The Alchemist's Ledger — August 2026
August 2026 — Issue No. 1
A monthly Distillation.
Read this first. The Alchemist's Ledger is published for entertainment purposes only. We do not provide investment advice, and nothing in this letter is a recommendation to buy or sell anything. We are dealers. Alchemy Trading Co. trades in every category covered here and holds positions in most of them. We publish our thinking after we've acted on it, not before. Do your own work.
Where We Disagree
Most of what you're reading about hard assets right now is wrong, or at least early in a way that's going to cost people money.
The story goes: silver's run is a physical shortage finally breaking through, so it's going higher. Crypto is finished, down roughly half on Bitcoin and more on Ethereum, capital gone to AI, thesis disproven. Natural diamonds are a dead category that lab-grown killed for good.
We don't buy any of it.
Silver's move is a paper squeeze and it's going to fail. Crypto is setting up the best accumulation window it's offered in two years. And natural diamonds are finding a floor right about now, badly over-extended to the downside.
Here's our thinking on each.
Metals
Gold is near $4,400/oz, up 28% on the year, after touching roughly $5,000 in March and giving back most of that into summer. Silver is around $65/oz. That's up 12% on the month and 70% on the year, after living above $70 in the spring and washing out under $60 in June. The Silver Institute has 2026 at a record 215 million ounce deficit, the sixth straight year of shortfall. Everybody's pointing at that number.
The number is real. What people are doing with it isn't.
Physical demand has been steady. Not weak, but steady, and steady demand doesn't produce a 70% year. What produced it was positioning. Paper traders got squeezed out of short books into a thin tape with a Fed hike debate and a Hormuz conflict feeding it headlines. Squeezes make real prices, but they're a positioning event, not a demand event, and positioning events give it back.
So our call is that silver pushes through $80 and doesn't hold. It might sit at $80 to $90 for a month or two, long enough for everyone to call the breakout confirmed. Then it gives most of it back. We want a rejection candle on the 3-day chart before we'd say it's done.
None of that means silver lacks legs. $300 is very possible. But this market has to digest these prices first, and that's going to take a lot longer than the bulls think. We're talking about years of range, not months of pause. Own metals for where they end up. Size the position so a two-year chop doesn't bother you, and stop watching the chart every morning.
The Rotation Nobody Accounts For
This is the piece that ties the letter together, and it's the reason our metals call and our crypto call aren't in conflict.
Metals and crypto both go up in an inflationary cycle. Over a long enough window they rise together, and that's what most people see. But inside the cycle they take turns. Crypto runs while metals consolidate. Then metals wake up and crypto goes quiet. Pull up the charts side by side over the last several inflation cycles and you'll see it without needing anyone to explain it to you.
We just spent a year with metals doing the work. Gold up 28%, silver up 70%. Crypto sat it out and got left for dead.
If we're right that silver is about to spike, fail, and then spend a long time digesting, then metals are going into their quiet leg. That's crypto's turn. We're not making a crypto call and a metals call separately here. It's one call with two sides.
Crypto
Bitcoin is near $63,000, down about 49% year over year. Ethereum is near $1,875, down 60%. ETF outflows, money rotated into AI, legislation stalled out. And the detail worth noticing: crypto didn't move on the soft inflation print this month while metals did.
Hold onto that “money rotated into AI” line. We'll come back to it, because we think it's the most misread fact in this market.
Most people read that as the thesis dying. We read it as the thesis running late, and late is what makes something cheap. The reason crypto ignored the inflation data is that the buyer at the margin right now is still a risk-asset buyer, not somebody worried about the dollar. That composition changes when the Fed gets forced to reverse, and a Fed hiking into 3.4% CPI with mediocre growth is a Fed setting up that reversal. You don't get to wait for the composition to change and still buy it cheap.
So we'd start a DCA. Not a lump sum and not a bottom call. Just steady accumulation into a market nobody wants, which is the only condition under which this asset has ever been cheap. Large caps are the safe end of it. The bigger upside sits in the ISO 20022 names, XRP ($1.00), HBAR ($0.066) and XLM ($0.158), which are built to speak the messaging standard the banking system is actually moving to. That's a specific thesis you can be proven wrong about, which puts it ahead of most of what gets pitched in this sector. And note what those quotes tell you. XLM is down about 64% on the year, HBAR about 73%. Nobody is excited about these right now, which is the entire point.
Now back to that AI rotation. Everyone treats AI as the thing that drained capital out of crypto. We think AI is going to be the thing that gives crypto a job.
Agentic AI needs money. Not as a metaphor. An agent that books travel, buys compute, pays for an API call, or hires another agent has to move value, and it has to do it without a person sitting there clicking approve. Every payment rail we have assumes a human is present. Cards, ACH, wires, all of it is built around an account holder, a manual authorization, and batch settlement windows. That design falls apart when software needs to transact constantly, in tiny amounts, across borders, with nobody watching.
Crypto handles it. Stablecoins settle continuously instead of in batches, move without FX friction, do micropayments at costs a card network can't approach, and don't require the agent to be a legal person with a bank account. Nothing sits in the middle taking a cut and adding latency. Speed is what AI needs, and removing the intermediary is how you get it.
This has already left the whiteboard. x402, which resurrects an unused HTTP status code so a service can bill an agent per request, went to the Linux Foundation in April 2026. Stripe and Tempo put out MPP for recurring and multi-agent settlement. AWS shipped agent microtransactions in Bedrock. Mastercard is paying up to $1.8 billion for BVNK to buy its way onto stablecoin rails. Visa built a command-line payment tool for Claude Code.
So the capital that left crypto to chase AI is going to find that AI needs crypto to function. That's not irony, it's a demand thesis, and we don't see it priced in anywhere.
The GENIUS Act is the other piece people are sleeping on. Signed July 2025, and the rules are being written now. The OCC put out its proposed regulation in Q1 2026, and the statute goes live the earlier of January 2027 or 120 days after final rules. Every compliant stablecoin has to hold 100% reserves in cash, deposits, or short-dated T-bills.
Follow that through. A dollar of stablecoin issued is a dollar of customer cash turned into T-bills bought. Tether was already sitting on roughly $127 billion in Treasury exposure and Circle around $60 billion. Put the stablecoin complex together and it would rank about the 18th largest holder of US Treasuries on earth, somewhere between Germany and Norway.
That is a monetary stimulation channel and it's being built in public. Washington has figured out that stablecoins create a structural, automatic buyer at the front of the Treasury curve. Every dollar that moves into a regulated stablecoin funds the government at short duration. Nobody has to vote on it and nobody has to call it QE.
The other half of it is control. A federally licensed, fully reserved, monthly-audited stablecoin is a programmable dollar. It can be frozen, traced, and eventually conditioned on behavior. Treasury already has proposed rules out on the illicit finance side. The infrastructure being stood up to support the Treasury market is the same infrastructure that gives an issuer a switch on your balance.
We think both things are true at once. Stablecoins are going to be a real source of monetary support, and they're going to be a real instrument of monetary control over individuals. If that's the direction, then owning the neutral, self-custodied version of the asset matters more, not less. Hold your own keys.
Going into the fall, crypto looks like better risk-adjusted value than metals to us.
Gemstones
The Big Three keep grinding and everybody already knows it. That part is priced.
Before any numbers, a rule we hold to. A per-carat price means nothing without the spec attached. Size, origin, color, clarity, and treatment are what determine value, not the variety name. Two stones both honestly called “ruby,” both one carat, can trade a hundred times apart. So when we quote, we quote the whole spec. Anyone handing you a bare per-carat number for a colored stone is either careless or selling you something.
Reference points as of this writing, fine commercial-to-investment material, eye-clean:
| Stone | Spec | Per carat |
|---|---|---|
| Ruby | Burma, unheated, fine color, 1–3 ct | $15,000–$80,000 |
| Ruby | Burma, unheated, exceptional “pigeon's blood,” 3 ct+ | $80,000–$500,000+ |
| Ruby | Mozambique, heated, fine color, 1–3 ct | $3,000–$8,000 |
| Ruby | Mozambique, commercial, heated, 1–3 ct | $500–$3,000 |
| Red spinel | Burma, unheated, fine saturated red, 1–3 ct | $5,000–$15,000 |
| Red spinel | Burma, unheated, commercial, 1–3 ct | $500–$1,500 |
| Tsavorite | Untreated (market standard), vivid green, 2–3 ct | $2,000–$4,000 |
| Tsavorite | Untreated, vivid green, 5 ct+ | $5,000–$8,000+ |
| Blue zircon | Heat-treated (market standard), fine saturated blue, 5–10 ct | $200–$300 |
| Blue zircon | Heat-treated, fine saturated blue, 10 ct+ | $300–$400 |
Now look at rows one and five. Burma origin, unheated, fine color, same 1 to 3 carat range. Ruby at $15,000 and up. Red spinel at $5,000 to $15,000. Same country, same treatment status, same size band, comparable beauty in hand, and a three-to-five times spread on name alone. That is the substitution trade, and stating it with the spec attached is the only way the comparison is worth anything.
Red spinel, blue zircon, and malaya garnet are the early runners, and malaya in particular has a lot of upside left in it. Mahenge spinel and tsavorite belong in the same conversation.
On malaya we'll be straight with you: we could not find a per-carat figure published to a standard we'd repeat here. Four sources, no usable spec. That is not a knock on the stone, it's a symptom of how thinly it's covered, and thin coverage is usually where the early money is.
Which raises the broader point. There is no real public price index for colored stones. Diamonds have RAPI, metals have spot, colored stones have auction records and opinions. That absence isn't a footnote, it's the opportunity. An unindexed market is an inefficient market, and inefficiency is where mispricing survives long enough for you to actually buy it.
The logic underneath all of this is about people, not geology. Consumers will always chase material that looks as good or better for less money. That has never been wrong yet.
Buy size and quality over quantity. One good stone beats ten commercial ones, because when a category re-rates the money goes to the top of the quality curve and skips the middle. Note in the table how the per-carat number climbs with size within the same spec. That is the whole argument for buying one better stone.
On treatment, and this applies to all colored stones: treated is fine. Most of the market is treated and there's nothing wrong with owning it. Untreated is a bit better for price stability, and that's the real argument for it, not that treated stones won't appreciate. The premium is real and it's large. Unheated Burma ruby runs 3 to 10 times its heated equivalent, and a GRS “pigeon's blood” call adds another 30 to 50%. Worth knowing that some stones are untreated as a market rule rather than as a premium: garnet and spinel are essentially never treated, while blue zircon is essentially always heated. Pay up for untreated where the category actually has a treated alternative, and don't refuse to own treated.
On diamonds we're going against the room, and the index just started agreeing with us.
Natural is finding a floor. Look at RAPI. In July all four major categories printed flat or positive, the first time that has happened since March 2025. It broke a 13-month losing streak for 1-carat goods. Half-carat stones had their best month since March 2025 at +1.8%. Thirty-pointers ran +4.2% in June and another +1.6% in July.
The year-over-year spread says the same thing from a different angle. Thirty-pointers are still down roughly 28% on the year while 3-carat stones are down about 2%. The damage is concentrated in small goods and the top of the market has essentially stopped falling. That is what a bottom looks like from the inside. It forms at the quality end first and works down.
Knight Frank's luxury index tells a similar story. It closed 2025 down 0.4%, a year of stabilization after two straight years of losses.
Then there's De Beers taking a $6.8 billion writedown. A writedown that size is what capitulation looks like on the way out, not pain still coming. Lab-grown held natural prices down for years, but the labs are far past saturated now. Synthetics can't keep falling without going under their own cost to produce, and once lab stops falling it stops dragging natural with it.
We've watched this before. It's cubic zirconia all over again. A synthetic floods in, everybody declares the natural stone finished, and then the market remembers that natural material holds value and the synthetic doesn't. Natural is over-extended to the downside here. That's a minority opinion and we're fine holding it.
Rare Coins
Rarity is doing what it should. Stack's Bowers sold an 1841 Seated Liberty dime for $341,600. Heritage placed a Continental Dollar at $1,312,500. Sourcing on key dates and better-grade classic US is tight right now, and tight sourcing is a demand signal before it's a complaint.
There's been a lot of price movement lately and we want to be clear that this isn't a bubble. Things are re-rating. Demand is strong, the floor isn't going to drop out, and the structure hasn't changed. Supply is finite, nobody is minting more of it, and an aging collector base keeps feeding estates into the market. This is a slow accumulation play and it should be treated like one.
It matters more if you believe our metals view. If spot goes nowhere for years, a coin priced on metal content is dead money while a coin priced on rarity keeps working. Worst case, coins drift sideways with metals. Best case, capital comes out of stagnant metal and moves into tangible collectibles.
And that capital is unlikely to go back to dollars and stay there. This is the part we'd underline. Money that already made the decision to leave fiat doesn't usually return to cash and sit in it, not with the Fed hiking into 3.4% inflation and still losing ground. It rotates inside the hard-asset world looking for whatever is working. So a long metals consolidation isn't a headwind for rare coins. It's what funds them. The premium over melt stops being a markup and becomes the whole asset.
Bank Notes
The top of the paper market is setting records. Stack's Bowers' spring US currency sale did nearly $3.6 million, with a 1928 $500 Gold Certificate at $204,000 and a 1934 $100 light green seal, serial #1, at $180,000. Notes are underfollowed and cheap next to comparable rarity in coins.
That said, there isn't a lot of upside here for the average collector, and we'd rather say so than pad the letter. The records are being set at the census level by a small group of people bidding on a small number of notes. Below that the market is thin and appreciation is unremarkable. If you love the material it's a great and inexpensive place to collect. If you're putting capital to work for return, there are better places on this page.
The Washington Tax
Washington repealed its sales tax exemption on precious metal and monetized bullion effective January 1, 2026. Bullion now carries retail sales tax plus B&O under the Retailing classification. Wholesale needs a reseller permit.
Don't file this as a regional footnote. What it actually did was kill in-state retail bullion and push buyers out of state, and move a lot of volume private and peer-to-peer. Demand didn't disappear. It relocated and went quiet.
That's the lesson worth taking. When a state makes the taxed channel uneconomic, the trade doesn't stop, it moves somewhere it isn't watched. States copy each other, especially on revenue. Washington is a template.
If you hold hard assets for control and privacy rather than just for return, this is the most important thing in this letter and it has nothing to do with price. The risk to your position over the next few years isn't that gold goes down. It's changes in how, where, and at what friction you're allowed to transact. Value liquidity that doesn't depend on one jurisdiction's mood.
How We'd Position From Here
Split it. Metals for the endgame, collectibles for the interim, and crypto for the leg we're about to enter.
Metals are a multi-year hold with a big number at the end and a long, boring middle that most people won't sit through. Own them, size them right, and stop watching. What actually pays you during the digestion is rarity and stones. Coins re-rating on their own merits, second-tier colored gems repricing toward the material they resemble, and a diamond market being left for dead about a year too early.
That split isn't just an allocation preference, it's a guess about where money goes. If silver spikes and fails into a long consolidation, a lot of capital comes off that trade, and it isn't going into dollars to wait. It goes looking for the next tangible thing. We'd rather already own that thing.
And crypto deserves a real position, not a rounding error. A lot of hard-asset people carry a token allocation they don't take seriously, or none at all, usually because the last cycle burned them or because it doesn't feel like a real asset. That's a mistake going into this particular fall.
Three things stack up at once here. The rotation says it's crypto's turn while metals digest. Agentic AI is quietly building a reason for these rails to exist that has nothing to do with speculation, and the infrastructure is being funded by Mastercard, Visa, Stripe and AWS while the price sits 49% off. And the GENIUS Act means the government is about to route real front-end Treasury funding through stablecoins, using the same rail that makes monetary control over individuals possible. If you believe that last part, self-custodied crypto isn't a speculation sitting on top of your hard-asset book. It's the same insurance you already bought gold for, against a different failure.
Size it so a 50% drawdown doesn't change your life, and so a 5x doesn't leave you wishing you'd bothered.
Spot and index figures as of publication, August 16, 2026. We put them in writing so these calls can be scored later. Colored stone figures are reference points for the stated spec only. Size, origin, color, clarity, and treatment determine value; change any one of them and the number changes with it.
The Alchemist's Ledger is published monthly by Alchemy Trading Co. for entertainment purposes only. We do not provide investment advice and nothing here is a recommendation to buy or sell any asset. We are dealers, we hold what we cover, and we say so at the top of every issue. Verify independently, buy certified, and take delivery.