Masters · not tickets
Best practices
These are practices, not tickets. The men on this page lost money when they ignored their own rules, and they were early when they kept them. A peak is not called. A trough is not called. Confirm two clocks. Do not catch the first knife.
Be fearful when others are greedy · 1956–
Warren Buffett
Price is what you pay. Value is what you get. Cash is a call option on tomorrow’s bargains.
The Buffett indicator on this desk is Yahoo ^W5000 over FRED GDP, a public cousin of the print he showed in Fortune in 2001. FRED WILL5000INDFC 404s. Z.1 nonfinancial equities over GDP is labeled separately and is never spliced onto Wilshire. He does not sell wonderful businesses because a multiple is rich, and he does not buy a falling tape because it is falling. He waits for a fat pitch with a margin of safety. 2008 was cash deployed into Goldman and GE. 2020 was the same temperament. A high ratio is a price, not a sell ticket.
2008 cash · 2020 temperament. Not a timer.
Mr Market is a servant · 1934–1976
Benjamin Graham
Margin of safety. Never confuse a quoted price with an intrinsic value.
Security Analysis and The Intelligent Investor are the method: buy a dollar for fifty cents, hold cash when the tape offers no bargain, and let Mr Market’s mood be a bid you can ignore. Graham’s earnings yield versus the bond is on the masters board. He would not average into a first −5%. He would wait for the quotation to be stupid, not merely red.
1930s net-nets. The quotation is the offer, not the value.
Invert, always invert · 1962–2023
Charlie Munger
Sit on your ass. Wait for the fat pitch. Avoid being stupid.
Inversion: list the ways a crash desk gets killed — catching the first knife, selling a wonderful business on a −15% print, using leverage into a rich CAPE, treating a peak as called. Then do none of those. Opportunity cost is the 10-year on this desk. Activity is not a virtue. A trough is not called because you are bored.
Avoid leverage. Avoid the first bounce. Avoid the story.
Maximum pessimism · 1939–2008
John Templeton
Buy at the point of maximum pessimism. Sell at the point of maximum optimism.
Templeton bought Europe in 1939 when war was the only story. The method is not “buy the dip.” It is wait until the tape is hated, the news is uniformly bad, and the quotation is below any reasonable value. Michigan sentiment inverted sits on the masters board as his cousin. Maximum pessimism is named after it has passed. A peak is not called. A trough is not called.
1939 Europe. 1970s. The hated tape, not the first red week.
Absolute return, cash as option · 1982–
Seth Klarman
Risk is the probability of permanent loss, not the last month’s volatility.
Margin of Safety (1991) is still the crash-desk book. Cash is not a residual. It is the option to buy what other people are forced to sell. High-yield OAS wide is his photograph on this terminal — credit, not a multiple. He would rather underperform a melt-up than participate in a crowded sleeve with no margin. This is not a relative-return ticket.
1998, 2008. Cash was the option. OAS wide was the fuse.
Second-level thinking · 1995–
Howard Marks
You can’t predict. You can prepare. Risk is what you don’t see.
The memos: cycles, second-level thinking, the pendulum from greed to fear. Marks does not call a peak. He asks where we sit in the cycle — credit spreads, covenant quality, the stories people tell. BAA minus AAA is his sleeve here. Easy money and tight OAS are the same photograph. The crash is usually the thing that was not in the model.
2007 memos. Cycle position, not a date.
Liquidity first · 1981–
Stanley Druckenmiller
Never invest in the present. Size when you are sure. Cut when you are wrong.
Soros’s successor at Quantum: liquidity drives the tape before earnings do. M2 year-over-year is the cousin on the masters board. He has said he never made money by buying a cheap multiple; he made it by being on the right side of liquidity. 2000 he shorted the Nasdaq after riding it. That is not a model you copy from a KPI. It is a temperament plus a central-bank photograph.
2000 Nasdaq. Liquidity first, valuation second.
The 200-day · 1980–
Paul Tudor Jones
Defense first. The 200-day is a risk switch, not a religion.
1987 is the speed crash on this desk. Jones’s method around that winter was risk first: if the tape is under the 200-day, size down; if the 200-day is rolling over, do not argue. The Tudor Jones sleeve here is S&P versus the 200-day. It is a photograph of trend, not a sell-everything ticket. He also said the most money is made sitting, not trading every wiggle.
1987. Speed. The 200-day as a switch.
Debt cycles · 1975–
Ray Dalio
Diversify. Know which debt cycle you are in. Cash is trash until it is king.
Principles and the long-term debt cycle: when debt is high and the central bank is out of cuts, the photograph rhymes with 1930s and 1970s more than with 1990s. Federal and household debt to GDP sit on the masters board. All-weather is ballast across sleeves — not a crash call. 2022 broke the 60/40 cousin because bonds and stocks sold together. That is on the aftermath board.
2008. 2022 60/40. The debt cycle is a map, not a date.
Valuation plus internals · 2000–
John Hussman
Rich valuations plus deteriorating internals are a warning. Rich valuations alone can run.
Hussman is the living warning that a high CAPE is not a timer. His method pairs valuation with market internals — breadth, credit, leadership. CAPE yield minus the 10-year is his cousin here. He has been early, which is the cost of treating a rich print as a peak. Use the sleeve as a photograph of how little future return is priced, not as a short ticket.
2000 and 2007 he was early-right. 2013–2021 he was early.
Mean reversion · 1977–
Jeremy Grantham
This time is never different in the distribution. It is often different in the timing.
GMO’s seven-year forecasts and the bubble essays: profit margins, CAPE, and the idea that bubbles have two parts — a superbull that humiliates the cautious, then a mean reversion that humiliates the late. Margins and Michigan sentiment are his sleeves. He named 2000 and 2008. He has also named bubbles that ran. Timing is the wound. The distribution is the method.
2000, 2008 named. Timing is the wound.
Stay the course · 1975–2019
John Bogle
Own the haystack, not the needle. Costs compound. Activity is a cost.
The crash practice is the least cinematic: a low-cost total-market fund, a written plan, and no selling because VIX is 40. Household equity share (Z.1) is the Bogle sleeve here — how much of financial assets sit in equities. He would rebalance, not time. He would not buy a story about oversold RSI. Stay the course is a practice only if the course was written before the winter.
Do not sell the haystack because the winter arrived.
Barbell, not the middle · 2001–
Nassim Nicholas Taleb
Antifragile: convex to disorder. The middle is where the crash lives.
Fooled by Randomness, The Black Swan, Antifragile. The crash practice is a barbell: mostly very safe, a little very convex, nothing that is mildly risky and leveraged. Cboe SKEW is the cousin on this desk — the price of tail insurance. A washed RSI is not a Taleb buy. A fat left tail with no insurance is the thing he would not carry. Optionality, not a forecast.
2008 as a fat tail. Insurance, not a dip.
Cut losses. Sit on winners. · 1907–1940
Jesse Livermore
The big money is in the sitting. Never average a loss.
Reminiscences of a Stock Operator is still assigned on crash desks. 1929 is the winter on the overlay. His method, when he kept it: wait for the tape to confirm, size, and do not argue with a trend that has turned. Averaging the first −5% is the opposite of his rule. He also blew up when he ignored it. The photograph is the rule, not the biography.
1907, 1929. Never average a loss.
Reflexivity · 1969–
George Soros
Markets influence the fundamentals they are supposed to discount. When wrong, get out.
The Alchemy of Finance: a boom is not a mispricing that waits; it changes earnings, credit, and the story, until it cannot. 1992 sterling is the famous trade. The crash practice is not the theory. It is the size-and-cut: when the thesis is broken, flatten. Reflexivity says a rich CAPE can get richer because the richness funds the earnings. That is why a peak is not called.
When the thesis is broken, flatten. Do not marry the analog.
Know what you own · 1977–1990
Peter Lynch
The stock does not know you own it. Do not sell a business because the market is down.
Magellan’s method was bottom-up: understand the business, ignore the macro analog. In a crash that is either liberation or ruin — liberation if you know the cash flows, ruin if you only knew the ticker. He told people the real risk was selling good companies in a winter to buy them back higher. This desk does not pick stocks. The practice is: if you cannot name why you own it, you will not hold it.
1987 he held businesses. The ticker is not the company.
Commodities and cycles · 1973–
Jim Rogers
Buy what is hated, if the cycle is turning. Ignore the street’s last five years.
Rogers is the commodity cycle on this desk: copper over gold as a real-activity cousin. 1973 was oil. A crash in equities is not always a crash in stuff, and 2022 was a reminder. He would not buy an index because RSI is 28. He would ask which cycle just ended. The photograph is relative, not a date.
1973 oil. 2022 stuff versus duration.
The bond king’s map · 2009–
Jeffrey Gundlach
Watch the curve, the dollar, and credit. Equities are often the last sleeve to know.
2s10s is his cousin here. Inverted curves are a recession photograph with a lag, not a sell ticket the week they invert. 2022 he was in the duration wreck with everyone else who thought bonds always hedge. Aftermath on this desk shows that winter: stocks and bonds together. The practice is: the curve is a map. Maps are not timers.
Curve as map. 2022: bonds did not hedge.
Read the footnotes · 2000–
Michael Burry
Do the work. Size the asymmetry. Being early is a cost, not a virtue by itself.
2000 he was in cash. 2005–08 he was in CDS on housing. Case-Shiller year-over-year is his sleeve. The practice that transfers: read the instrument, size so you can be early, and do not need the crowd. The practice that does not: copy the last trade. Housing 2005 is not AI 2026. A peak is not called because a doctor in San Jose was right once.
2008 housing. The footnotes, not the last win.
Stack expected returns · 2011–
Antti Ilmanen
Add the building blocks. Do not tell a story about a multiple.
Expected Returns stacks an earnings yield, a real rate, and an excess. Tonight’s cousin is 100 over Shiller CAPE minus the 10-year TIPS yield (FRED DFII10). Below one percentage point the equity sleeve is paying little over inflation-linked. Two points is ordinary. This is not AQR’s engine and not a forecast. The practice that transfers: write the stack before you need a story. The practice that does not: treat a low excess as a sell ticket.
A stack, not a timer. The yield is the offer.
Process over story · 1988–2024
Jim Simons
If you cannot write the rule, you do not have an edge. Stories are for after.
Medallion is closed and not a model for this desk. The transferable practice is brutal and useful in a crash: the rule is written before the VIX is 40. If your “buy oversold” is a feeling, it is not a rule. If it is RSI ≤ 30 plus VIX backwardation plus OAS wide plus a second clock, it is a rule — and even then it is a photograph of ended tapes, not a ticket. Process survives the winter. Stories do not.
Write the rule before the winter. Do not invent it at the low.
What transfers to this desk
- Write the rule before the VIX is 40. Simons. Do not invent it at the low.
- Confirm two clocks. Livermore inverted: never average a loss on one sleeve.
- Cash is an option (Klarman), not a residual. Ballast, do not forecast (playbook).
- Rich CAPE is a price (Hussman, Grantham, Buffett). It is not a peak.
- Stay the course only if the course was written (Bogle). Otherwise you will sell the haystack.
- Haven is not automatic. 2022 bonds and bitcoin both sold. Dalio’s 60/40 cousin broke.
Tonight · the rules as lights · Rates · Credit · FX · Liquidity · Live guide · Social · When it happens · Back to the terminal