Macro watching — tape vs tone

1. The headline tape

One call, full breakdown plus 5d and 30d rolling for context.

curl -sH "$H" "https://genka.dev/latest/fiidii/latest" | jq '.data | {trade_date, fii, dii, rolling}'

Live snapshot 2026-04-24:

DII absorbed ~86% of the FII cash exit in trailing 30 days — the typical India pattern. The actionable detail is in fii_opt_cr (-83,054 cr): when option positioning is that one-sided, vol is mispriced. The 30d cumulative via /fiidii/cumulative gives the same data as a time-series chart against Nifty close.

2. Is the macro tone in concalls bearish or bullish?

Cross-reference: if FII flow is -₹70K cr over 30 days, are managements echoing the same caution? Search concalls for the macro-headwind vocabulary.

for q in "US+rate+cuts" "China+demand" "global+uncertainty" "input+cost+inflation"; do
  curl -sH "$H" "https://genka.dev/latest/concalls/search?q=$q&limit=3" \
    | jq -r ".data | { q: \"$q\", count: .count, top: [.results[0:3][].symbol] }"
done

Each query returns a count + top 3 companies talking about that theme. Patterns: q=US+rate+cuts tops with MAHABANK on FII-flow commentary; q=China+demand tops with SAIL + HEG (steel spillover); q=premiumisation tops with HINDUNILVR (FMCG resilient). When the tape is negative AND multiple sectors are calling out the same headwind, signal is real.

3. The synthesis call

When you want a paragraph, not a hit list. concall_answer runs RAG over the corpus + returns citations linked back to fragment ids.

curl -sH "$H" -X POST -H "Content-Type: application/json" \
  -d '{"question":"What are Indian managements saying about FII outflows and rupee weakness over the last quarter?","limit":5}' \
  https://genka.dev/latest/concalls/answer

Returns {answer, citations[]}. Each citation carries an id so [N] markers in answer are dereferenceable via /concalls/{id}. LLM tier — ~1.2× upstream cost. Use this when you want "the management voice condensed into a paragraph"; for everything else, plain concall_search is faster and cheaper.

4. The early-reversal signal

A negative 30d fii_fut_cr with deep-negative fii_opt_cr is the “short futures, long downside puts” book. When that flips to positive on a single day's data, that's an early reversal signal.

curl -sH "$H" "https://genka.dev/latest/fiidii/cumulative" | jq '.data.data | reverse | .[0:5] | .[] | {trade_date, daily_fii_cash: .daily.fii_cash_cr, daily_fii_fut: .daily.fii_fut_cr, nifty: .market.nifty_close}'

Returns the most recent 5 trading days as a series. Watch for a single-day positive flip in daily.fii_fut_cr alongside Nifty stabilising; that's typically the first sign of position cover before the cash flow turns. The cumulative.fii_cash_cr running total tells you how far they've already leaned out — the further negative, the more the squeeze potential.