Know the tape, cold.

The Fed raised 25bp to 4.00% at its September 15-16, 2026 meeting. Futures price another move as more likely than not, and the 10-year stood at 5.17% on 25 September 2026. Equities are up double digits on the year anyway, at a CAPE of 41.25, and the VIX is subdued. Gold is down on the month, which fits the rates side: higher yields raise the cost of holding an asset that pays no coupon. Stocks are priced for a Fed that stops. Rates are priced for one that keeps going. Both cannot be right.

Note that the VIX and dollar rows carry no red or green. Both measure magnitude rather than direction. A rising VIX is not bad and a falling one is not good, and coloring them as if they were is the most common way this table gets misread.

The VIX is also the one row with no year-to-date figure, on purpose. It is mean-reverting: it spikes and falls back to roughly the same place, so a change measured from an arbitrary starting date describes where January happened to sit, not anything about this year. The level is the information. Oil and gold do get a year-to-date, but read the note beside them first, because a futures return depends on which contract basis you use and the two answers differ by more than most people expect.

Flip trailing P/E upside down and you get the earnings yield: 1 ÷ 26.26 = 3.81%, as of 25 September 2026. That is what the index earns for each dollar you pay. The 10-year Treasury pays 5.17% with no risk, so stocks yield 1.36 points less than a bond with no default risk. The easy mistake is to call that a sell signal. It means investors are paying for earnings growth a bond cannot give them, so the question to ask is how much growth that gap assumes.

So equities currently yield 1.37 points less than government bonds. That spread is normally positive, because you are paid for taking equity risk. Negative means the market is pricing substantial earnings growth to justify the price, or investors are accepting a worse deal than cash offers. Saying "the equity risk premium is negative" and knowing what produces it is a genuine differentiator.

The orange spike to 70.9 in 2009 is not a bubble. It is a denominator collapse. Earnings nearly vanished, so trailing P/E went vertical while stocks were at their cheapest in a generation. That is the entire argument for CAPE.

Ten companies are 39.4% of a 500-stock index, and the largest alone is 7.5%. Buying "the market" is a large, undiversified bet on a handful of semiconductor and platform businesses, and index earnings growth is increasingly their earnings growth.

Everyone says this is a narrow market driven by a few mega-caps, and the concentration number supports it. The test is equal-weight against cap-weight. This year cap-weight is ahead by a few points, so the giants are doing more of the lifting than the average stock. Technology alone is 38.69% of the index by weight as of 25 September 2026. The easy mistake is to hear "narrow" as "everything else is falling". Equal-weight is up double digits too. Narrow leadership means the average stock is lagging. It does not mean it is losing.

Both are true and not in conflict. Concentration measures weight. Breadth measures participation. Holding two apparently contradictory facts and explaining why both hold is exactly the reasoning an interviewer tests for.

Read whether the stock rose, not whether the company beat. That column kills the instinct every student arrives with: that a beat is good news and a miss is bad news. Companies guide analysts down to a number they can clear, so beating is the base case. The bar that actually matters is what investors privately expected, and it does not appear in this table. A stock that falls on a beat tells you the private bar was higher. A stock that rallies on a miss tells you it was lower.

Notice there is no relationship between the size of the beat and the size of the move. Caterpillar beat EPS by 31.8% and rose 5.0%. Walmart beat by 9.5%, raised full-year guidance, and had its worst day in four years. Lilly beat by 27.4% and rose 6.8%, while Arista beat by 14.6% and moved 1.0%. If the beat drove the reaction, this column would sort with the one beside it. It does not.

What moved these three was the part of the release the table cannot hold. Walmart's US comparable sales grew 2.6% against roughly 3.6 to 3.7% expected, the slowest in six years, and the guidance raise leaned on a tariff refund rather than on trading. Cisco beat on both lines and guided above consensus, then said gross margin fell to 66.3% from 68.4% because memory costs rose and the sales mix shifted toward lower-margin AI hardware. AMD's print was fine and the outlook was not. In all three cases the market repriced the forward number, not the reported one.

So when an interviewer asks how a company did, the answer is never the beat. It is what the beat was made of, what management said about next quarter, and what the stock did with both.

The 2-year tracks where the market thinks policy is going. The long end is growth plus inflation plus term premium. When the 2-year rises 10bp and the 10-year falls 1bp, the market is saying: tighter policy now, no better growth later. That is a bear flattener, and it is hawkish.

All four by name. Bull steepener, front end falls fastest, cuts coming. Bear flattener, front end rises fastest, which is what we have. Bear steepener, long end rises fastest, inflation or supply worry. Bull flattener, long end falls fastest, a growth scare.

Never quote a probability without its timestamp. These move hourly, and an interviewer who checked this morning will catch a stale number. Say "as of Friday's close it was roughly a coin flip" rather than a false-precision figure.

Monetary policy does not work through the fed funds rate directly. It works by transmission. The policy rate moves bond yields, which move credit spreads, lending standards, mortgage rates and equity valuations, which move spending and hiring months later.

Two readings on this table deserve a second look. The NFCI is negative and getting more negative: below zero means looser than the historical average, so a fall of 0.026 over the month means conditions loosened further while the Fed debated hiking. And bank lending standards at exactly 0.0% is not a missing value. It is a net percentage, tighteners minus easers, and the July survey netted out even. The Fed's own write-up says banks "left standards basically unchanged, on net."

When the rate is held at a restrictive level but conditions loosen anyway, transmission is leaking. That is exactly why three members dissented to hike: from their seat, holding is not holding. It is easing. Saying "the policy rate is restrictive but financial conditions aren't" puts you ahead of most candidates.

The two hiring counts disagree. Official payrolls rose +162k in August 2026 against a Trading Economics consensus of 56K, while ADP had private hiring at just 38K. The two use different samples, so some gap is normal. A gap this wide means neither one alone is the labor market. Claims settle the firing side: 197,000 in the week to 19 September 2026, with a four-week average of 202,250, so firms are not letting people go. Unemployment held at 4.1% while participation rose to 61.6%. More people were looking for work and the rate still did not rise, which is a strong reading and not a flat one.

The consumer is what actually deteriorated. Retail sales fell 0.6% against +0.1% expected, with the control group that feeds GDP down 0.4%. Michigan sentiment dropped to 51.7 from 55.2. Consumption is roughly two thirds of GDP, so that matters more than one payroll print.

And PPI came in soft at 0.0% against +0.2% expected. Producer prices sit upstream of consumer prices and several PPI components feed straight into the PCE calculation. Soft PPI alongside hot core PCE is the one piece of evidence pointing toward inflation cooling from here.

The bar is what printed. The horizontal tick across it is what the street expected. The distance between them is the surprise, and that is what markets actually price. August 2026 printed +162k against a Trading Economics consensus of 56K, so the gap is 106,000. A strong number everyone forecast moves nothing. A decent number nobody expected moves a lot. That is why two bars of the same height can mean opposite things in different months.

The shape matters more than the 4–3 record. The four beats cluster January to May. Three of the last four missed, and both large misses sit in the recent stretch. A run of surprises in one direction is what repriced the front end of the curve.

Then read the "Now" column. May was first reported at +172k and stands at +63k, cut 109,000 across two rounds. June went +57k to +20k. But note the honest nuance: January, March and April all revised up, by 99,000 combined. This is not systematic overstatement. It is that the most recent, least-mature months keep getting cut, which is a different and more useful claim.