Āagman Research Handbook
A practical guide to discovering and analyzing stocks, options, mutual funds, and commodities with natural language in Āagman's Research section.
Screening answers "what matches?" Analysis answers "what do I think?"
1. The mental model
Āagman's Research section handles two kinds of questions from the same workspace. Screening answers "Which instruments match this rule?" and returns a table. Analysis answers "What do I think about this instrument?" and returns scores, scenarios, and written reasoning.
You don't need to switch tabs or open a different tool. Type your prompt in the Research workspace, and Āagman figures out whether you're screening or analyzing based on what you ask.
What happens under the hood
You type a prompt in the Research workspace.
Āagman classifies your intent — screening or analysis.
The system resolves symbols, universes, metrics, and indicators.
It fetches market data and runs the query.
It returns a structured response: a filterable table for screens, a score card and written analysis for research.
A caution on what Research can and can't tell you
A screen finds instruments that match a rule right now. It doesn't tell you whether the rule has predictive value, whether the match will persist tomorrow, or whether you should act on it.
An analysis is not a recommendation. The score summarizes evidence across multiple dimensions, not a buy/sell signal. Scores change as prices and fundamentals change — treat them as one input alongside your own judgment.
2. Screening or analyzing? Two intents, one workspace
Both happen inside the Research section. Āagman reads your prompt and decides which mode to use.
| You want to... | What to type | What you get |
|---|---|---|
| Find stocks matching a technical condition | Screen NIFTY 50 for RSI below 30 | Filterable table |
| Find stocks matching a fundamental condition | Show me stocks with PE below 15, ROE above 20% | Filterable table |
| Get options with unusual IV or OI | Show unusual IV Rank on NIFTY options | Filterable table |
| Filter mutual funds | Show me large-cap mutual funds | Filterable table |
| Understand one stock in depth | Research on RELIANCE | Score card + analysis |
| Compare stocks head-to-head | Compare RELIANCE and TCS | Comparison + recommendation |
| Get bull / bear / base scenarios | Bull bear base case for RELIANCE | Three scenarios |
| Score a mutual fund | Score SBI Bluechip Fund | Score card + analysis |
| Decide between two investments | Which is better: SBIN or ICICIBANK? | Comparison + recommendation |
Start with a screening prompt to build a short list, then follow up with an analysis prompt on the finalists — all in the same workspace.
3. How do I write my first screener query?
Ask Āagman to find stocks or instruments that match your criteria, such as RSI below 30 or market cap above a threshold.
Open the Research section and paste this.
You'll see a criteria summary, a paginated table of matching stocks with RSI values, sortable headers, a CSV export button, and a coverage note telling you how many stocks were screened.
What to notice
RSI below 30 is a classic oversold signal. In a bull market, this list may be short or empty. In a correction, it may be long. The screen tells you what is oversold; it doesn't tell you whether to buy. Follow up with an analysis prompt on any stock that looks interesting.
4. What makes a good screener prompt?
A good screener prompt states the universe, filters, sorting, and output format you want.
Every screener prompt contains at least three parts.
4.1 The universe
The pool of instruments the filter runs over.
- Index: NIFTY 50, BANKNIFTY, FINNIFTY
- Sector: IT stocks, banking stocks, pharma stocks
- Market-cap bucket: large-cap stocks, mid-cap stocks
- Custom list: RELIANCE TCS INFY HDFCBANK SBIN
- All equities: all NSE stocks
Always specify the universe. "Screen for RSI below 30" is ambiguous — Āagman doesn't know whether you mean 50 stocks or 2,000.
4.2 The filter
| Filter type | Examples |
|---|---|
| Technical | RSI below 30, MACD bullish crossover, close above 200-day SMA |
| Fundamental | PE below 15, ROE above 20%, Piotroski score above 7 |
| Options | IV rank above 80, unusual OI buildup |
| Mutual fund | large-cap funds, flexi-cap funds with 3-year return above 15% |
4.3 The condition
- below 30
- above 20%
- crossed above
- any time in the last month
4.4 Optional modifiers
- top 10 or bottom 5
- sorted by return
- with market cap above ₹50,000 crore
5. Screener pattern library
Each pattern gets a concept, a prompt, variations, and a common failure mode.
RSI oversold
Concept. Find stocks that have been sold off aggressively and may be due for a bounce.
- Add a trend filter: ...and price above 200-day SMA
- Use RSI(21) for smoother signals
- Look for exits: RSI above 70
Oversold stocks can stay oversold. Without a trend filter, this list can catch falling knives.
MACD crossover
Concept. Find stocks where short-term momentum is turning up relative to longer-term momentum.
- Add a trend filter: ...and price above SMA(200)
- Use the zero-line crossover
- Look for bearish crossovers
MACD crossovers are lagging. In choppy markets they fire constantly and most signals are false.
Breakout
Concept. Find stocks breaking above a recent resistance level.
- Add volume: ...and volume is 1.5x the 20-day average
- Use 52-week high
- Short breakouts: crossed below their 20-day low
False breakouts are common. Volume confirmation helps but doesn't eliminate the problem.
Candlestick pattern
Concept. Use price-pattern recognition to find potential reversal or continuation signals.
- Bullish engulfing pattern today
- Hammer pattern and RSI below 40
- Morning Star pattern this week
Candlestick patterns are rare and produce very few matches on any given day. Combine with support/resistance or trend filters.
PE + ROE
Concept. Find reasonably priced, profitable companies.
- Add market cap: ...and market cap above ₹10,000 crore
- Add debt filter: debt-to-equity below 0.5
- Use PEG ratio below 1
Low PE can mean a "value trap." ROE alone doesn't reveal excessive leverage — add a debt filter.
Piotroski score
Concept. Joseph Piotroski's F-score identifies financially strong companies using nine binary tests across profitability, leverage, and operating efficiency.
- Combine with a PE filter
- Combine with RSI on the survivors
Piotroski scores are backward-looking. A perfect historical score can still face future headwinds.
Cross-screener
Concept. Use fundamental filters for quality, then technical filters for timing.
Combining filters can leave zero results. Loosen the technical filter first — timing is harder to force than quality.
Historical scan
Concept. Find stocks that met a condition at any point during a lookback window, not just today.
Historical scans can return many symbols — add a recency or severity filter to narrow results.
Custom universe screen
Concept. Run a filter only on symbols you already care about.
If a symbol in your list has no data for the period, it's skipped and noted in the coverage caption.
Options — IV rank
Concept. Find options where implied volatility is unusually high or low relative to the past year.
IV rank tells you volatility is high or low relative to history — not direction, or whether current IV is justified.
Options — OI buildup
Concept. Identify strikes where option traders are building significant positions.
OI buildup shows activity, not direction. Combine with price action for context.
Mutual fund screen
Concept. Find mutual funds by category, AMC, or performance.
- Flexi-cap funds with 3-year return above 15%
- Top 5 mid-cap funds by 5-year return
Sorting by past returns is the most common mistake in fund selection — a top-returning fund may have taken the most risk to get there.
6. How do I run my first research query?
Ask for a deep dive on a single stock, mutual fund, or commodity and Āagman returns fundamentals, technicals, and news context.
In the same Research workspace, paste this.
You'll see an overall score (e.g., 72/100), a written analysis, dimension breakdowns (momentum, valuation, quality, risk), bull/bear/base scenarios, and a SEBI disclaimer at the bottom.
The analysis doesn't give a buy or sell recommendation. The score summarizes the opinion; the text explains it. Your job is to weigh the evidence against your own thesis.
7. Anatomy of a research prompt
7.1 The instrument
- Single stock: RELIANCE, HDFCBANK, TCS
- Multiple stocks: RELIANCE vs TCS
- Mutual fund: SBI Bluechip Fund
- Index: NIFTY 50
7.2 The output shape
| Shape | Example prompt | What you get |
|---|---|---|
| Score card | Analyze RELIANCE | Overall score + dimension breakdown |
| Scenarios | Bull bear base case for RELIANCE | Three probability-weighted outcomes |
| Comparison | Compare RELIANCE and TCS | Side-by-side scores + recommended pick |
| Ranking | Rank RELIANCE, TCS, INFY | Ordered list with edge for each |
| Valuation | Is IRCTC overvalued? | Valuation-focused analysis |
7.3 The depth
- Quick: Score RELIANCE
- Deep: Deep dive on TCS — fundamentals, technicals, and outlook
- Focused: RELIANCE ka valuation analysis do
If you don't specify depth, Āagman defaults to a standard score card. Ask for a deep dive when you want the full picture.
8. Understanding research scores
Analysis prompts return an overall score out of 100 and a breakdown across dimensions.
| Dimension | What it measures | High score means | Watch out for |
|---|---|---|---|
| Momentum | Price trend strength, relative performance | Uptrend, outperforming peers | Momentum reverses hardest at turning points |
| Valuation | PE, PB, earnings yield vs. growth | Cheap relative to fundamentals | "Cheap" can mean the market sees problems |
| Quality | ROE, ROCE, margin stability | Strong, stable fundamentals | Doesn't protect against sector disruption |
| Risk | Drawdown history, volatility, beta | Relatively low risk | Low historical risk ≠ future stability |
| Growth | Revenue and earnings growth trajectory | Growing meaningfully | Growth at any price often disappoints |
The overall score is a weighted average, not a buy/sell threshold — a score of 80 isn't necessarily better than 65 if the 65 has lower risk and better valuation for your portfolio. Scores change: a stock scored 75 last week can be 68 this week. Don't anchor on a number — recheck before acting.
9. Research pattern library
Single-stock research
Concept. Get a full picture of one stock — fundamentals, technicals, catalysts, and risks.
For thinly traded or newly listed stocks, data may be sparse and the analysis less reliable.
Bull / bear / base case
Concept. Get three probability-weighted scenarios instead of a single verdict.
- Bull — upside target and the reasoning behind it
- Bear — downside target and the reasoning behind it
- Base — most likely outcome with a target price
Scenario targets are estimates, not forecasts — useful for the range of outcomes, not as price predictions to trade against.
Stock comparison
Concept. Compare two or more stocks on the same dimensions and surface a recommended pick.
The "winner" depends on the dimensions Āagman weights. State your priority explicitly: "...prioritize valuation."
Stock ranking
Concept. Rank a list of stocks by a common criterion.
Rankings require comparable data. Newly listed or missing-data stocks can be ranked unfairly.
Deep dive
Concept. Combine fundamentals, technicals, and outlook into one comprehensive report.
Deep dives are long. If you only care about one angle, ask for that specifically for a sharper answer.
Mutual fund research
Concept. Score a mutual fund across multiple dimensions and get an investment opinion.
Fund names can be ambiguous — "SBI Bluechip" could mean direct or regular plan. Use the full scheme name.
Valuation question
Concept. Ask whether a stock is over- or undervalued at current levels.
Valuation is inherently subjective. Āagman gives a structured opinion, not a fact.
Hinglish research
Concept. Ask in Hindi-English mixed language.
Very informal or broken Hinglish may not resolve correctly — keep the stock name clear.
10. Prompt engineering
10.1 Be explicit about the universe
10.2 Use standard indicator names
10.3 Ask for the output shape you want
- Screen ... → table
- Analyze ... → score card + text
- Bull bear base case for ... → scenarios
- Compare A vs B → comparison table
- Rank ... → ordered list
10.4 Avoid ambiguity
10.5 For analysis, state what you care about
10.6 For comparisons, state your priority
10.7 Handle empty screener results
11. Reading the results
11.1 Screener output
- Criteria summary — what was screened
- Paginated table — ticker + requested metrics
- Sortable headers — click to reorder
- Best-in-column highlight — top value marked
- CSV export — download full results
- Row quick-look — click a row for a sparkline and indicator snapshot
- Coverage note — how many symbols screened and how many skipped
11.2 Analysis output
- Overall score — e.g., 72/100
- Dimension scores — momentum, valuation, quality, risk, growth
- Written analysis — structured reasoning
- Scenarios — bull / bear / base (if asked)
- Comparison table — for multi-instrument queries
- Confidence level — how much data backed the opinion
- SEBI disclaimer — not investment advice
11.3 Red flags
- Empty screener result — your filter may be too strict or the condition hasn't occurred in this universe
- Very few matches with extreme values — the result may be noise rather than signal
- Analysis score without reasoning — ask for a deeper explanation; the score alone is not actionable
- One stock dominating a ranking — check whether it's genuinely better or just more volatile
12. How to iterate
Screener iteration
Start broad. One filter, one universe. See what comes back.
Add filters one at a time. If RSI < 30 gives you 15 stocks, add a trend filter and see how many survive.
If you get zero results, loosen the technical filter first. Quality and valuation filters are worth keeping strict; timing filters are more flexible.
Export and track. Download your screen as CSV. Run it again a week later and compare.
Feed finalists into analysis. A screen gives you a list — follow up with Analyze or Compare in the same workspace.
Analysis iteration
Start with a quick score before going deep.
If the score is interesting, ask for scenarios. Bull bear base case tells you the range of outcomes.
If you're comparing, do it explicitly — use Compare A vs B vs C in one prompt.
Recheck before acting. Scores move with markets.
Challenge the opinion. The output is an input, not an instruction.
13. Troubleshooting
"No results found"
- Filter is too strict
- Condition hasn't occurred in the requested universe
- Data is missing for the period
Fix: loosen the filter, widen the universe, or extend the date range.
"Too many results"
Filter is too loose, or universe is too broad. Fix: add a secondary filter or limit to top 10.
"Symbol not found"
- Use exact NSE symbols: RELIANCE, not Reliance Industries
- For mutual funds, use the full scheme name: SBI Bluechip Fund Direct Plan Growth
"Ambiguous query"
Āagman may ask for clarification. Answer with specifics — universe, timeframe, indicator parameters.
Analysis returns a generic answer
Your prompt was probably too open-ended. Instead of "Analyze RELIANCE," try "Analyze RELIANCE's valuation and momentum." Add context: "...after the recent 15% correction."
Unsupported asset
Āagman covers Indian equities, options, mutual funds, and MCX commodities. It doesn't cover international stocks, crypto, forex, or bonds.
14. Appendix
14.1 Supported asset classes
- Equities & indices (NSE/BSE)
- Options (NIFTY, BANKNIFTY, FINNIFTY, MIDCPNIFTY, SENSEX, BANKEX, stock options)
- Mutual funds (AMFI scheme codes)
- Commodities (MCX futures)
14.2 Supported screener metrics
Technical: RSI, MACD, EMA, SMA, ATR, ADX, Bollinger Bands, Stochastic, OBV, candlestick patterns, MAX/MIN, z-score
Fundamental: P/E, ROE, ROCE, debt-to-equity, Piotroski score, market cap, dividend yield, earnings quality
Options: IV rank, open interest, PCR
14.3 Supported analysis outputs
- Single-stock score card
- Bull / bear / base scenarios
- Multi-stock comparison and ranking
- Mutual fund score card
- Valuation-focused analysis