Board of Advisers · Product Pitch

The MTG Portfolio, and what to fund next

Nineteen Magic: The Gathering assets already ship. Here is the honest read on each, then a ranked slate of the next builds that would actually move the needle. Two of the eight pitches have since been funded and delivered, and one of them came back with a finding that changes its own ceiling.

TO: The CEO  ·  FROM: Product  ·  RE: turning one-offs into one compounding line  ·  original 2026-07-08  ·  status revision 2026-07-18

Status revision · 2026-07-18

Ten days on, two pitches are delivered and live. Every route below was checked this morning and returned HTTP 200.

PitchStatusWhat shipped
03SHIPPED Craft ranker unpinned from Mardu. MTGA Craft Planner takes any color combo, ranks 25 archetypes by mythic win rate, and walks a collection-aware craft path over 60 staples.
08SHIPPED Creator database live at Creator Standard Decks: 83 decks, 13 creators, each with decklist link, colors and archetype. Today it also sorts by price to build and wildcards to build.
01, 02, 04, 05, 06, 07openUnchanged. Pitch 01 is still the recommendation.
The finding that matters: Pitch 08 asked for "a slow polite Moxfield harvester." Politeness turns out not to be the obstacle. Moxfield answers 403 behind Cloudflare on both its API host and its public deck page after the 301, and Aetherhub answers 403 the same way. Rate limiting does not change that. Of 83 decks, only the 7 on MTGGoldfish expose a machine-readable list, so 76 read "list walled" rather than carry a fabricated price. The remaining decklists need a signed-in browser pass, which is a different and more expensive ask than the one this memo originally made.

What we already have

Corrected inventory: roughly nineteen live MTG routes, not the six I first showed. The earlier count missed every "marvel", "commander", and codenamed page. Grouped below, each link verified live.

Data pipeline · 10 pages

Marvel set suite

Real Scryfall cards joined to LLM set-review grades and dated market signals, the crown jewel.

Dossier · Standard review · Commander review · Spec targets · Spot-check audit · Sealed builder · Draft targets · Jumpstart pair ranker · Best-cards for Standard · Mardu Thanos sealed
Arena rankers · 2 pages

MTGA craft planners

Wildcard crafting priority for Arena, ranked by winning-deck count, read against your collection. The planner takes any color combo; the original is pinned to Mardu.

Craft Planner (any colors) · Mardu ranker
Creator database

Creator Standard Decks

83 recent Standard decks from 13 tracked creators, filterable by color and by spike-or-brew, sortable by paper price to build and by Arena wildcards to build.

Open the database →
Table trackers · 5 pages

Life & game-state trackers

In-game companions for the table, multiplayer life and score.

Life Station (4P) · Life Station v3 (3P) · Two-Headed Giant · Commander Life Toolbox · Write It Down
Commander decks · 2 pages

Commander deck trackers

Per-deck tracker pages built for specific Commander archetypes.

Gossamer (Faerie) · Gravecrawl (Zombie)
Local & community

Events & community

Local play discovery and a community-deck status page.

KC MTG Events · Community Deck status
Creative asset

Magic: The Badgering

Thirty movie-quote comeback cards, a finished party game. Sellable as-is.

See the deck →

The pitch, ranked

Ordered by value-to-effort. The top three are the ones I would fund this quarter.

PITCH 01 · fund first

Grade the graders: the post-release accuracy backtest

Builds on: Marvel suite · the dossier admits it does NOT validate card-eval accuracy or market outcome.

The Marvel set has now released. Re-pull real market prices and score every LLM grade and every spec pick against what actually happened. Output one page: "our A-tier called X, the market did Y, our hit rate was Z%."

Why fund it: this is the single credibility unlock. Right now the whole suite is a schema-validated opinion. A proven track record turns it into a reason to trust the next set review, and it is the only pitch here that makes every other MTG product more valuable at once.

Value very high Effort low · data + one join already exist Reuses fetch_signals.py
The ask: one build pass to re-pull prices and publish the scorecard.
PITCH 02 · fund first

Turn the Marvel pipeline into a reusable Set Review Engine

Builds on: Marvel suite · the pipeline (fetch → grade → join signals → dashboards → spec targets → deckbuilder) is already set-agnostic.

Parametrize the pipeline by set code. Point it at the next Standard or Commander set during spoiler season and it produces the same five artifacts automatically. Four to six sets ship per year.

Why fund it: this is the productization move. We built the hard part once for Marvel. Every future set becomes a repeatable, timely drop instead of a from-scratch project, and paired with Pitch 01 each drop carries a track record.

Value high · recurring Effort medium · refactor, not rebuild Reuses generate_workflow.py, build_dashboards.py
The ask: lift the set code into config, run it on the next spoiled set as the pilot.
PITCH 03 · delivered 2026-07-08

Generalize the craft ranker beyond Mardushipped

Builds on: MTGA Mardu Craft · the engine already ranks by deck-count, it is just pinned to three colors.

Let the user pick any colors or archetype and get the same wildcard priority list off live metagame decklists. Then add the collection-aware step it already has the data for: "given your wildcards, here is the cheapest path to a Tier-1 deck."

Why fund it: the Mardu page proves the format works and people use it. Unpinning the colors turns one deck's tool into every Arena player's tool with almost no new logic.

Value high Effort low · parametrize existing ranker Reuses update_mtga_mardu.py, collection.json
The ask: add a color/archetype selector and the shortest-craft-path planner.
Delivered. Both halves of the ask are live in the MTGA Craft Planner: pick any color combo, get every archetype that fits inside it ranked by mythic win rate, then a collection-aware craft path ordered by how many top archetypes run each card, with a running wildcard cost. Built from the Untapped pull over 25 archetypes and 60 staples. The honest limit the page states itself: the staple list skews colorless, so colors drive the archetype ranking harder than they drive the craft list.
PITCH 04 · next up

Make KC Events live, with a "today" button and calendar subscribe

Builds on: KC MTG Events · the "today" button was already requested in a prior session.

Wire real store event feeds, add the "today" filter, and ship an ICS export so events land on the user's phone calendar. Then the cross-sell: flag which nearby stores stock current spec targets under market.

Why fund it: it is the one asset with a concrete unmet request already on record, and it is the natural bridge that connects the events data to the spec-target data.

Value medium-high · local, sticky Effort medium · needs a feed source
The ask: pick a store feed, add today + ICS, then link spec targets to inventory.
PITCH 05 · next up

Spec-target price alerts

Builds on: Marvel spec-target sheet · today it is a static snapshot.

Re-pull signals on a schedule and fire a phone alert when a spec target crosses a buy or sell threshold. The daily-pull machinery already exists elsewhere in the toolchain.

Why fund it: a snapshot is a guess, an alert is a reason to open the app. It converts a one-time page view into a standing utility.

Value medium Effort low · scheduler + threshold
The ask: a daily signal re-pull plus a threshold-crossing notifier.
PITCH 06 · easy money

Package Magic: The Badgering as a print-and-play product

Builds on: Magic: The Badgering · thirty cards, already designed.

Export the deck as a print-ready PDF and list it on the existing shop lane. The thirty cards already exist, so this is only layout and a listing.

Why fund it: it is the lowest-effort item on the board and the only one that can return money directly. A built asset, sitting idle.

Value medium · direct revenue Effort very low
The ask: a print PDF and one shop listing.
PITCH 07 · the tie-together

One MTG hub, wired into the canonical board

Builds on: all five assets, which currently live as scattered routes.

A single launcher tile linking every MTG page, added into the existing indexes-board rather than a new board.

Why fund it: five good pages nobody can find is worth less than five good pages behind one door. It also makes the portfolio legible for exactly this kind of review next time.

Value medium · discoverability Effort very low
The ask: one hub tile, folded into /indexes-board.
PITCH 08 · delivered 2026-07-11, extended 2026-07-18

The creator-sourced winrate databaseshipped, with a caveat

Builds on: the yt_trans skill (already pulled an mtg-mono-black transcript corpus) + the Marvel/Mardu data layer + the YouTube politeness rule (slow, randomized pulls).

Comb the latest content from named MTG creators. For each video, grab the linked Moxfield decklist slowly and politely, one at a time. Aggregate a database keyed on win rate, color identity, and archetype, then mine the transcript for what the creator actually claims about the deck. Match the spoken claims to the list and attach a confidence score to each entry.

Why fund it: our current grades and craft ranks lean on preorder prices and raw deck counts. Creator winrate plus a decklist plus the creator's own words is a richer, opinion-weighted signal, and the confidence score keeps a hype video from outvoting a tested list. It feeds Pitch 01, Pitch 03, and Pitch 05 at once.

Value high · new signal source Effort high · scraper + join + confidence model Reuses yt_trans, Moxfield lists
The ask: a slow polite Moxfield harvester, a winrate/color/archetype DB, and a transcript-claim confidence match.
Delivered. Creator Standard Decks is live: 13 creators, 83 decks, each with its decklist link, colors, archetype, and a spike-or-brew read. The 2026-07-18 pass added the two build-cost sorts the craft planner implied. Price to build is paper USD for the full list from the cheapest Scryfall printing of every card. Wildcards to build is what Arena charges after subtracting the 8,795 owned card names resolved out of the collection export, bucketed by rarity and sorted on rare-plus-mythic, which is the number that actually gates a build. Range across the seven priced decks runs $90 to $575, and 4 to 30 rare-plus-mythic.
Two caveats the CEO should hear. First, the harvester premise failed: Moxfield and Aetherhub both return 403 behind Cloudflare, so 76 of 83 decks show "list walled" instead of a price. Politeness was never the blocker, and unwalling them means a signed-in browser pass. Second, the transcript-claim confidence model in the original ask was never built. YouTube's transcript API returned RequestBlocked, so colors and archetype come from video descriptions, not spoken claims. The opinion-weighted signal this pitch was really about is still open.

If you fund one thing

The recommendation has not moved. Fund Pitch 01. It is still low effort, still the credibility unlock, and now that 03 and 08 have shipped it is the only remaining item that raises the value of everything else at once. Revised sequence, with the two delivered items struck:

  1. Backtest the Marvel grades against real post-release prices, publish the scorecard. Still the one to fund.
  2. Parametrize the pipeline and pilot it on the next spoiled set.
  3. Unpin the craft ranker from Mardu, add the shortest-craft-path planner. Shipped 2026-07-08.
  4. Ship the KC Events "today" button and ICS subscribe.
  5. Package Badgering as a print-and-play while the above build, then wire the MTG hub.

What the two shipped pitches taught us

Both delivered items came back with a constraint worth carrying into the next funding round.

Three constraints, priced

ConstraintEvidenceWhat it costs us
Decklist hosts are walled 403 Moxfield · 403 Aetherhub · 200 MTGGoldfish Only 7 of 83 decks can be priced. Any pitch that assumes bulk decklist access needs a signed-in browser budget, not a politeness delay.
Transcripts are blocked RequestBlocked The confidence model that made Pitch 08 interesting is unbuilt. Descriptions carry colors and archetype; they do not carry what the creator claims about win rate.
The collection snapshot ages stamped 2026-04-22 Wildcard math drifts the moment you craft anything. Mitigated for now with an editable budget field on the page rather than a rebuild.

The honest read: the data layer we own outright is Scryfall, which is why price-to-build works for every deck whose list we can see, while win rate and creator opinion stay out of reach. A pitch to fix decklist access would unlock Pitch 08's real value and is not on this slate yet.