On-chain data

Borrowing Against QQQx on Kamino at Max LTV Leaves a 2.8% Cushion, and QQQ Fell 3.04% in 24 Hours on 14 October

Kamino lends stablecoins against tokenized stocks on Solana. At the main market's max LTV, a SPYx or QQQx loan is liquidated by a fall of under 3%, and QQQ fell more than that inside one day last week. The script to check any token and any loan is included.

By

In India you can take a loan against your shares from a bank. If the shares fall, the bank asks you to add money. If you don’t, it sells your shares to get its loan back.

Kamino does the same thing on Solana with tokenized stocks, and no one calls you first. You deposit a token like QQQx, you borrow USDC against it, and a program watches the price. When the loan gets too big for the stock under it, anyone can repay part of your debt and take your stock at a discount.

So the question for you is simple. How far can the stock fall before that happens? We read the risk settings of every tokenized stock on both of Kamino’s xStocks markets, and we checked them against what the shares actually did last week.

Key findings

How a Kamino loan gets liquidated

Two numbers decide it. Max LTV is the most you can borrow, as a share of what your collateral is worth. Liquidation LTV is the line where the loan can be closed. LTV means loan to value, so a 70% LTV means you borrowed 70 dollars for every 100 dollars of stock.

Your LTV goes up when the stock goes down, because the loan stays the same and the stock is worth less. The cushion is how far the stock can fall from where you borrowed before your LTV hits the liquidation line.

Formula
Cushion 1 minus (your LTV divided by liquidation LTV)
Liquidation price amount borrowed divided by (tokens held times liquidation LTV)

When the line is crossed, a liquidator can repay up to one fifth of your debt in one go. In return they take your stock worth that amount plus a bonus. The bonus is the penalty you pay, the same mechanism as on Aave. On Kamino the price comes from Chainlink Data Streams, and for most tokens Kamino rejects a price older than five minutes or more than 5% away from its TWAP.

The cushion on each token

These are the settings Kamino’s API returned on 20 October 2026. The cushion column assumes you borrowed at max LTV.

The main xStocks Market. A liquidation here costs you a penalty of 5% to 10%.

Token Max LTV Liquidation LTV Cushion
SPYx 73% 75% 2.7%
QQQx 70% 72% 2.8%
GOOGLx 60% 70% 14.3%
TSLAx 55% 65% 15.4%
NVDAx 55% 65% 15.4%
AAPLx 40% 50% 20.0%
HOODx 30% 40% 25.0%
CRCLx 30% 40% 25.0%
MSTRx 30% 40% 25.0%

The Sentora xStocks market. A liquidation here costs a flat 10%.

Token Max LTV Liquidation LTV Cushion
SPYx 72% 79% 8.9%
QQQx 70% 78% 10.3%
NVDAx 62% 73% 15.1%

Look at the top two rows. SPYx and QQQx get the highest max LTV on the main market, and that is why they have the thinnest cushion. The max LTV sits right under the liquidation line. The single stocks lower down get less to borrow and much more room.

What the shares did last week

A cushion means nothing without a fall to compare it to. This table takes the worst fall inside any 24 hours, in regular trading only, from 12 to 19 October 2026. The last column says how much of the max LTV cushion on the main market that fall used up.

Share Worst 24-hour fall Main market cushion Cushion used
QQQ 3.04% 2.8% 109%
SPY 1.50% 2.7% 56%
NVDA 5.43% 15.4% 35%
TSLA 5.16% 15.4% 34%
GOOGL 4.20% 14.3% 29%
AAPL 1.91% 20.0% 10%
MSTR 11.12% 25.0% 44%

QQQ is the only one over the line. Say you deposited QQQx and borrowed at the 70% max when QQQ was at 738.28. Your liquidation price was 717.77, and the next morning QQQ traded at 715.84.

No, this was not a crash week. SPY’s worst 24 hours used more than half of SPYx’s cushion, and that was a 1.50% fall.

We record our own copy of these prices. chaotic.markets saves the share price Jupiter reports for every token every 5 minutes. For QQQx it logged 738.19 at 10:15 New York time on 14 October and 716.20 at the same time on 15 October, a 2.98% fall, past the 2.8% line too. The live gap between QQQx and its share is on the QQQx data page.

Borrow less, get more room

The cushion grows fast as you borrow less. Same token, same market, same liquidation line, only your LTV changes.

Your LTV SPYx cushion (liquidation 75%) QQQx cushion (liquidation 72%)
73% 2.7%
70% 6.7% 2.8%
60% 20.0% 16.7%
50% 33.3% 30.6%
40% 46.7% 44.4%

The max LTV is the most Kamino lets you borrow. On SPYx and QQQx a loan at that max can be closed on an ordinary day. The script below prints the liquidation price for your own loan size.

Two markets, two prices

Kamino runs two lending markets that take xStocks. The main xStocks Market lends USDC against nine stock tokens. The Sentora xStocks market is a curated market that lends PYUSD against SPYx, QQQx and NVDAx only, and Kamino’s API marks it hidden from its main list.

On 20 October 2026 xStocks Market Sentora xStocks
You borrow USDC PYUSD
Borrow rate 6.9% 0.9%
Share of the pool already lent 92% 21%
SPYx cushion at max LTV 2.7% 8.9%
Penalty if liquidated 5% to 10% 10%

Rates move with how much of the pool is lent, so check them before you borrow. Two more limits from the same API. GOOGLx deposits on the main market were nearly full, and METAx was closed to new deposits.

What this does not tell you

Six trading days is a short history. A calm week can make any cushion look safe, and one bad day can break it.

We used regular trading hours only. Kamino’s oracle is Chainlink, and its prices will not match Yahoo’s 5-minute bars to the cent. Kamino says that outside market hours it only accepts prices inside a band around the last close, but it has not published how wide that band is. What a weekend does to your loan, as covered in what the weekend price says about Monday’s open, depends on that band.

Interest also adds to your debt every day, so your LTV creeps up even when the stock does not move. And Kamino can change these settings, so run the script before you trust any number here.

We make no claim about why any share moved.

Run it yourself

You need Python 3.9 or newer. There is nothing to install and no key to set. On Windows, Python’s zoneinfo needs one package first, pip install tzdata. Save this as kamino_cushion.py.

"""How far can a tokenized stock fall before Kamino liquidates a loan against it?

usage: python3 kamino_cushion.py                              every xStock on both Kamino markets
       python3 kamino_cushion.py loan SPYx 10 4000            10 SPYx held, 4,000 dollars borrowed
       python3 kamino_cushion.py drop QQQ 2026-10-12 2026-10-19   worst 24-hour fall of the share
"""

from __future__ import annotations

import json
import sys
import urllib.request
from collections import deque
from datetime import date, datetime, time, timedelta
from time import sleep
from zoneinfo import ZoneInfo

KAMINO = "https://api.kamino.finance"
BARS_URL = "https://query1.finance.yahoo.com/v8/finance/chart/{ticker}?interval=5m&period1={start}&period2={end}"
NEW_YORK = ZoneInfo("America/New_York")
MARKETS = {
    "xStocks Market": "5wJeMrUYECGq41fxRESKALVcHnNX26TAWy4W98yULsua",
    "Sentora xStocks": "8BNUWRSibVasaAmhYpBCFpGgMisGKfVAf9ho3Cmf6vjr",
}
STABLES = {"USDC", "USDG", "PYUSD"}


def _get(url: str) -> object:
    """Fetch a URL and parse the JSON body. Tries twice, then raises."""
    request = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0 kamino-cushion/1.0"})
    for attempt in range(2):
        try:
            with urllib.request.urlopen(request, timeout=20) as response:
                return json.load(response)
        except Exception:
            if attempt:
                raise
            sleep(2)
    return None


def _collateral(market: str) -> list[dict]:
    """Every stock reserve in a market that takes deposits, with its risk settings and price."""
    out = []
    for reserve in _get(f"{KAMINO}/kamino-market/{market}/reserves/metrics"):  # type: ignore[union-attr]
        if reserve["liquidityToken"] in STABLES:
            continue
        stats = _get(f"{KAMINO}/reserves/{reserve['reserve']}/stats")
        supply = float(stats["totalSupply"])  # type: ignore[index]
        if stats["status"] != "Active" or float(stats["maxLTV"]) == 0 or not supply:  # type: ignore[index]
            continue
        out.append({
            "token": reserve["liquidityToken"],
            "max_ltv": float(stats["maxLTV"]),  # type: ignore[index]
            "liq_ltv": float(stats["liqLTV"]),  # type: ignore[index]
            "penalty": (float(stats["minLiquidationBonus"]), float(stats["maxLiquidationBonus"])),  # type: ignore[index]
            "price": float(stats["totalSupplyUsd"]) / supply,  # type: ignore[index]
        })
    return out


def cushion() -> None:
    """Print, for each collateral, how far it can fall from a loan taken at max LTV."""
    print(f"{'market':16} {'token':7} {'max LTV':>8} {'liq LTV':>8} {'cushion':>8}  penalty")
    for name, market in MARKETS.items():
        for c in sorted(_collateral(market), key=lambda c: -c["max_ltv"]):
            drop = (1 - c["max_ltv"] / c["liq_ltv"]) * 100
            low, high = (p * 100 for p in c["penalty"])
            penalty = f"{low:.0f}%" if low == high else f"{low:.0f}% to {high:.0f}%"
            print(f"{name:16} {c['token']:7} {c['max_ltv']:8.0%} {c['liq_ltv']:8.0%} {drop:7.1f}%  {penalty}")


def loan(token: str, held: float, borrowed: float) -> None:
    """Print the LTV and liquidation price of one loan on each market that takes the token."""
    for name, market in MARKETS.items():
        for c in _collateral(market):
            if c["token"].lower() != token.lower():
                continue
            value = held * c["price"]
            ltv = borrowed / value
            liq_price = borrowed / (held * c["liq_ltv"])
            print(f"{name}: {c['token']} at ${c['price']:,.2f}, collateral ${value:,.2f}, LTV {ltv:.1%}")
            if ltv > c["max_ltv"]:
                print(f"  over the {c['max_ltv']:.0%} max LTV, Kamino would not open this loan")
            print(f"  liquidated at ${liq_price:,.2f}, a fall of {(1 - liq_price / c['price']) * 100:.1f}%\n")


def drop(ticker: str, start: date, end: date) -> None:
    """Print the worst fall of a share inside 24 hours, regular trading only, from 5-minute bars."""
    lo = datetime.combine(start, time(), NEW_YORK)
    hi = datetime.combine(end + timedelta(days=1), time(), NEW_YORK)
    url = BARS_URL.format(ticker=ticker, start=int(lo.timestamp()), end=int(hi.timestamp()))
    result = _get(url)["chart"]["result"][0]  # type: ignore[index]
    closes = result["indicators"]["quote"][0]["close"]
    bars = []
    for stamp, price in zip(result["timestamp"], closes):
        at = datetime.fromtimestamp(stamp, NEW_YORK)
        if price and time(9, 30) <= at.time() < time(16):
            bars.append((at, price))
    worst = (0.0, bars[0], bars[0])
    highs: deque[int] = deque()  # indexes of the highs inside the last 24 hours, highest first
    for i, (at, price) in enumerate(bars):
        while highs and bars[highs[-1]][1] <= price:
            highs.pop()
        highs.append(i)
        while bars[highs[0]][0] < at - timedelta(hours=24):
            highs.popleft()
        fall = price / bars[highs[0]][1] - 1
        if fall < worst[0]:
            worst = (fall, bars[highs[0]], (at, price))
    fall, (t1, p1), (t2, p2) = worst
    print(f"{ticker}: {len(bars)} bars, {start} to {end}, regular trading only")
    print(f"worst 24-hour fall {fall * 100:.2f}%: {p1:.2f} at {t1:%a %d %b %H:%M} to {p2:.2f} at {t2:%a %d %b %H:%M} New York")


if __name__ == "__main__":
    args = sys.argv[1:]
    if not args:
        cushion()
    elif args[0] == "loan" and len(args) == 4:
        loan(args[1], float(args[2]), float(args[3]))
    elif args[0] == "drop" and len(args) == 4:
        drop(args[1], date.fromisoformat(args[2]), date.fromisoformat(args[3]))
    else:
        sys.exit(__doc__)

First, the cushion on every token. Run it with no arguments.

python3 kamino_cushion.py
market           token    max LTV  liq LTV  cushion  penalty
xStocks Market   cbBTC        75%      80%     6.2%  1% to 10%
xStocks Market   SPYx         73%      75%     2.7%  5% to 10%
xStocks Market   QQQx         70%      72%     2.8%  5% to 10%
xStocks Market   GOOGLx       60%      70%    14.3%  5% to 10%
xStocks Market   TSLAx        55%      65%    15.4%  5% to 10%
xStocks Market   NVDAx        55%      65%    15.4%  5% to 10%
xStocks Market   AAPLx        40%      50%    20.0%  5% to 10%
xStocks Market   HOODx        30%      40%    25.0%  5% to 10%
xStocks Market   CRCLx        30%      40%    25.0%  5% to 10%
xStocks Market   MSTRx        30%      40%    25.0%  5% to 10%
Sentora xStocks  SPYx         72%      79%     8.9%  10%
Sentora xStocks  QQQx         70%      78%    10.3%  10%
Sentora xStocks  NVDAx        62%      73%    15.1%  10%

The main market also takes cbBTC, which is why it shows up. Next, your own loan. Pass the token, how many you hold, and how many dollars you borrowed.

python3 kamino_cushion.py loan SPYx 10 4000
xStocks Market: SPYx at $774.00, collateral $7,739.98, LTV 51.7%
  liquidated at $533.33, a fall of 31.1%

Sentora xStocks: SPYx at $774.00, collateral $7,739.98, LTV 51.7%
  liquidated at $506.33, a fall of 34.6%

The price is Kamino’s own, the one it values your collateral at, so it moves every time you run this. Last, the worst 24-hour fall of a share over any dates in the last 60 days, which is as far back as Yahoo keeps 5-minute bars.

python3 kamino_cushion.py drop QQQ 2026-10-12 2026-10-19
QQQ: 468 bars, 2026-10-12 to 2026-10-19, regular trading only
worst 24-hour fall -3.04%: 738.28 at Wed 14 Oct 09:50 to 715.84 at Thu 15 Oct 09:50 New York

Run drop with SPY, NVDA, TSLA, GOOGL, AAPL and MSTR to get the rest of the table above.

Buying the token first

To borrow against QQQx or SPYx you need the token in your own wallet. On chaotic.markets you buy it in one click from that wallet: open the QQQx page or the SPYx page, press Buy, connect your Solana wallet, type the amount of USDC and approve once. The tokens settle to your wallet and chaotic.markets never holds them, so you can deposit them on Kamino yourself. chaotic.markets adds no fee. You still pay the pool fees inside Jupiter’s route, the Solana network fee, and any price over the share, which each data page measures live.

Newsletter

Get chaotic.markets updates and new data articles by email.

You confirm by email, and you can unsubscribe from any message. The list is kept by Kit.