A subscription letter for stocks, crypto, and physical metals — gold and silver. Every call shows the full chain of evidence that produced it. No story without the numbers. No call without a grade.
Issue · Thu Nov 12, 2026 · Built from the research engines' last completed runs (Wed Nov 11, 2026) · Next issue after tonight's runs
How a call is made — the wire map
Tap to expand▸
This is the exact pipeline the three research engines already run every day. A subscriber can audit any call against these steps.
Step 1 · Sources
Public data only
Stocks: SEC EDGAR — 13F filings from five tracked managers (Berkshire, Duquesne, Pershing, Appaloosa, Scion), insider Form 4 filings, ARK daily holdings. Crypto: CoinGecko prices cross-checked against Coinbase spot. Metals: live spot prices, CFTC Commitments of Traders, World Gold Council, ETF holdings, real dealer premiums on physical coins.
▼
Step 2 · Intake
Signals and claims enter separately
Engine signals come from the data itself. Outside calls — a guru post, a viral video, a tip — enter as CLAIMS, not signals. A claim has no weight until it survives the next two steps.
▼
Step 3 · Premise check
Fact-check the claim against primary sources
If the story behind a call isn't true — the filing doesn't say that, the transfer didn't happen, the number is invented — the call dies here, and the letter says why.
▼
Step 4 · Rules screen
The trend decides, not the narrative
Every asset is scored on fixed rules: price vs its 50-day and 200-day averages, 30-day momentum, an RSI penalty for overheated entries (metals use the 20-day vs 50-day trend; crypto drops the earnings leg stocks use). A BUY needs a score of +2 or better. Oversold alone is never a buy signal.
▼
Step 5 · Smart-money cross-check
Is real money aligned — or against?
13F position changes, insider buying clusters, futures positioning (COT), ETF flows, physical premiums. This step also catches traps: a famous "new position" that turns out to be put options — a bet against the stock — gets reported as exactly that.
▼
Step 6 · The decision gate
BUY · HOLD · AVOID — or no ticket at all
Most days, most assets fail the gate. "No BUY qualified today" is a published result, not a failure — it's the filter subscribers are paying for.
▼
Step 7 · Ticket terms
A BUY is issued as an exact ticket
Entry limit, window, position sizes at $500 and $1,000, a target, a stop, and a time exit — fixed before the fact, never edited after. Full anatomy below.
▼
Step 8 · Publish & grade
Every call is logged and graded in public
Calls are scored against their benchmark — stocks vs the S&P 500 (SPY), crypto vs Bitcoin, metals vs spot. Misses stay on the record. A signal that fails grading gets dropped, on the record: our Kronos model scored 47.5% on direction and was removed.
Anatomy of a BUY ticket
Tap to expand▸
When the gate passes an asset, this is the exact shape of the call a subscriber receives.
Entry limit
Reference price × 1.01 — never chase above it
Window
Stocks: next session, expires after 3 trading days. Crypto: opens immediately, 72 hours. Expires instantly if price runs more than 5% above the limit
Size
Shown for $500 and $1,000 (whole shares for stocks; ounces plus premium for physical metal)
Target
Limit × 1.10
Stop
Limit × 0.93
Time exit
Out after 20 trading days (stocks/metals) or 20 calendar days (crypto), whichever exit comes first
Provenance
The signal's own reference price and how far price has already moved since the signal
This issue's headline: no BUY tickets. All three lanes ran Wed Nov 11 and the gate passed nothing — stocks quiet, crypto in a pullback, metals without an uptrend. Below is what the engines actually said, asset by asset. Tap any card to see the full reasoning.
This issue — the board
Data as of the Wed Nov 11, 2026 engine runs. Prices are each lane's reference prices from those runs.
In the live letter, each card opens its own page. Here they're stacked below, exactly as a subscriber would read them.
Bitcoin — HOLD
Crypto lane · Screen run Wed Nov 11, 2026 · Benchmark: itself (crypto grades against BTC)
Source
CoinGecko daily price, cross-checked against Coinbase spot at run time. Reference price $73,826.54.
Rules screen
Price is above its 200-day average (+1) but below its 50-day average (−1). 30-day return −11.8% (−1). RSI 31 — not oversold enough to matter, not overheated. Total score: −1.
The gate
A BUY needs +2. At −1, BTC is in no-man's land: the long-term trend hasn't broken, the short-term trend hasn't healed.
Verdict
HOLD. No ticket issued. Buying here means fighting the short-term trend; selling here means fighting the long-term one.
What changes this call: BTC reclaims its 50-day average with 30-day momentum back above zero — that's a +2 score and a BUY ticket with full terms.
Ethereum — AVOID
Crypto lane · Screen run Wed Nov 11, 2026 · Reference price $2,050.05
Below the 200-day average (−1), below the 50-day average (−1), momentum negative (−1). Score: −3 — the AVOID line is −2.
The trap this avoids
RSI is 24 — deep oversold. Every cycle, "it's cheap now" calls appear at exactly this point. This system's rule, written before the fact: oversold alone is not a buy signal. Falling assets can stay oversold all the way down.
Verdict
AVOID. No buy at any price while both averages sit above the price in a downtrend.
What changes this call: a base — price back above the 50-day average first, then the 200-day, with momentum positive. Until then, every bounce is treated as a bounce, not a reversal.
Solana — AVOID
Crypto lane · Screen run Wed Nov 11, 2026 · Reference price $82.93
Below both averages, momentum deeply negative. Score: −3.
Verdict
AVOID — the same read as ETH. The whole crypto board this run: 4 HOLDs, 6 AVOIDs, 0 BUYs. When the benchmark itself (BTC) can't pass the gate, altcoins don't get the benefit of the doubt.
What changes this call: SOL outperforming BTC back through its 50-day average — relative strength first, ticket second.
Gold — AVOID (for now)
Metals lane · Run Wed Nov 11, 2026 · Spot $3,981.80/oz · GLD $365.17 · Graded against spot
Source
Live spot verified twice in the run; futures positioning from CFTC; demand data from the World Gold Council.
Rules screen
The trend test fails: the 20-day average is not above the 50-day trend. Spot has slid from ~$4,140 (Nov 9) to $3,981.80 in two sessions.
Smart-money cross-check
Two stories, honestly separated. Structural demand is real: central banks bought an estimated 321 tonnes in Q3 2026 (+42% year over year). Positioning is cooling: managed-money futures length is +103,614 contracts but down 7,373 over four weeks — the fast money is trimming, not chasing. GLD holds 1,042 tonnes, roughly flat on the month.
Verdict
AVOID. The long-term story and the entry timing are different questions. The story says gold has real buyers; the trend says this isn't the week to join them at any price.
What changes this call: the 20-day average back above the 50-day trend — then a ticket prints with entry, target, and stop, in spot and physical (ounces + dealer premium) terms.
Silver — AVOID · the watch metal
Metals lane · Run Wed Nov 11, 2026 · Spot $51.46/oz · SLV $46.65 · Gold/silver ratio 77.4
Source
Live spot, CFTC positioning, dealer premium checks on physical Silver Eagles.
Rules screen
Same failed trend test as gold — no uptrend, no ticket. Spot down from $53.52 (Nov 9) to $51.46.
Why it's the watch
Managed-money net length is only +4,035 contracts — near 2.5-year lows — but it rose 21.5% in one week off that small base. Crowded trades unwind; empty ones have room to run. At a 77.4 gold/silver ratio, silver is also the historically cheaper metal of the pair.
Verdict
AVOID today, first in line tomorrow. Positioning is turning before price is. The system waits for price to confirm — that's the discipline — but subscribers see the turn being tracked, not discovered late.
What changes this call: the same trend flip as gold. Silver usually moves first and faster — in both directions, which is exactly why the gate exists.
Alphabet — smart-money signal, honestly labelled
Stock lane · 13F tracker, run Wed Nov 11, 2026 · Source: SEC EDGAR filing, Berkshire Hathaway 13F-HR filed Aug 14, 2026 (quarter ended Jun 30, 2026)
The signal
Berkshire raised its Alphabet position +45.2% in shares, to 78.79M shares worth $28.16B — its largest increase this filing. A second Alphabet share class position rose +658.3% to $9.61B.
The delay, printed on the label
13F data is a quarter-end snapshot filed up to 45 days later. This position is months old news by the time anyone can read it — the letter never presents it as a live trade.
Insider cross-check
The Form 4 lane in the same run: 30 fresh insider filings parsed — 0 open-market purchases, 27 sales, and no 3+ insider buying cluster anywhere in the window. Insiders, in aggregate, are not confirming a broad buy signal.
Verdict
Context, not a ticket. Smart-money moves shape the watchlist; they never override the rules screen. Alphabet gets watched harder because Buffett added — it gets bought only if the trend gate passes it.
Why subscribers get this: most letters sell "Buffett bought X" as a buy signal. The receipts version shows the filing date, the lag, and the contradicting insider lane — then lets the rules decide.
Palantir — the trap, flagged
Stock lane · 13F tracker, run Wed Nov 11, 2026 · Source: SEC EDGAR, Scion Asset Management 13F
The headline version
"Michael Burry takes a $912.1M position in Palantir." True — and useless without the next line of the filing.
What the filing actually says
The position is in put options — contracts that profit when the stock falls. Burry isn't buying Palantir; he's betting against it. The engine reads the option flags in the filing table, so this is caught mechanically, not by opinion.
Verdict
Reported as a bearish signal — the opposite of the headline. Anyone who copied the headline version bought the wrong side of one of the most-watched traders alive.
This one example is the product: the data is public and free. The money is in reading it correctly — and showing the work.
Get the next issue
One letter per engine cycle: every call, every ticket, every reasoning chain — including the days the answer is "buy nothing." Track record published, misses included.