diff --git a/site/index.html b/site/index.html index 0ebdfc8..d47037f 100644 --- a/site/index.html +++ b/site/index.html @@ -1,390 +1,38 @@ - + - - - -Quantum Asset Selection — Quip Network - - - - - - - - - - -
-
- - Qubitrefill - - -
-
- - -
- -
-
-

Bitrefill · Quantum Asset Selection

-

Pay Bitrefill with quantum-picked crypto.

-

- We frame your portfolio as a QUBO — assets as nodes, couplings as edges — - then race classical and quantum solvers to choose the best crypto to spend on gift cards, - mobile top-ups, and eSIMs. -

-
- See how it works - View Repo ↗ -
-
-
- Assets as wells on a quantum cost surface, with reinforcing and opposing couplings -
-
-
- - -
-
-
- -
-
-
- - -
-
-
-

Asset selection as an optimization landscape

-

Find the lowest point on the cost surface

-

- Every basket is a point on a surface shaped by expected return, risk, and market - conditions. The optimizer descends toward the deepest wells — the allocations that - cost the least to hold and trade. -

-
- deeper well = more optimal allocation -
-
-
- A mesh cost surface with several wells, each marked by a point -
-
-
- - -
-
-
-

Formulating the QUBO

-

A graph of assets and their couplings

-

- Nodes are assets; each carries a weight h (its standalone appeal). - Edges are the J couplings between assets — how holding one affects another. -

-

Topology is irregular — a node can couple to many others, not just four.

-
- positive J — reinforcing - negative J — opposing -
-
-
- A graph of crypto assets connected by reinforcing (green) and opposing (red) couplings -
-
-
- - -
-
-

From problem to persisted portfolio

-

One straight line: estimate → solve → report

-

- Build a PortfolioProblem from recent returns (μ = trend, Σ = risk) and the - slider knobs (γ, w_max, w_min); race the solvers for the best feasible weights; value money - at spot, split it into token units, save, and report. + + + + Qubitrefill has moved + + + + + + + +

+ Qubitrefill has moved to + + quipnetwork.github.io/qubitrefill + .

- Six-step pipeline: build problem, race solvers, reallocate, persist, report, emit events -
-
- - -
-
-
-

Why pool optimization pays off

-

Less slippage now, more capital later

-

- Optimizing across the pool routes weight toward the assets that are cheapest to trade - under current conditions — so each rebalance loses less to slippage. -

-

- That edge compounds: you still buy the least-weighted assets, but you keep more of every - dollar over the long run. -

-
+8.9%capital retained vs. naive equal split (12 rebalances)
-
-
- Capital retained over rebalances: optimized routing vs naive equal split, +8.9% retained -
-
-
- - -
-
-
-

Assets as wells, constraints as nodes

-

The optimizer settles into the deepest wells

-

- Each asset bends the cost surface into a well; couplings - (green = reinforcing, - red = opposing) and the constraint nodes between - them shape it. -

-

The search settles into ZEC and SOL — the two deepest wells.

-
-
- Crypto assets rendered as wells on a 3D cost surface, with ZEC and SOL deepest -
-
-
- - -
-
-
-

Get started

-

Three steps to first run

-

- Install the skill into Claude Code, connect it to the hosted quantum backend, then - ask in plain English. There's no backend to run yourself. -

-
-
-
- 1 -

Connect & install

-

Connect the bitrefill MCP, then drop qupick into Claude Code.

-
claude mcp add --transport http bitrefill \ - https://api.bitrefill.com/mcp -mkdir -p .claude/skills -cp -R skills/qupick .claude/skills/
-
-
- 2 -

Register & connect

-

- Point .mcp.json at the hosted server and register once — your API key is - emailed to you. Paste it into the Authorization header and reconnect. -

-
url https://qupick.quip.network/mcp -auth Bearer <key-from-email>
-
-
- 3 -

Just ask

-

- The agent picks your worst-performing asset, settles the bill, then retunes the rest of - the basket around it. -

-
"Buy a $20 Steam card and pay with my worst-performing crypto."
-
-
-
- Full setup guide → README ↗ - View Repo ↗ -
-
-
- - - - - - - \ No newline at end of file + +