic

Start with price/volume data → Apply pattern/indicator filters → Rank by momentum/setup quality

Common technical filters: $\text{RSI} = 100 - \frac{100}{1 + \frac{\text{Avg Gain (14 days)}}{\text{Avg Loss (14 days)}}}$

Screen for oversold: $30 < \text{RSI} < 40$ (potential reversal candidates) $\text{ADX} = \text{Directional Movement Index} > 25 \text{ (trending market)}$ Why these?

RSI 30-40: Price has fallen sharply (oversold), but not crashed (RSI < 20 often = disaster). Mean-reversion edge.
ADX > 25: Strong trend exists. Momentum strategies need trends to work; ADX < 20 = chopy/sideways = avoid.

Example technical screen (Momentum Breakout):

Filter 1: Price >50-day MA (uptrend)
Filter 2: Volume today > 2× 20-day avg volume (breakout confirmation)
Filter 3: Close within 5% of 52-week high (near resistance)
Filter 4: ADX > 30 (strong trend)
Filter 5: Market cap > ₹1,000 crore (avoid pump-and-dump)

Why this works: You're catching stocks breaking to new highs WITH volume confirmation (institutional buying) during strong trends. Momentum continuation edge.

Deriving a Screen from Strategy (First Principles)

Step 1: Define your edge hypothesis

Example: "Mean reversion in oversold quality stocks outperforms over 3-6 months."

Step 2: Translate hypothesis to measurable filters

"Oversold" → RSI < 35 OR Price < 20-day BB lower band
"Quality" → ROE > 15%, consistent earnings growth
"3-6 months" → Use daily data, not intraday (matches time horizon)

Step 3: Set universe boundaries

Market cap > ₹500 crore (liquidity for your position size)
Average volume > 500,000 shares (can enter/exit without slippage)

Step 4: Rank output

Primary: RSI (most oversold first)
Secondary: ROE (highest quality first)

Result: Top 10 stocks are the MOST oversold HIGHEST quality names → manual review for catalysts/news.

Worked example Worked Example: Designing a Dividend Growth Screen

Strategy hypothesis: "Companies that consistently grow dividends outperform because they have pricing power, disciplined capital allocation, and management confidence." Translate to filters:

Filter 1: Dividend yield > 2% (meaningful income)
Filter 2: 5-year dividend CAGR > 10% (growth, not just yield)
Filter 3: Payout ratio < 60% (sustainable, room to grow)
Filter 4: ROE > 15% (generates cash to fund dividends)
Filter 5: Debt/Equity < 0.5 (won't cut dividends in downturn)

Why each filter?

Yield > 2%: Eliminates growth stocks that pay token dividends. We want income + growth.
5Y CAGR > 10%: Historical commitment to growing payout (management quality signal).
Payout < 60%: If payout is 90%, no room to grow. <60% means they COULD pay more but choose to retain—signals confidence in reinvestment.
ROE > 15%: Must generate returns to fund dividends. Low ROE = dividend is unsustainable.
D/E < 0.5: High debt → cash goes to interest, not dividends. In recession, dividend gets cut.

Output: Run on NSE 500 → returns 12stocks → manually check: any recent dividend cuts? Management credibility? Sector headwinds?

Why this specific combo works: You've combined yield (income now), growth (income tomorrow), quality (sustainable business), and safety (won't blow up). Each filter removes a failure mode.

Common Screening Platforms

Platform	Best For	Key Feature
Screner.in	Indian fundamental	Free, clean financial data, custom screens
Chartink.com	Indian technical	Real-time intraday scans, pattern detection
Finviz	US stocks	Fast, visual momentum/value presets
TradingView	Global technical	Pine Script custom indicators, backtest screens

The Multi-Stage Screening Funel

Example funel (Momentum strategy):

Stage 1 (5,000 → 300):
- Price > ₹20 (avoid penny stocks)
- Avg volume > 100,000 shares (can trade500 shares without slippage)

Stage 2 (300 → 25):
- Price > 50 MA (uptrend)
- Volume spike > 2× average (breakout)
- RSI < 70 (not overbought yet)

Stage 3 (25 → 5):
- Check news: is volume spike due to scandal? (eliminate)
- Check chart: clean breakout or false break? (visual confirmation)
- Check sector: is entire sector spiking? (context)
- Position size: can I risk 1% on this stop level? (risk management)

Why the funnel matters: You use computational power (stages 1-2) for 95% of work, reserve human judgment (stage 3) for the 5% that matters most.

Common mistake Common Screening Mistakes

Mistake 1: Over-optimization (curve-fitting)

What students do: Add15 filters, backtest on 10 years, achieve 95% win rate on paper.
Why it feels right: More filters = more precision, right?
Why it's wrong: You've fit noise, not signal. Every added filter reduces sample size.15-filter screen might match only 3stocks ever—pure luck.
The fix: Use 3-5 core filters aligned with ONE edge. Test on out-of-sample data (different time period). If performance colapses, you overfit.

Mistake 2: Ignoring base rates

What students do: Screen for "stocks up 10%+ on high volume" daily, find 50 stocks, assume all will continue.
Why it feels right: Big move + volume = momentum must continue.
Why it's wrong: Base rate of continuation after10% spike is ~40% (most revert). Your screen finds A LOT of false positives.
The fix: After screening, calculate: "Of the last 100 stocks my screen found, how many actually continued?" If <50%, add discriminating filters (trend strength, sector momentum).

Mistake 3: Static screens in dynamic markets

What students do: Use same screen (e.g., "P/E < 10") in 2025 bull market as in 2020 bear market.
Why it feels right: P/E < 10 is undervalued in any market.
Why it's wrong: In bull markets, P/E < 10 often = broken business (why so cheap?). In bear markets, P/E < 10 = opportunity. Context matters.
The fix: Adjust thresholds by market regime. Bull market: tighten valuation, add momentum. Bear market: loosen valuation, add quality/safety.

Building a Custom Screen: Step-by-Step

Example: Building a "Post-Earnings Momentum" screen

Step 1: Hypothesis "Stocks that beat earnings AND raise guidance tend to continue upward for 1-2 weeks as analysts revise targets upward."

Step 2: Measurable criteria

Earnings announcement in last 5 days
EPS beat consensus by >5%
Forward guidance raised (requires manual check)
Price up >3% on announcement day
Volume >2× average on announcement day

Step 3: Implementation

On Screner.in or similar:
1. Filter → Corporate Actions → Earnings Date (last 5 days)
2. Filter → EPS vs. Estimate > 5%
3. Filter → Price Change (earnings day) > 3%
4. Filter → Volume (earnings day) / Avg Volume > 2
5. Export list → manually check guidance (read press release)

Step 4: Ranking

Primary: Size of EPS beat (bigger surprise = stronger momentum)
Secondary: Volume ratio (higher = more conviction)

Step 5: Backtest

Run screen weekly for 3 months
Track: How many from screen are up 5%+ after 2 weeks?
If<60%, revise filters (maybe >5% beat is too low, needs >10%)

Step 6: Live trading

Run screen every Monday morning (post-weekend earnings)
Take top 3, enter on Tuesday open
Stop loss: -3% (below earnings day low)
Target: +7% or exit after 10 days

Integrating Screening Into Your Workflow

Daily routine example (Swing trader):

8:00 AM: Run momentum screen (breakouts from yesterday)
8:15 AM: Manual review top 10 (news, chart quality)
8:30 AM: Watchlist created (5 stocks)
9:15 AM: Market open, monitor watchlist for entry triggers
3:30 PM: Market close, run end-of-day screen (new setups for tomorrow)
Evening: Backtest/refine screen based on week's results

Why this rhythm works: You're using screens to generate ideas (morning), then applying discretion (chart/news check), then waiting for the market to confirm (entry trigger). Screening doesn't make the decision—it surfaces candidates.

Recall Explain to a 12-Year-Old

Imagine you want to find a good restaurant in a huge city with 5,000 restaurants. You can't visit all of them, so you use filters on a food app:

"Show me only Indian restaurants" (now800 left)
"4+ star ratings" (now 150 left)
"Under₹500 per person" (now 30 left)
"Open right now" (now 8 left) Now you can actually READ the menus and reviews for those8 and pick the best one. Without filters, you'd waste weeks visiting random restaurants.

Stock screening is the same. The stock market has 5,000+ stocks. You create filters like "price going up," "company making profit," "not too expensive," etc. The screen shows you maybe 10-20 stocks that match. THEN you study those 10-20 carefully to find the 1-2 best trades.

The screen doesn't pick the stock for you—it just removes the4,990 stocks that definitely WON'T work, so you can focus your brain on the 10that MIGHT work.

Connections

4.2.01-Types-of-stocks-to-trade – Screening helps filter by stock type (large/mid/small cap)
4.2.07-Understand-stock-selection-criteria – Selection criteria become screen filters
4.3.02-Technical-indicators-and-patterns – Technical screens use indicators like RSI, MACD
4.4.01-Fundamental-analysis-basics – Fundamental screens use P/E, ROE, etc.
4.5.03-Position-sizing-strategies – Screen output must be filtered by position-sizing constraints
4.6.01-Backtesting-trading-strategies – Screens MUST be backtested before live use

#flashcards/stock-market

What is stock screening? :: The systematic process of filtering thousands of stocks through quantitative criteria (price, volume, ratios) to identify 10-50 candidates matching your specific trading strategy, automating 90% of the elimination work.

Why do we screen stocks instead of analyzing all 5,000?

Time efficiency (manual screening takes 166 hours), consistency (eliminates human bias/fatigue), and opportunity cost (focus analysis time only on high-probability setups).

What are the two main screening paradigms?

(1) Fundamental screening using financial ratios like P/E, ROE, Debt/Equity for value/growth investing, and (2) Technical screening using price/volume patterns like RSI, ADX, moving averages for momentum/swing trading.

What does P/E < 15 as a screen filter achieve?

Identifies stocks trading at "cheap" valuations relative to earnings (value hypothesis: undervalued stocks tend to mean-revert upward over time).

What does ROE > 15% as a screen filter achieve?

Finds companies efficiently generating profit from shareholder equity (quality screen: high ROE suggests competitive advantage and sustainable growth).

What is the 3-stage screening funnel?

Stage 1 (broad filter: liquidity/universe → 5,000 to 200 stocks), Stage 2 (strategy filter: core criteria → 200 to 20 stocks), Stage 3 (manual filter: discretion/context → 20 to 5 trades).

Why use RSI between 30-40 in a technical screen?

Identifies oversold conditions (price fell sharply) without catastrophic crashes (RSI < 20 often means broken business), capturing mean-reversion edge.

Why add ADX > 25 to a momentum screen?

Ensures strong trending market exists (momentum strategies need trends to work; ADX < 20 indicates choppy sideways market where momentum fails).

What is over-optimization in screening?

Adding too many filters (10+) that fit historical noise rather than signal, resulting in screens that match only 2-3 stocks ever (curve-fitting) and fail out-of-sample.

Why must screens be adjusted by market regime?

Same thresholds (e.g., P/E < 10) mean different things in bull vs. bear markets (in bulls, ultra-low P/E often = broken business; in bears = opportunity), so context-aware adjustment improves edge.

What is the "base rate" mistake in screening?

Finding many stocks matching criteria (e.g., 50 stocks up 10%+) but ignoring that historical continuation rate is only40%, leading to false positives—solution is calculate screen's actual hit rate and refine.

How do you derive a screen from a trading strategy?

(1) Define edge hypothesis, (2) Translate to measurable filters, (3) Set universe boundaries (liquidity/cap), (4) Rank output by primary/secondary criteria, (5) Backtest and refine.

What's the difference between filtering and ranking in a screen?

Filtering removes stocks not meeting criteria (binary yes/no), ranking orders survivors by priority (e.g., most oversold first, highest ROE first) to focus analysis on best candidates.

Why include market cap > ₹500 crore in screens?

Ensures liquidity—can enter/exit positions without excessive slippage or difficulty finding counterparty, crucial for actually trading the screen's output.

What is a "discretionary overlay" in screening workflow?

The manual Stage 3 review where you check news, chart quality, sector context, and risk/reward on the10-20 algorithmic survivors before making final 3-5 trade decisions.

Concept Map

Hinglish (regional understanding)

Intuition Hinglish mein samjho

Stock screening samajhne ke liye ek simple analogy hai: maan lo tumhe ek badi library mein se ek best book dhoondhni hai, par5,000 books hain. Tum sab nahi padh sakte, tohum filters lagate ho — "fiction only," "published after 2020," "4+ star ratings." Bas10-15 books bachti hain, aur

What is Stock Screening?