value-complex

Name: value-complex
Author: mphinance

15stars3forks

How to install value-complex?

mphinance/alpha-skills/value-complex

$npx -y skills add mphinance/alpha-skills --skill value-complex

Installs into the current project.

Summary

Avantis value-factor tracker — small-cap (AVUV) and large-cap (AVLV) value in one read, straight from TickerTrace's per-fund holdings. Groups by fund tier and by position lifecycle (new/exited, conviction tilt, up-streaks), ranked by weight, and surfaces the sector themes both fu

Files · 1

View on GitHub

SKILL.md

1	# Value Complex — Avantis small + large value, by weight
2
3	This skill reads Avantis's two flagship value ETFs straight from TickerTrace's
4	per-fund endpoint (`/fund/<TICKER>`) and lays out what changed under the hood,
5	grouped the way the value factor actually moves:
6
7	- AVUV — small-cap value ($12.5B, ~794 holdings) — the flagship.
8	- AVLV — large-cap value ($3.2B, ~272 holdings).
9
10	> This is the per-fund holdings view that the daily stock-recap skill does
11	> not pull. stock-recap only hits TickerTrace's cross-fund aggregates
12	> (`/briefing`, `/institutional`, `/divergences`). This skill is the deep,
13	> by-weight, per-fund value read.
14
15	## What it groups, and why
16
17	For each fund (small tier → large tier), three lifecycle buckets:
18
19	1. 🆕 New / Exited — positions TickerTrace flags `type=NEW` or `REMOVED`.
20	Passive Avantis funds rarely open or close a name day-to-day, so when they do
21	it's meaningful. This is the literal "new adds."
22	2. 📈 Conviction tilt — active over/under-weighting, inflow-stripped.
23	Every name in an inflow day shows a positive `sharesDelta` (new money buys the
24	whole book pro-rata), so raw adds are noise. We compute each name's share growth
25	`sharesDelta/previousShares`, take the fund-wide median as the inflow tide,
26	and rank by the excess over that tide. Positive excess = the fund is genuinely
27	leaning into the name beyond mechanical inflow; negative = leaning out.
28	3. 🔥 Up-streaks — names on multi-day winning runs (TickerTrace `streaks`).
29	This is the cleanest momentum signal — independent of fund mechanics entirely.
30
31	Then the payoff, 🔗 cross-tier synthesis:
32	- Sector convergence — sectors that are streaking and/or tilting up in
33	both the small and large fund. (Exact-ticker overlap is rare because the
34	funds hold different cap tiers, so convergence is read at the sector/theme
35	level — e.g. "transports leading in both.")
36	- Flow tell — all-buys with zero trims = inflows into value (creations,
37	risk-on); a genuine mix of adds and trims = active rebalancing.
38
39	## How to run it
40
41	```bash
42	node "$HOME/.claude/skills/value-complex/scripts/gather.mjs"
43	```
44
45	- Writes `runs/<date_time>/report.md` + `raw/fund_AVUV.json`, `raw/fund_AVLV.json`.
46	- Prints the report to stdout and a final JSON line with `{reportPath, rawDir}`.
47	- No key needed — `/fund/<TICKER>` is public on TickerTrace. Presents a browser
48	User-Agent (the edge WAF 403s a bare Node UA). ~5–10s.
49	- Override the base with `TICKERTRACE_API_URL`; funds with `VC_FUNDS=AVUV,AVLV,AVMV`.
50
51	## Present it to Mike
52
53	Read the generated `report.md` and give a tight read, in this order:
54
55	1. The flow tell first — are both funds taking inflows (value getting bought)
56	or net trimming? That frames everything.
57	2. New / Exited names per fund — rare, so always call them out by name.
58	3. Conviction tilt — the inflow-stripped over-weights. These are the names the
59	fund is actually leaning into, not just buying with the tide. Name the top 3–4
60	per tier and what sector they're in.
61	4. Streaks — the momentum leaders, longest runs first.
62	5. The cross-tier theme — the one or two sectors strong in both small and
63	large value. This is the headline ("the whole value complex is leaning into X").
64	Tie it to the tape: what's it rotating out of (usually mega-cap growth).
65
66	## Notes / gotchas
67
68	- `weightDelta` is not a clean signal on its own — it blends price drift with
69	share-count change. The report leads with the inflow-stripped excess instead;
70	it shows weightDelta as secondary context only. Don't present weightDelta as
71	"conviction" without the excess check.
72	- Data is daily and can lag — the report prints TickerTrace's `asOfDate`;
73	surface it so Mike knows how fresh it is.
74	- Cross-tier convergence is sector-level, not ticker-level — small and large
75	funds rarely share a ticker; agreement shows up as a shared theme.
76	- Each run is a point-in-time snapshot under `runs/`; nothing is overwritten and
77	old folders are safe to delete (gitignored under `.claude/`).

People who liked this also viewed

6 picks

Type

value-complex

15stars3forks

View on GitHub

How to install value-complex?

mphinance/alpha-skills/value-complex

$npx -y skills add mphinance/alpha-skills --skill value-complex

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Value Complex — Avantis small + large value, by weight
2
3	This skill reads Avantis's two flagship value ETFs straight from TickerTrace's
4	per-fund endpoint (`/fund/<TICKER>`) and lays out what changed under the hood,
5	grouped the way the value factor actually moves:
6
7	- AVUV — small-cap value ($12.5B, ~794 holdings) — the flagship.
8	- AVLV — large-cap value ($3.2B, ~272 holdings).
9
10	> This is the per-fund holdings view that the daily stock-recap skill does
11	> not pull. stock-recap only hits TickerTrace's cross-fund aggregates
12	> (`/briefing`, `/institutional`, `/divergences`). This skill is the deep,
13	> by-weight, per-fund value read.
14
15	## What it groups, and why
16
17	For each fund (small tier → large tier), three lifecycle buckets:
18
19	1. 🆕 New / Exited — positions TickerTrace flags `type=NEW` or `REMOVED`.
20	Passive Avantis funds rarely open or close a name day-to-day, so when they do
21	it's meaningful. This is the literal "new adds."
22	2. 📈 Conviction tilt — active over/under-weighting, inflow-stripped.
23	Every name in an inflow day shows a positive `sharesDelta` (new money buys the
24	whole book pro-rata), so raw adds are noise. We compute each name's share growth
25	`sharesDelta/previousShares`, take the fund-wide median as the inflow tide,
26	and rank by the excess over that tide. Positive excess = the fund is genuinely
27	leaning into the name beyond mechanical inflow; negative = leaning out.
28	3. 🔥 Up-streaks — names on multi-day winning runs (TickerTrace `streaks`).
29	This is the cleanest momentum signal — independent of fund mechanics entirely.
30
31	Then the payoff, 🔗 cross-tier synthesis:
32	- Sector convergence — sectors that are streaking and/or tilting up in
33	both the small and large fund. (Exact-ticker overlap is rare because the
34	funds hold different cap tiers, so convergence is read at the sector/theme
35	level — e.g. "transports leading in both.")
36	- Flow tell — all-buys with zero trims = inflows into value (creations,
37	risk-on); a genuine mix of adds and trims = active rebalancing.
38
39	## How to run it
40
41	```bash
42	node "$HOME/.claude/skills/value-complex/scripts/gather.mjs"
43	```
44
45	- Writes `runs/<date_time>/report.md` + `raw/fund_AVUV.json`, `raw/fund_AVLV.json`.
46	- Prints the report to stdout and a final JSON line with `{reportPath, rawDir}`.
47	- No key needed — `/fund/<TICKER>` is public on TickerTrace. Presents a browser
48	User-Agent (the edge WAF 403s a bare Node UA). ~5–10s.
49	- Override the base with `TICKERTRACE_API_URL`; funds with `VC_FUNDS=AVUV,AVLV,AVMV`.
50
51	## Present it to Mike
52
53	Read the generated `report.md` and give a tight read, in this order:
54
55	1. The flow tell first — are both funds taking inflows (value getting bought)
56	or net trimming? That frames everything.
57	2. New / Exited names per fund — rare, so always call them out by name.
58	3. Conviction tilt — the inflow-stripped over-weights. These are the names the
59	fund is actually leaning into, not just buying with the tide. Name the top 3–4
60	per tier and what sector they're in.
61	4. Streaks — the momentum leaders, longest runs first.
62	5. The cross-tier theme — the one or two sectors strong in both small and
63	large value. This is the headline ("the whole value complex is leaning into X").
64	Tie it to the tape: what's it rotating out of (usually mega-cap growth).
65
66	## Notes / gotchas
67
68	- `weightDelta` is not a clean signal on its own — it blends price drift with
69	share-count change. The report leads with the inflow-stripped excess instead;
70	it shows weightDelta as secondary context only. Don't present weightDelta as
71	"conviction" without the excess check.
72	- Data is daily and can lag — the report prints TickerTrace's `asOfDate`;
73	surface it so Mike knows how fresh it is.
74	- Cross-tier convergence is sector-level, not ticker-level — small and large
75	funds rarely share a ticker; agreement shows up as a shared theme.
76	- Each run is a point-in-time snapshot under `runs/`; nothing is overwritten and
77	old folders are safe to delete (gitignored under `.claude/`).

People who liked this also viewed

6 picks

Type