agentleFS
ShubhamkumartrichyShubhamkumartrichy/claude-trading-skills-india/skills/market-breadth/SKILL.md

market-breadth

Skill64 starsChanged 17 days agoMITView on GitHub

Market-wide breadth for Indian equities: how many NSE/BSE stocks are actually joining the index's move. From a folder of stock CSVs (NIFTY 50/500, midcap or smallcap constituents), a yfinance symbol list, or an exchange advances/declines file, breadth.py reports % of stocks above 20/50/200 DMA, the uptrend ratio, advances/declines with an A/D line and divergence against NIFTY, new 52-week highs/lows, the McClellan oscillator and summation index, breadth thrusts, sector participation and a 0–100 health score. Works offline from CSV.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    india-market-analyst: put the breadth label next to the regime in the context line.

    line 111

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Market Breadth (India)

The index tells you where a few large weights went. Breadth tells you what the typical stock did. A NIFTY high made by five heavyweights while most stocks slide is a weaker market than the chart shows.

1. Run it

python3 scripts/breadth.py status --csv-dir nifty500/ --benchmark-csv "NIFTY 50.csv" --members ind_nifty500list.csv
python3 scripts/breadth.py status --csv-dir nifty500/ --universe-file ind_nifty50list.csv --benchmark-csv nifty.csv
python3 scripts/breadth.py status --ad-csv nse_adv_dec.csv          # exchange counts, no stock files
python3 scripts/breadth.py status --universe-file ind_nifty500list.csv --save-dir nifty500/   # yfinance
python3 scripts/breadth.py history --csv-dir nifty500/ --sessions 60 --out breadth.csv --json

Stock CSVs are daily OHLCV, one per stock, file stem = symbol (RELIANCE.csv, .NS is stripped), from ../technical-analysis/scripts/ta.py fetch, a broker export, or --save-dir (caches the Yahoo download: ~500 requests for NIFTY 500). Rows with blank or null prices are skipped. --pattern filters the folder; --universe-file restricts it to a constituent list, so one NIFTY 500 folder gives NIFTY 50, NIFTY 500 and midcap breadth. --members (symbol,sector, or the Industry column of a niftyindices list) adds sector participation. Also: --as-of DATE (replay), --hl-window 252, --div-window 20, --div-min-pct 1.0, --benchmark NAME (yfinance; NIFTY by default when fetching).

2. Getting the data

  • Constituents: niftyindices.com, each index page, "Index Constituent" CSV (for example ind_nifty500list.csv, with Company Name, Industry and Symbol columns). Pass it straight to --universe-file or --members.
  • Exchange-wide A/D (--ad-csv): NSE and BSE show daily advances, declines and unchanged counts, and 52-week high/low lists, on their market-data pages. nseindia.com blocks scripts, so copy the numbers by hand or find them with WebSearch in market-wrap reports, citing the source. Columns: Date,Advances,Declines[,Unchanged,New 52W Highs,New 52W Lows,NIFTY 50]. Any case or spacing works, and the index column feeds the divergence check. This mode gives A/D, McClellan, thrust and highs/lows but no DMA participation (that needs per-stock prices).

3. What it measures

MetricDefinition in breadth.py
% above 20/50/200 DMAStocks with close > N-day SMA ÷ stocks with ≥ N bars. Change over 5 and 20 sessions in points
Uptrend ratio% of stocks with close > 50 DMA > 200 DMA (stocks with ≥ 200 bars)
Advance / declineClose vs that stock's previous close; equal = unchanged. A/D line = running sum of A − D (read its slope, not its level)
DivergenceOver --div-window sessions: index up ≥ 1% while the A/D line falls = bearish; index down ≥ 1% while it rises = bullish
New 52w highs/lowsToday's high above every high of the prior 251 sessions (lows mirrored). Only stocks with 252 bars count
McClellanRANA = 1000 × (A − D)/(A + D). Oscillator = EMA19(RANA) − EMA39(RANA) (smoothing 0.10/0.05, SMA-seeded). Summation = running sum of the oscillator from the first value
Breadth thrust10-session EMA of A/(A + D) closes above 0.615 having been below 0.40 within the previous 10 sessions (Zweig's thresholds). "Armed" = below 0.40 recently, window still open
Sector participationA sector participates when ≥ 50% of its members are above their 50 DMA
Circuit suspectsZero-range bar (high = low) with a move ≥ 1.9%: probably locked at a price band

Ratio adjustment lets a 50-stock list and exchange-wide counts share one McClellan scale. The summation level depends on where the data starts, so read its direction and zero crossings.

4. Health score (0–100)

ComponentWeightScaling
Participation250.2 × %>20 DMA + 0.4 × %>50 DMA + 0.4 × %>200 DMA
Uptrend ratio1510% → 0, 60% → 100
Participation trend1020-session change in %>50 DMA: −25 pts → 0, +25 → 100
A/D line1520-session (A − D)/(A + D): −0.25 → 0, +0.25 → 100
McClellan10oscillator −100 → 0, +100 → 100
Highs vs lows1510-session new highs ÷ (highs + lows); none of either = 50
Sectors10% of sectors participating

Missing components (no members, short history, --ad-csv) are dropped and the rest re-weighted. "Coverage" shows how much weight was scored, and under 50% the read is A/D-only. Labels: 75+ Broad participation, 60–74 Healthy, 40–59 Mixed, 25–39 Narrow / weak, below 25 Very weak. The scalings are judgement calls, not fitted values, so compare a universe with its own history (history).

5. Reading breadth in India

  • Index vs broad market. The top six names (HDFC Bank, ICICI Bank, Reliance, Infosys, Bharti Airtel, L&T) are about 40% of NIFTY 50, so heavyweight-led rallies are common (2018: smallcaps fell hard while NIFTY made highs). Run one folder three ways: NIFTY 50, NIFTY 500, midcap/smallcap lists. NIFTY near highs with NIFTY 500 %>50 DMA under 40% and falling, or a bearish A/D divergence, is a narrowing market; the script flags an index above its 50 DMA with under 40% of stocks above theirs.
  • Survivorship bias. Today's constituent list omits stocks dropped after falling and includes ones added after rallying, so past breadth looks stronger than it was. Treat long history as rough.
  • Circuit-locked stocks. A stock frozen at its 5/10/20% band counts as an advance or decline but nobody could trade it. Clusters of lower circuits in SME, ASM/GSM or small names exaggerate declines, so check the circuit-suspect line before reading a washout.
  • F&O ban names (open interest above 95% of the market-wide limit) still trade in cash and stay in the counts. Their sharp moves are stock events, not breadth.
  • How many stocks. Each stock is 100/N points: 2 points in NIFTY 50, 0.2 in NIFTY 500. Use 100+ for stable percentages and 200–500 for McClellan and thrust, or use --ad-csv exchange counts. Ten stocks give a sketch, not a signal.

6. How other skills use this

  • market-regime sets the exposure ceiling; breadth adjusts within it. Confirmed uptrend with a score under 40 or a bearish divergence: act one notch lower (fewer, smaller new positions). In a Correction or Rally attempt, count a thrust as supporting evidence for a follow-through day. Score 60+ with %>50 DMA rising backs the uptrend.
  • market-pulse: the "Breadth" line quotes exchange A/D for the day, NIFTY 500 %>50/200 DMA, the score and label, and any divergence or thrust. Don't infer breadth from index points.
  • india-market-analyst: put the breadth label next to the regime in the context line. In a narrow market, prefer leaders in participating sectors and cut size in small caps.
  • breakout-planner: in a narrow tape (under 40% above the 50 DMA) demand stronger relative strength.
  • sector-rotation has per-sector member breadth (its breadth sub-command); this is the market view.

Caveats

  • Stale or suspended stocks (no bar on the last date) are excluded and listed. Use split-adjusted daily bars after 15:30 IST: a raw split looks like a crash, a partial day distorts A/D.
  • Thresholds (0.40/0.615 thrust, 19/39 McClellan, 52-week window) are US-born conventions applied to Indian data. Replay past turns with --as-of (Mar 2020, mid-2022, early 2025) before you rely on them.

Output

Breadth: [universe, N stocks / exchange A/D] to [date] | score s/100 [label] (coverage c%)
%>20/50/200 DMA a/b/c (20-session change) | uptrend u% | A/D adv/dec, 20s net, divergence [flag]
NH/NL today and 10s | McClellan osc, summation trend | thrust [state] | sectors k/m participating
Read: [one line] | Regime fit: [how it shifts the market-regime exposure ceiling]

Educational analysis, not investment advice. Data: [sources + timestamps].

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.