Open the app

Your whole portfolio. One limit. Published in advance.

Borrow against every Stock Token you hold, with the liquidation price stated before you sign and a price that keeps moving when the market doesn’t.

Open the appTestnetWhat we measured

Tapehouse is not a broker. A Stock Token is a debt security of its issuer, not a share: it follows the price and gives you no shareholder rights.

TAPEHOUSE

Your liquidation price, stated before you sign

141.08NVDA

If NVDA falls to 141.08 we sell 12% of your SPY, not your whole position.

10,000.00 USDG borrowed, backed by all four holdings as one position.

Holdings backing the loan
TokenPriceBandValue
NVDA182.50±1.1%21,900.00
TSLA396.12±1.6%17,825.40
AAPL241.08±0.9%14,464.80
SPY766.14±0.6%13,790.52
Whole portfolio67,980.72

Example portfolio · figures are illustrative

You own it. You just can’t use it.

  1. A tax bill for touching your own money.
  2. Four holdings, treated like four strangers.
  3. A loan that still thinks it’s Friday afternoon.
  4. A Sunday when your own portfolio won’t pick up.
  5. Learning your liquidation price the day it arrives.

Borrowing so you don’t have to sell is not a niche. At Morgan Stanley’s wealth business, clients owe $109.2 billion in securities-based and other lending, against $31.2 billion in margin loans. On-chain, every Stock Token is still priced alone. Where that comes from.

The market closes on Friday. Your loan doesn’t.

We measured it before we built it. Every weekend the reference price feed on this chain goes silent for 55.7 to 80.8 hours, while an around-the-clock source kept answering 288 times out of 288. People don’t stop either: 29.9% of SPY transfers happen on a Saturday or a Sunday.

55.7hours since the reference feed last spoke

FRISATSUNMONReference equity feedAround-the-clock sourceFRI 16:18 UTCMON 00:00 UTC288 OF 288 TEN-MINUTE POINTS

One measured weekend, 28–31 August 2026, drawn to scale. Over the Labor Day weekend the same silence lasted 80.8 hours. How we measured it.

They price one stock. We price the portfolio.

Most lenders here apply one flat limit to each Stock Token and add the results up. A rule like that cannot tell a spread portfolio from a single bet. Move the weights and see what it misses.

Four blocks of different sizes standing on one beam, balanced level on a single fulcrum.
32%
26%
22%
20%

Priced one by one

42,487.95

62.5% · USDG

Priced as a portfolio

48,745.67

71.7% · USDG

Spread across holdings that don’t fall together, the portfolio carries more than its parts.

Illustrative model on a 67,980.72 USD example portfolio. It is not an offer of credit. The parameters Tapehouse lends on are published with the backtest. See the assumptions.

A number you can check. Not a promise.

If you lend, build a wallet, or run a protocol on this chain, you need a price for Stock Tokens on a Sunday. Here is ours, with everything required to prove us wrong.

A plug with three pegs hovering just above the block it fits into.
How the band is built
Three things widen it: live sources that disagree, a price that is moving fast, and fewer sources answering. A source that has gone quiet for the weekend is set aside, never averaged in. The parameters are public, so anyone can recompute last Sunday’s band and check it against ours.
What happens when a source fails
The band widens and the seal says DEGRADED, with the count of sources still alive. It never fills the hole with a number it doesn’t have.
How you integrate it
Each feed answers the aggregator interface your contracts already read, and adds one call that returns the band, the session and the live-source count.
What changes, and when you hear about it
Parameters change only with notice, on-chain, ahead of taking effect. You cannot build on something that moves without warning, so it doesn’t.
Consumer.solInterface preview
// Read a Stock Token price together with its band.ITapehouseFeed feed = ITapehouseFeed(NVDA_FEED); // The interface your contracts already read.(, int256 price, , uint256 updatedAt, ) = feed.latestRoundData(); // One more call: the band, the session, the live sources.(uint256 lower, uint256 upper, Session session, uint8 live)    = feed.latestBand(); require(session != Session.Degraded, "source degraded");// Lend against the conservative edge, not the midpoint.uint256 collateralPrice = lower;

The interface is a preview and may change before mainnet.

We were wrong 8 of 30 times. Here they are.

We gave our band the worst input we could find, the on-chain exchange price on its own, across 9 weekends, and graded it against the first fresh price when the market came back. It missed 8 times. Anyone can rebuild every row from the chain.

A tall stack of sealed cards with one pulled out of line.

The live record

The band as the product runs it, fed by an around-the-clock source that answers every ten seconds. It is one weekend old: 3 of 3 reopenings inside.

Live record of weekend reopenings
ReopeningTokenOff byBandResult
21 Sep 2026NVDA+0.14%±0.55%Inside
21 Sep 2026TSLA−0.05%±0.55%Inside
21 Sep 2026AAPL+0.14%±0.55%Inside

This first weekend was graded after the fact, from our own minute-by-minute capture: the band at 23:59 UTC against the price published at 00:00. Later entries are sealed and timestamped before the market reopens, and marked as sealed.

The stress test

The same band, fed only a thirty-minute average of on-chain trades. At a weekend that price drifts away from the real one, because few people are trading. These are the 8 times it left the band.

Reopenings where the band was wrong
ReopeningTokenOff byBandResult
03 Aug 2026NVDA+0.62%±0.55%Wrong
24 Aug 2026NVDA−0.75%±0.66%Wrong
31 Aug 2026NVDA−0.64%±0.55%Wrong
24 Aug 2026TSLA+0.57%±0.55%Wrong
31 Aug 2026TSLA−0.64%±0.55%Wrong
31 Aug 2026AAPL−0.59%±0.55%Wrong
24 Aug 2026SPY−0.59%±0.58%Wrong
31 Aug 2026SPY−1.24%±0.55%Wrong
Show all 30 reopenings
Every reopening in the stress test
ReopeningTokenOff byBandResult
27 Jul 2026NVDA−0.24%±0.55%Inside
03 Aug 2026NVDA+0.62%±0.55%Wrong
10 Aug 2026NVDA−0.20%±0.67%Inside
17 Aug 2026NVDA−0.23%±1.48%Inside
24 Aug 2026NVDA−0.75%±0.66%Wrong
31 Aug 2026NVDA−0.64%±0.55%Wrong
08 Sep 2026NVDA−0.26%±0.55%Inside
14 Sep 2026NVDA+0.34%±0.55%Inside
21 Sep 2026NVDA+0.19%±0.55%Inside
27 Jul 2026TSLA+0.25%±0.55%Inside
03 Aug 2026TSLA−0.42%±0.55%Inside
17 Aug 2026TSLA+0.12%±0.55%Inside
24 Aug 2026TSLA+0.57%±0.55%Wrong
31 Aug 2026TSLA−0.64%±0.55%Wrong
08 Sep 2026TSLA−0.33%±0.55%Inside
14 Sep 2026TSLA−0.48%±0.55%Inside
21 Sep 2026TSLA+0.22%±0.55%Inside
03 Aug 2026AAPL−0.38%±0.91%Inside
10 Aug 2026AAPL+0.03%±1.19%Inside
17 Aug 2026AAPL−0.09%±0.72%Inside
24 Aug 2026AAPL−0.03%±0.77%Inside
31 Aug 2026AAPL−0.59%±0.55%Wrong
08 Sep 2026AAPL−0.07%±0.55%Inside
14 Sep 2026AAPL+0.25%±0.55%Inside
21 Sep 2026AAPL+0.04%±0.55%Inside
24 Aug 2026SPY−0.59%±0.58%Wrong
31 Aug 2026SPY−1.24%±0.55%Wrong
08 Sep 2026SPY−0.48%±0.55%Inside
14 Sep 2026SPY−0.02%±0.55%Inside
21 Sep 2026SPY+0.03%±0.55%Inside

For scale: doing nothing, and carrying Friday’s price through the weekend, was off by 0.34% at the median and as much as 1.33%, over 48 reopenings. How to rebuild these tables.

The bell rings on Monday. You didn’t wait for it.

Open an account on testnet, deposit the whole portfolio, and read your liquidation price before you sign anything.

Open the appTestnet