In the second week of March 2020, one of the largest bond funds in the world traded 5% below the value of what it held. It was LQD, a $30 billion investment-grade corporate bond ETF, and in the two years before that day it had never strayed even 1% from its worth.

A bond ETF lets an ordinary investor buy a slice of hundreds of corporate bonds in one ticker that trades all day like a stock. It sits in most retirement and brokerage accounts. Its promise is easy access to a market the reader could never trade directly.

The machine we are drawing is the plumbing that keeps the ETF's price honest. It is a swap system run by a handful of banks. When it works, the ETF tracks its bonds to the penny. When it strains, the price and the bonds come apart.

Bond ETFs have become a default holding. Trillions of dollars now sit in funds that trade all day while holding bonds that may not trade for days. The wrapper is liquid. The contents are not.

That gap is bridged by a small group of firms called authorized participants, or APs. They are the only ones allowed to create new ETF shares or turn shares back into the underlying bonds. Everyone else just trades shares with other traders.

In March 2020, the APs pulled back. Faced with a frozen bond market and their own limits, they stopped closing the gap. The ETF price fell well below the value of its bonds, and stayed there until the Federal Reserve stepped in.

A radical "light-speed" device is turning AI as we know it into "Accelerated AI".

It's smaller than a fingertip. It's made of glass. And it could make AI up to 100 times faster and up to 100 times more energy efficient.

Jensen Huang, Nvidia's CEO, says this device is shattering the limitations of AI and without it, AI can't scale.

Google Ventures says it's the future of AI compute.

And Sequoia Capital calls it a "holy grail".

Yet most Americans have never heard of it…

Wall Street insider Jason Bodner – the same man who called Nvidia at $4.50 – believes this device could launch a whole new wave of AI winners.

And to prove it, he's giving away his #1 stock involved with it – for free.

Click here for all the details.

Start with the two prices. Every ETF has a market price, set by buyers and sellers on the exchange all day, and a net asset value, the worth of the bonds it holds. The whole design exists to keep those two numbers together.

The APs are what keep them together. If the ETF trades above its bonds, an AP buys the bonds, hands them to the issuer, and gets new ETF shares to sell at the higher price. If it trades below, the AP runs it in reverse. Each trade pockets the gap and closes it.

The load-bearing part is that this is voluntary. No AP is required to step in. They do it only when the arbitrage profit beats the cost and risk of trading the underlying bonds. When bonds get hard to trade, that math stops working, and the APs simply stop.

Now the mismatch. The ETF promises instant liquidity to its holders, but its bonds trade rarely and by phone, one dealer at a time. In calm markets the AP smooths this over. In a panic, the AP is asked to buy bonds nobody wants, using its own balance sheet, exactly when its balance sheet is stretched.

And the APs wear two hats. The same banks that arbitrage the ETF are the dealers who make markets in the bonds. When their own capital is squeezed, both jobs shrink at once. The fund's price-keeper and the bond market's buyer are the same stressed institution.

The first joint is the AP's willingness. Because stepping in is optional, the mechanism is strongest when it is needed least and weakest when it is needed most. The discount that opened in March 2020 was not a broken fund. It was APs deciding the trade was not worth it.

The second joint is who eats the gap. The ordinary holder cannot redeem with the fund; only APs can. So a seller in a panic takes whatever the exchange offers. If that is 5% below the stated value of the bonds, that is the price. The loss lands on whoever needs cash now.

The third joint is the conflict of interest. An AP can also use creations and redemptions to offload its own unwanted bond inventory into the fund, rather than to close the price gap. When keeping the peg and managing its own book point in opposite directions, the AP is not required to choose the peg.

What persists: the mechanism in normal markets, which works remarkably well. On an ordinary day, APs keep large, popular bond ETFs within a hair of their NAV, and the ETF actually prints a price for bonds that would otherwise sit stale for days. The wrapper adds real transparency.

What can break: the peg in a cash crunch, precisely when the reader wants out. LQD's 5% discount in March 2020 closed only after the Fed announced it would buy corporate bonds and the fund jumped about 6% in a day. The gap shut because of a rescue, not because the machine self-corrected.

This machine connects to two others the reader's money touches. It rests on the corporate bond market, where the actual borrowing happens, and it leans on the same dealer balance sheets that clear Treasuries and repo. When those balance sheets fill up, the ETF's price-keeper goes missing at the worst time.

The next reading arrives at the next real liquidity squeeze, and until then in the daily premium-and-discount data every issuer publishes. Those numbers will show whether the gap between the liquid wrapper and its illiquid contents is staying shut, or quietly widening.