JOHNSON ECONOMIC TIER
JET Index · U.S. market valuation gauge
← Back to the dashboard

What each tier has meant, historically

The JET Index boils three market gauges — the yield curve, Shiller PE, and the excess CAPE yield — into one number, then rounds that number to a letter from F (worst) to A (best), with S held in reserve for conditions that have never actually occurred. The letter itself isn't a forecast — it's a snapshot of valuation and macro backdrop — but 49 years of history show each tier carries a distinct, measurable track record for what tends to happen next. This page exists to turn “today's reading is ___” into “here's what that has actually meant historically, and why.”

The same letter can arise for different reasons, though, so the first question worth asking of any reading isn't just “what tier is this” but “which gauge is actually driving it.” A tier caused by an expensive stock market behaves differently than the same tier caused by a flat yield curve or a thin earnings yield — the composite number looks identical, but the underlying story, and the historical precedent, is not.

The big picture, at a glance

Risk tiers (F, E, D) describe conditions historically tied to above-average odds of a correction. Opportunity tiers (C, B, A) describe fairly-to-favorably priced conditions, where the dominant story has been further upside, alongside a real but lower-odds chance of a correction along the way. Moving down the alphabet means the market is progressively more “priced for trouble”; moving up means progressively cheaper and safer — though as the numbers below show, cheap doesn't mean upside is guaranteed, and expensive doesn't mean a crash is imminent.

Risk tiers (F/E/D): correction likelihood, timing, size, and the cost of being wrong

Each row below is one episode — a continuous stretch spent in that tier — not one month, so a 14-month stay in D counts once, not 14 times. “Correction likelihood %” is the share of those episodes followed by a real 10%+ drop in the S&P 500 within 24 months; it's a yes/no read on whether a correction of that minimum size happened at all. The last column covers both outcomes: how deep the correction went on average when one did happen, and — when it didn't — what you'd have missed out on by staying defensive through a reading that never actually led anywhere (“never a false alarm” means F has never once produced one to measure a cost from).

TierCorrection likelihood %Avg. months to correctionDepth / false-alarm cost
F100%10.043.7% deep · never a false alarm
E80%14.034.5% deep · ~34% missed upside if false alarm
D60%21.126.0% deep · ~78% missed upside if false alarm

Opportunity tiers (C/B/A): upside, timing, and the cost of being wrong

“Being wrong” here means a correction happened anyway despite the safe reading. “Penalty” is what that correction actually cost — how deep it went, and how long it took to get back to even. (The same splits that matter for F/E/D — how a reading got there, how long it's persisted — also show up in C/B/A; they aren't broken into separate rows here because the samples get thin once you slice this far.)

TierAvg. upside %Avg. months to the peakPenalty if wrong (depth, time to recover)
C80.5%46.6−22.7%, ~15.4 months to recover
B117.2%61.7−17.7%, ~9.2 months to recover
A137.1%62.3−19.0%, ~10.7 months to recover

F Extreme risk, and exceptionally rare

In short: the rarest and most severe reading the index produces — in the only instance on record, a real correction followed every single time.

F only happens when all three gauges are extreme at once: the market overvalued, the curve weak, and earnings yield negative. Any one gauge being bad isn't enough — all three have to line up together, which is hard to produce by accident.

Example: it's happened once — Feb–Sep 2000, right at the dot-com top, with PE above 42 and earnings yield actually negative. Since it's a single episode, there's no sub-pattern to mine; the rarity itself is the signal.

Implications:

E High risk

In short: severe — roughly an 8-in-10 historical chance of a 10%+ correction within two years.

Within E, what matters isn't the label but how far below the D boundary the reading sits. A reading barely tipped into E usually means just one gauge softened a little — a wobble, not a trend. A reading sinking deep into E, close to F, usually means multiple gauges are deteriorating together — a broader, more durable kind of weakness.

Example: in 1999, PE and earnings yield were both sinking hard (PE low 40s, earnings yield negative) — a deep E that was really pre-F — and the dot-com crash followed within months. Compare 2024–2026, where PE and the curve are only mildly over the line and earnings yield is still fine — a borderline E that's really pre-D, and no crash has followed.

Implications:

D Borderline risk

In short: the most common warning sign — a 60% historical chance of a correction, but the odds swing a lot depending on which gauge is actually driving the reading.

D is the tier where which input is weak matters more than the tier itself, because the two ways into D mean different things economically. If Shiller PE is pulling the score down, that's a valuation problem — price has run ahead of earnings — and valuation problems tend to eventually resolve through a correction. If instead the yield curve and earnings yield are doing the dragging while PE stays reasonable, that reflects interest-rate conditions rather than a mispriced market, and it can persist for years without needing to correct at all.

Example: Dec 2020/Jan 2021 was the first kind — PE alone was rich (~34) while the other two were fine — and the 2022 bear market arrived about 22 months later. Mid-1995 was the second kind — the curve was flat and earnings yield thin, but PE sat at a reasonable 22-25 — and it took over three years to see any real pullback.

Implications:

C Fair value

In short: fairly valued. A correction happens about a third of the time within two years — more often avoided than not — alongside real average upside.

C spans the whole “fairly priced” range, so where a reading sits inside that range shows how much room is left before the market crosses into stretched territory. Near the B side, there's genuine room left to improve. Near the D side, it's one soft print away from a downgrade — usually a sign the rally itself is running low on fuel.

Example: July 1993 sat comfortably inside C, near the B side, with 158% of upside still ahead. January 2018 sat barely inside C, near the D side, with only 4% left — the “Volmageddon” correction followed within weeks.

Implications:

B Favorable conditions

In short: favorably priced. A correction still happens about half the time within two years, but it tends to be shallow, and the average upside is large.

A B that all three gauges agree on is a different thing from a B produced by averaging one good gauge with one mediocre one. The first means broadly healthy conditions; the second means a market that's good in some ways and merely okay in others, blended into a tier it doesn't fully earn.

Example: the genuine three-way B's (Aug/Oct 2010, Oct 2011, right after the financial crisis) averaged about 160% upside before the next top. The blended B's averaged only about 80%.

Implications:

A Excellent conditions

In short: the most favorable conditions on record. Correction risk is real but lower than any other tier, and average upside is the highest of all.

A tells you conditions are excellent right now, but it has no memory of how long they've already been excellent — so the same A reading can mean very different amounts of remaining opportunity. A fresh A describes a market early in a recovery, with most of the next bull run still ahead. A long-sustained A describes a market that's already spent a lot of that runway, even though the number hasn't moved.

Example: March 2009 — literally the bottom of the financial crisis, the month A first lit up — had roughly 280% of upside still ahead of the eventual 2018 peak. By mid-1983, 18 months into an A stretch that began in 1982, only about 100% remained before the market's next real top in 1987.

Implications:

S Unprecedented opportunity

The composite has never actually reached S in this history (1977–present), so there's no mechanism to describe.

Caution: most of the stories above are one or two illustrative episodes, not dozens of independent trials — the mechanism is likely real, but don't treat the exact numbers as precise. The tables above are the full data these stories are drawn from.

Caveats — what this isn't

← Back to the dashboard