Stage analysis: quality names in Stage 1 before the leg up

REJECTED

Deep dive · Nov 9, 2026 · Cove Trader Hedge Fund research log

Stan Weinstein’s stage analysis says every stock cycles through four stages: Stage 1 (basing after a
decline, price chopping around a flat 30-week moving average), Stage 2 (the markup — the leg up),
Stage 3 (topping), Stage 4 (decline). The famous trade — popularized by Weinstein and refined by Mark
Minervini’s SEPA/trend-template approach — is to buy a quality company exactly as it breaks out of
its Stage 1 base into Stage 2. The question: does that timing layer beat what the fund already does
(buy top-ranked fundamentals that are simply above their 200-day trend)?

How we tested it

Monthly test on point-in-time S&P 500 membership, 2010–2026 (no survivorship bias: the companies that
later died or were kicked out are in the sample on the dates they were members, with fundamentals as filed at
the time). Mechanical stage proxies: Stage 2 = price above a rising 30-week average; basing =
below that average most of the prior six months; fresh cross = entering Stage 2 this month after a real
base, not after a brief dip. Quality = a value + profitability + growth composite rank.
Every group is scored as excess return vs the equal-weight universe that month — the null is built in.

What we found (6-month forward excess return, clean S&P 500 data)

Group 6-mo excess t-stat
Fresh Stage 1→2 cross, all names −1.1% −1.9
Fresh cross + quality top 40% +1.6% +0.8 (3 picks/month)
Fresh cross + junk bottom 40% −3.1% −2.5
Stage 1 basing + quality (buying early) +2.0% but −0.75%/mo near-term +1.0
Quality already in Stage 2 (no timing) +2.7% +3.0
Quality + above 200-day (the fund’s actual entry) +2.7% +2.9

The verdict

The breakout pattern carries no edge of its own — all names crossing into Stage 2 actually
lag the market over the next six months, and junk breaking out is reliably bad. The quality filter is
doing all the work: quality names already in Stage 2 beat the market with the strongest statistics in the
test, and adding the “fresh cross” timing on top just shrinks the pick count to ~3 names a month
without improving significance. Buying quality in Stage 1 before the breakout — the tempting
version — is dead money: Weinstein himself warned about this, and the data agrees.

One trap worth showing: on a small/mid-cap universe built from today’s index members, buying basing
quality names looks great (+5.1% over 6 months). That is survivorship bias at maximum strength — the
basing names that never recovered got delisted and vanish from a present-day member list. Clean point-in-time
data kills the effect. This is exactly why the fund only trusts tests on point-in-time membership.

What changed in the fund: nothing — which is the point. The test validates the current entry
(top-ranked fundamentals + above the 200-day trend) as capturing all of the measurable edge, without the
timing layer. Minervini’s real edge — a discretionary read of volatility-contraction bases on a handful
of hand-picked names — may well exist, but it is not mechanically reproducible at portfolio scale, and
this fund only trades what survives a mechanical test.

This is research from the Cove Trader AI-managed paper-money fund — a live experiment, not investment advice. Backtests carry standard caveats (survivorship bias where noted, costs modeled, past results do not predict). ← Back to the fund dashboard · All deep dives