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

market-regime

Skill64 starsChanged 17 days agoMITView on GitHub

Reads the market regime of NIFTY, BANKNIFTY, SENSEX (or any index/ETF CSV) from the index's own tape: distribution-day clusters with 25-session/5%-rally expiry, the correction → rally attempt → follow-through cycle, 50/200 DMA and 52-week drawdown context, and optional India VIX. Outputs a 0–100 risk score, a regime label, an exposure ceiling (% of capital, max new positions) and the levels that would change the label. Handles Yahoo's missing index volume via ETF/futures proxy or a clearly flagged price-only mode. regime.py works offline from CSV.

3 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. 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Don't recommend fresh breakouts in Correction.

    line 93

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “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 Regime

Decide how much risk the market allows before picking stocks. Most swing breakouts fail when the index is distributing or correcting, so the regime sets the exposure ceiling for every other skill.

Commands

python3 scripts/regime.py status --csv nifty.csv --volume-csv niftybees.csv --vix 14.2
python3 scripts/regime.py status --symbol NIFTY --vix 14.2          # yfinance; ETF volume auto-fetched
python3 scripts/regime.py distribution --csv nifty.csv --volume-csv niftybees.csv --all --stalling
python3 scripts/regime.py ftd --csv banknifty.csv --volume-csv bankbees.csv --events 15
python3 scripts/regime.py trend --symbol SENSEX
python3 scripts/regime.py status --csv nifty.csv --as-of 2026-06-05 --json   # replay a past date

Inputs match ../technical-analysis/scripts/ta.py (--csv, --candles, --symbol). Useful flags: --ftd-pct, --dd-pct (0.2), --dd-window (25), --correction-pct (5), --ignore DATE,DATE, --expiry-day tue|thu, --price-only, --no-proxy, --json. Run after 15:30 IST on daily bars.

Volume: the India data problem

Yahoo's ^NSEI and ^NSEBANK usually carry zero or patchy volume. The script checks this (≥80% of recent sessions must have non-constant volume) and then:

  1. --volume-csv / --volume-symbol: volume from another series matched by date. Use NIFTYBEES or BANKBEES (ETF), or continuous NIFTY futures exported from your broker or TradingView. Futures volume jumps in the monthly rollover week (last Tuesday), so treat DDs there with suspicion.
  2. With --symbol NIFTY or BANKNIFTY it fetches NIFTYBEES/BANKBEES volume automatically.
  3. Otherwise it runs price-only and says so. In this mode a heavy-selling day is a fall of at least 0.75% closing in the lower half of its range, and a follow-through needs only the gain and an upper-half close. Report the result as lower-confidence. Never call it "distribution" without volume.

Sessions after an abnormally thin one (Muhurat trading, special Saturday sessions) are not judged, because beating a thin session's volume proves nothing.

Signals and how to read them in India

SignalRule in regime.pyIndian reading
Distribution dayClose −0.2% or worse on higher volume than the prior session. Counts for 25 sessions incl. the day, or until a close 5% above it4–5 inside 25 sessions = institutions selling into strength. Carries more weight when NSE provisional data shows a multi-day FII cash-selling streak. Heavy DII/SIP buying often turns FII selling into flat or small-down days
Stalling day (--stalling, half weight)Gain of 0–0.4% on higher volume near the 25-session high, closing in the lower half of the rangeChurning at the top. Common before results-season disappointments
CorrectionClose 5% below the cycle peakStarts the bottoming watch
Rally attemptFirst up close after the correction low. Fails if that low is undercutDays 1–3 prove nothing. Short-covering into weekly expiry often creates them
Follow-through (FTD)Day 4 or later of the attempt, gain ≥ threshold on higher volumeConfirms a new uptrend. It fails if price later closes below the rally low; many do
Trend/durationClose vs 50/200 DMA, 50 DMA slope, drawdown from 52-week high, sessions since high/lowDepth: <3% near highs, 3–5% pullback, 5–10% correction, 10–20% deep, 20%+ bear
India VIX (--vix)<13 calm, 13–17 normal, 17–22 elevated, 22–30 stressed, 30+ panicVIX ≥ 22 caps exposure at 50%

FTD threshold default. The script uses 1.5 × the 250-session standard deviation of daily returns, clamped to 1.0–2.0%. NIFTY's daily σ is usually 0.7–0.9%, so this lands near 1.1–1.3%, and higher for BANKNIFTY. A fixed US-style 1.5–1.7% hurdle would almost never fire on NIFTY. Pass --ftd-pct to fix it (for example 1.2 for NIFTY or 1.5 for BANKNIFTY).

Event days. Pass Budget day, RBI policy days, election-result days, MSCI/FTSE rebalance days (huge closing-auction volume) and Muhurat sessions to --ignore if they distort volume. DDs that land on the weekly-expiry weekday are tagged (--expiry-day: tue for NIFTY, the default; thu for SENSEX). Weigh them lighter if your volume source is futures or the ETF. A market-wide circuit-breaker day (±10/15/20% on NIFTY or SENSEX, see indian-market-structure) is tagged. It means the regime has already broken, so don't wait for a DD count.

Regime → exposure

LabelWhenExposure ceilingNew positions
Confirmed uptrendNo correction, < 5 DDs, above 50 DMA, score < 50100% (score < 25) / 80%5 / 4
Confirmed uptrend (fresh FTD)FTD in the last 10 sessions and score < 6050%3 starter-size
Uptrend under pressure≥ 5 DDs in 25 sessions, ≥ 4 in 15, below 50 DMA, or score ≥ 5060% / 40% if score ≥ 502 / 1
Uptrend under pressure (fresh FTD)FTD in the last 10 sessions but score ≥ 6030%2 pilots
Rally attemptBouncing from the correction low, no FTD yet25%1 pilot, half size
Correction≥ 5% below the cycle peak, no rally attempt20% (10% if 10%+ deep)0

The risk score combines distribution (35 points), trend (30), drawdown (20) and VIX (15 if given), scaled to 100, minus 10 during a live follow-through. It is a heuristic, not a probability. "What would change the label" lists the exact closes to watch: DD expiry and clearing levels, the 50 DMA, the correction trigger, the FTD close needed, and the rally low.

How other skills use the regime

  • india-market-analyst: put the label, score and ceiling in the first line of any trade view. Don't recommend fresh breakouts in Correction.
  • position-sizing: scale risk per trade with the regime. Confirmed: 1%. Under pressure: 0.5–0.75%. Rally attempt or fresh FTD: 0.25–0.5% pilots. Correction: no new risk. Total deployed capital stays below the ceiling.
  • trailing-stops / stop-loss-strategies: under pressure, tighten trails to 1.5 × ATR and book partial profits. In Correction, honour every stop.
  • technical-analysis: a stock setup with a weak regime needs strong relative strength vs NIFTY.
  • fno-options-analysis: a correction with high VIX favours defined-risk spreads over naked selling.

Caveats

  • The index is not your portfolio. Midcaps and smallcaps can diverge for months. Run it on NIFTY MIDCAP 100 or SMALLCAP data too if that is where you trade.
  • It uses daily closes only. Intraday reversals and gap risk (global cues, GIFT Nifty) are not seen.
  • ETF volume includes market-maker and arbitrage flow. It is a participation proxy, not institutional flow. Cross-check with FII/DII figures via market-pulse.
  • Thresholds are conventions, tuned for NIFTY's lower volatility. Check --as-of replays on past corrections (Mar 2020, Oct 2024–Mar 2025) before you trust the defaults.

Output

Regime: [label] | risk score s/100 | data as of [date] IST close, volume: [source / price-only]
Exposure ceiling: x% | max new positions: n | stance: ..
DDs: d5 a | d15 b | d25 c (oldest drops out in k sessions; clears ≥ ..) | FTD state: ..
Trend: vs 50 DMA ..% | vs 200 DMA ..% | drawdown ..% (depth class), .. sessions since high
Watch: [levels that change the label]

Educational analysis, not investment advice.

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.