agentleFS
ShubhamkumartrichyShubhamkumartrichy/claude-trading-skills-india/skills/momentum-scanner/SKILL.md

momentum-scanner

Skill64 starsChanged 17 days agoMITView on GitHub

Short-term momentum on NSE/BSE stocks across a whole universe: momentum-burst days (a 4%+ up-close out of a quiet base on rising volume, near the high, above a rising 50 DMA, not already extended), episodic pivots (a 5%+ opening gap on 3x+ volume that holds the gap-day low, with a day-1/day-2 read and a catalyst checklist), and a 1-99 relative-strength rating with the RS line vs NIFTY. Flags upper-circuit closes, 2%/5% band stocks and low delivery, applies a ₹-crore liquidity floor, then ranks a short watchlist and hands the best names to breakout-planner. momentum.py works offline from CSVs.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 0 are distinctive, and 4 are 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.

  • 4

    Very specific

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

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

    Delivery %: If the CSV has an NSE delivery column (DELIV_PER, % Dly Qt to Traded Qty, or delivery quantity), a burst on below-average …

    line 103
  2. Very specific:

    Run ../market-regime/scripts/regime.py status (market-regime).

    line 112
  3. Very specific:

    Run scan on the universe, read the reasons, and drop flagged names.

    line 114
  4. Very specific:

    Run the trading-discipline gate (../trading-discipline/scripts/discipline.py) before any order.

    line 118

≈ “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

Momentum Scanner (NSE/BSE)

Find what is moving now: a burst out of a quiet base, a news gap the market re-prices, or a leader whose RS line makes a new high before its price. This skill finds candidates; breakout-planner reads the base and builds the order.

Commands

python3 scripts/momentum.py burst --csv-dir data/ --days 5 --show-rejects
python3 scripts/momentum.py pivot --csv-dir data/ --catalyst DIXON=results --catalyst KAYNES=order-win
python3 scripts/momentum.py rs --csv-dir data/ --benchmark-csv data/nifty.csv --top 20
python3 scripts/momentum.py scan --csv-dir data/ --benchmark-csv nifty.csv --bands bands.csv --regime confirmed
python3 scripts/momentum.py scan --symbols DIXON,KAYNES,TRENT,POLYCAB --benchmark NIFTY   # yfinance

The universe flags match breakout.py scan: --csv-dir + --pattern (file name = symbol), --csv FILE (repeatable) or --symbols. Also --as-of YYYY-MM-DD (replay a date), --min-turnover-cr (5), --bands FILE (SYMBOL,band with 2/5/10/20, or 0 for F&O names with no fixed band) and --json. Use 260+ daily bars (64 is the minimum for RS). Blank or zero candles are skipped. Bad input exits 2 with a one-line error:.

Reading burst (last 5 sessions, --days)

TestDefault (flag)
MoveClose ≥ +4% vs the previous close (--min-move-pct)
VolumeAbove the previous session and above the 50-day average (the 50 sessions before)
CloseIn the top 70% of the day's range (--min-close-pos)
Quiet base3+ sessions right before it inside an 8% high-low band (--min-base-days, --max-base-pct), with daily ranges no wider than the 20 sessions before (--max-range-ratio 1.0)
Not extendedUp no more than 6% in the 3 sessions before (--max-run-pct)
TrendClose above a 50 DMA that is higher than 10 sessions ago
Liquidity≥ ₹5 crore average traded value over the 20 sessions before the burst, so a one-day spike can't pass an illiquid stock

Score /100: move 25, volume 20, close 15, base length 10, range contraction 15, 50 DMA above 200 DMA 10, delivery at or above average 5, minus flag penalties. Plan reference: buy above the burst-day high, stop at its low. A burst is a 3-5 day trade: book partials fast, exit if the low breaks. --show-rejects lists near-misses and why (extended, illiquid, no base).

Reading pivot (last 10 sessions)

Open ≥ 5% above the previous close (--min-gap-pct) on ≥ 3× the 50-day volume (--min-vol-mult), with no close below the gap-day low since. Day 1: strong if it closed above the open near the high, weak if sellers took it back to the low. Day 2: follow-through, holding, tested the low, or failed. A gap that gets filled, or a stock already up 40%+ in the 60 sessions before, scores lower. The open comes from NSE's pre-open call auction (orders 09:00-09:08, trading from 09:15), so don't chase the first minutes. Buy above the gap-day high or the opening-range high, stop under the gap-day low.

Catalyst checklist. Pass what you have verified: --catalyst SYM=results|order-win|block-deal|approval|rating-upgrade|other (a bare --catalyst results applies to every name). Results beats with margin expansion and order wins large against revenue carry the most weight. For a block deal, check who sold. A gap with no catalyst is a burst, not a pivot, and in small caps it can be operator-driven.

Reading rs

Weighted return = 0.4 × 3-month + 0.2 × 6-month + 0.2 × 9-month + 0.2 × 12-month (63/126/189/252 sessions, cumulative). The latest quarter sits inside all four, so recent strength counts most. The rating is the percentile rank inside your universe, from 1 to 99. Rank NIFTY 500 or a full watchlist, because ten names give coarse ratings. Stocks with less history are re-weighted over the horizons they have and marked partial. With a benchmark, the RS line (stock ÷ NIFTY) is checked for a new 52-week high in the last 5 sessions (--rs-high-days). New high ahead of price is the leader's tell. A price high without an RS-line high is a laggard riding the index. The output also shows where the benchmark itself would rate (use NIFTY Midcap 150 for midcaps).

scan: the watchlist

Bursts, pivots and liquid RS leaders (RS ≥ 85 --leader-rs, within 10% of the high, above the 50 DMA, RS line at a high) go into one list ranked by signal strength (60%), RS rating (30%) and RS-line and high bonuses, minus circuit and band penalties. Each name gets a one-line reason. The top tradable names (no circuit or band flag) get a ready breakout.py check command (see ../breakout-planner/scripts/breakout.py) to find the base, pivot, stop and order. --regime takes the market-regime label and prints the gate: Correction means no new momentum trades.

India specifics

  • Results season (mid-Jan, Apr-May, Jul-Aug, Oct-Nov) is when most real pivots happen. Gaps around the result date are the signal. A burst just before results is a coin toss.
  • Upper circuit. A close at the high on a band-sized move (2/5/10/20%) is flagged UC? (inferred) or UC (band from --bands). Buyers queued unfilled, so the close isn't a price you could pay, and the next open usually gaps. --exclude-locked drops these closes.
  • 2%/5% bands (flag band2/band5, inferred when the tape keeps closing exactly ±2/5%) mean surveillance (ASM/GSM), trade-for-trade (BE/T2T) or SME. These names have no intraday or BTST, need 100% upfront margin and often have no MTF, and their exits can lock at the lower circuit. Operator and SME pumps live here: vertical moves, circuit after circuit, low delivery, no news. Skip them.
  • Delivery %. If the CSV has an NSE delivery column (DELIV_PER, % Dly Qt to Traded Qty, or delivery quantity), a burst on below-average delivery is flagged as intraday churn. --min-delivery-pct makes it a filter.
  • Expiry-week noise. Bursts in the week of the last-Tuesday F&O monthly expiry are tagged expiry-wk. Rollover and short covering in F&O names, and NIFTY's weekly Tuesday expiry, can fake one-day strength. Wait for a second day.

Workflow

  1. Run ../market-regime/scripts/regime.py status (market-regime). Trade bursts only when the regime allows new longs, and stay inside its position count.
  2. Run scan on the universe, read the reasons, and drop flagged names.
  3. Hand the top names to breakout-planner (check, then plan) for the stop, trigger and GTT.
  4. Size with position-sizing (../position-sizing/scripts/risk.py). Bursts are short trades, so check the round trip with trading-costs.
  5. Run the trading-discipline gate (../trading-discipline/scripts/discipline.py) before any order. After entry, trailing-stops takes over.

Output

 # Symbol  Close  RS  vs 52wH  Score  Signals [flags]   +  "Why:" one reason per name
Next: python3 .../breakout.py check --csv SYMBOL.csv --benchmark-csv nifty.csv

Limits: daily bars only, so there's no intraday pre-open data or live circuit status. Bands are inferred unless you pass --bands, and an F&O stock up exactly 10% at its high can be a false UC?. Catalysts are whatever you supply. RS ranks only the universe you give it.

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.