MANH › analyze
MANH · Analyze
A portfolio-specific passage was removed from the public build.
This pass was commissioned to settle two specific questions: (a) does the $150–210 entry zone survive a from-scratch re-derivation, and (b) is [8.5] conviction consistent with a WATCH verdict? Both are answered in §3 and §4. The short version: the zone was never a zone, and the answer to (b) is that conviction and verdict were both slightly wrong, in the same direction.
1. What This Updates
The August 27 baseline was written at $223.76, one day after a new all-time-since-ATH intraday high of $227.03. It held conviction at [8.5], held the verdict at WATCH, re-anchored the entry ladder upward, and named four things to watch: (1) the Q3 print, (2) whether insider selling accelerated, (3) whether the $195–210 zone got tested, (4) whether consensus targets converged to spot or spot fell to meet them.
Three of those four resolved during this window, and all three resolved in the baseline's predicted direction. That is a real result and it is recorded as a force in the baseline's favour before anything else is judged.
Event list, 2026-08-27 → 2026-09-28
A portfolio-specific passage was removed from the public build.
Only one item on that list is a business event — the Editions launch — and it is unquantified. The rest is an 10.8% drawdown made entirely of multiple compression against an unchanged guide.
The sharpest single observation of the window: estimate revisions in it were upward (Zacks FY26 +$0.14, five analysts up; Baird +$42 and Barclays +$38 on price targets) while the stock fell 11.0%. That is a multiple-compression signature, not an estimate-cut signature, and it is the fact the whole adjudication below turns on. This is a Price-row pass, not a Structural-row pass — with one genuine exception, the margin row, which turned out to have been mis-stated in the baseline rather than changed by any new event.
2. The Delta Ledger
30 baseline claims carried the verdict and were re-tested. Status counts: 13 CARRIED · 6 REFRESHED · 2 DRIFTED · 3 SUPERSEDED · 1 RETRACTED · 5 UNTESTED (+9 NEW).
❌ RETRACTED
| # | Claim (baseline) | The error | Cause |
|---|---|---|---|
| 21 | Trim 46x fwd — "implies ≈$251.6 at the current $5.47 FY26 non-GAAP guide-mid EPS; recomputes automatically as EPS climbs" |
The valuation judgment (~$252 ceiling) was defensible. The encoding was wrong, and it has been rendering a wrong number on the live site since August 27. site.py's trim_from_multiple() computes EPS = price / prof["forwardPE"] — it has no independent EPS field. Yahoo's forwardPE for MANH is 32.38, which divides an EPS of $6.1432 — the FY2027 consensus, not the FY26 guide-mid of $5.47 the baseline wrote the multiple against. So the site renders 46 × ($199.63 / 32.38) = $284, against an intended ~$252: +12.6%, silently. |
Vendor trap, and a known one. pitfall-multiple-trim-inherits-the-broken-vendor-field was written 2026-08-12 — fifteen days before the baseline — and says in terms: "Before writing a multiple-trim, verify the specific vendor field that basis will use." The baseline did not. It also flagged the wrong-fiscal-year problem elsewhere in its own text ("Vendor forward-EPS again on the wrong fiscal year; used the company's own FY26 guide") and then wrote a trim that depends on exactly that field. It caught the trap for the multiples table and missed it for the trim. |
The fix, applied below: a ceiling of ~$230 must be written as 37x fwd, because fwd selects
Yahoo's FY27-based field. 37 × ($199.63 / 32.38) = $228. Writing "42x fwd" — the same ceiling
expressed on the FY26 guide — would render $259 and repeat the defect. Both expressions of the
judgment are stated in §6 so a human reader is not misled by the smaller-looking number.
🔄 SUPERSEDED
| # | Claim | Old | New | Why |
|---|---|---|---|---|
| 18 | Fair value band | $195–250, midpoint $222 (= 36–46x FY26 guide-mid $5.47) | $180–230, midpoint $205 (= 33–42x FY26) | Re-derived from scratch. Following pitfall-stale-fair-value-is-most-costly-on-winners rule 3, inputs and assumptions are reported separately: the inputs did not move against the band — they moved slightly for it. FY26 guide unchanged at $5.47; FY27 consensus of $6.14 (+12.3%) is a new input and it is supportive. The band fell because the assumption changed: the 36–46x multiple range was set on Aug 27 into a rate-cut-hope tape (the baseline's own "Aug 14 cooling PPI/CPI basket rally" row). Since then the Fed has raised 25bp to 3.75–4.00% with ~60% odds of another on 10/28, and the 10yr sits at ~4.96%. A zero-yield name at 44x ex-SBC free cash flow does not hold a 46x ceiling into a hiking cycle. Cut to 33–42x. Cross-check: $180–230 is 29–37x the FY27 consensus $6.14, which is where a 12%-grower with an intact wide moat and no dividend belongs. |
| 20 | Entry ladder T1 $195–210 (40%) / T2 $175–195 (35%) / T3 $150–170 (25%) — collapsed in frontmatter to entry: 150-210 |
Anchored to the $195–250 band | T1 $180–195 (40%) / T2 $162–180 (35%) / T3 $140–160 (25%); frontmatter entry: 180-195 |
Price rows never carry forward. Re-derived against the new $180–230 band. Separately, the frontmatter flattening is itself corrected — see §3(a). |
| 29 | Verdict WATCH — "DO NOT CHASE, HOLD ON WATCHLIST, NO INITIATION" | WATCH | DEFER | Not a change of opinion about the business; a change in what the report is honestly able to decide. See §4(b). |
📉 DRIFTED
| # | Claim | Old | New | Breaks at |
|---|---|---|---|---|
| 12 | "Margin-compression flag (Q1 FY26 op margin 23.0% vs 25.2%) — still the last data points on record; next test is Q3." | Q1 FY26 = 23.0% | The Q2 FY26 figure was already on record and the baseline did not use it. Recomputed from the quarterly income statement, as-reported operating margin runs 27.09% → 27.49% → 24.78% → 23.01% → 22.24% (Q2 FY25 → Q2 FY26): four consecutive declines, and Q2 FY26 at 22.24% sits 24bp above the 22% break trigger. | The trigger — op margin <22% for two consecutive quarters — is one weak quarter from firing. Basis caveat, and it is material: Q2 carried an $8.263M restructuring charge (the 6% workforce reduction). Ex-that charge the margin is 25.02%, i.e. a recovery, not a decline. The baseline never stated which basis its own trigger runs on. It is resolved here as GAAP as-reported, which is the conservative reading and the one consistent with the trigger's wording. |
| 27 | "All five break triggers — none fired, none testable this pass." | None near firing | Four of five remain untestable. The fifth (op margin <22% 2Q) is now within 24bp on a GAAP as-reported basis and was already within 24bp when the baseline said none were near. | See above |
🔁 REFRESHED
| # | Claim | Old (Aug 27) | New (Sep 28) | Source |
|---|---|---|---|---|
| 16 | Yahoo EV field is stale; correct EV ≈$12.91B | Vendor understated by ~$450M (and by $2.2B on Jul 29) | Vendor EV $11.88B vs manual (MktCap $11,639M + debt $53.9M − cash $186.1M) = $11.51B — vendor now overstates by $373M. | The claim sharpens: the field is not "stale low," it is unreliable in both directions. Three consecutive passes, three different errors, two different signs. Manual recompute used throughout. |
| 17 | Multiples at $223.76 | Fwd P/E 40.9x · Trailing 64.1x · EV/FY26E rev 11.1x · P/FCF 34.9x · P/adjFCF 49.6x · Rule of 40 ≈32.6 (fails) | Fwd P/E 36.5x (on FY26 guide-mid $5.47) · Trailing 57.2x (TTM GAAP $3.49) · EV/FY26E rev 9.9x · P/FY25 FCF 31.1x (yield 3.21%) · P/adj FCF 44.3x (ex-SBC, yield 2.26%) · Rule of 40 ≈30.6 (still fails) | .mcp/fin.py, corrected EV. Yahoo reports fwd P/E of 32.4x. That is the wrong fiscal year — see N2. Using it would understate the multiple on the guided year by 12.6%. |
| 19 | Bogle scenarios from a 40.9x start | 41x → 10–12%/yr; 34x → 7–8%; 28x → 2.7–4.7%; 22x → −1.7% to +0.3% | From 36.5x: holds 36.5x → 10–12%/yr; drifts to 30x → 6–8%; reverts to 25x → 2.7–4.7%; reverts to 22x (2022-style reset) → 0.4–2.4%/yr | Recomputed. Every rung improved ~2pts vs. Aug 27 purely on the lower entry multiple — this is the honest pro-MANH consequence of the drawdown. |
| 25 | Consensus target $213.90 mean, 4.4% below spot | 4.4% below | $220.70 mean / $220.00 median, 10.6% ABOVE spot ($199.63). Range widened to $180–260 (low: DA Davidson-adjacent; high: Baird), 10 analysts. | The baseline explicitly predicted this resolution — "watch whether it closes the gap (bullish confirmation) or spot pulls back to meet it (the more likely resolution for an unowned watchlist name)." Spot pulled back. Recorded as a force in the baseline's favour. |
| 26 | Three insider sales since Jul 29, ≈$1.1M combined, no buying | Net sellers into strength, small size | Zero insider transactions of any kind since 2026-08-11. No sales through the entire $220+ window, no buying on the way down to $199. Widened: zero insider buys in the trailing 12 months — every open-market transaction in a year is a sale; the only acquisitions are $0-cost grants and gifts. Insider ownership 1.35%. | The baseline's #2 watch item — "a jump back to five- or six-figure share blocks at $220+ would be the tell" — did not happen. A predicted bearish signal that failed to appear is a force in favour, and is recorded as one. The absence of buying at −11%, against a 12-month record of no buying at all, is a real offset that keeps this from being a strong positive. |
| 28 | Conviction [8.5] | 8.5 | [8.0] | See §4. |
✅ CARRIED
Re-tested, still true. None had a falsifying event available this pass unless noted.
A portfolio-specific passage was removed from the public build.
Also carried by non-application: Graham IV = $14.55 vs a $199.63 price — the model is broken here, not the stock. BVPS is $2.70 because a decade of buybacks consumed the equity account; the assets are code and customer relationships. Discarded, as in both prior passes. DYT and DDM remain N/A — no dividend, no plan for one, zero current yield. This matters for a book rotating toward income: MANH's total return is 100% dependent on multiple maintenance.
⏳ UNTESTED
| # | Claim | Passes deferred |
|---|---|---|
| 6 | Blue Yonder / Panasonic is the real competitor — watch win rate; a fall to the low 60s is the earliest observable damage | No win-rate data landed. 3rd consecutive pass. |
| 11 | Cloud subscription +26.2%, cloud = 42.5% of revenue, non-cloud −0.5% and shrinking as a share | Next data point is Q3. 2nd pass. |
| 14 | Buyback pace — Q2 share count −3.4% yoy, "worth watching if it holds" | No new quarterly share count. 3rd consecutive pass, and it was on the July baseline's own open-item list. |
| 22 | RPO $2.47B, +23% yoy, 2.1x forward-revenue coverage | Next data point is Q3. 2nd pass. |
| 24 | Active Agents >10% installed-base penetration, 100% pilot-to-subscription conversion, win rates >70% | Explicitly deferred to Q4 FY26 by the July baseline. 3rd consecutive pass. |
Five rows, three of them deferred for a third consecutive pass. This is called out per the protocol: the system keeps deferring the same questions because the calendar has not delivered them, not because anyone forgot. But three passes is the point at which it stops being a calendar problem and starts being a conviction problem — see §4(b).
🆕 NEW
| # | Finding | Strength |
|---|---|---|
| N1 | The margin compression is identifiable growth spend, and it can be sized. Q2 FY26 vs Q2 FY25: Selling & Marketing $19.98M → $30.70M, +53.7%, while R&D was flat at −0.3% ($34.87M → $34.77M). The entire operating-margin drag is sales capacity, not cost inflation and not AI R&D. This corroborates with a number management's claim that sales-capacity investment made a year ago is what produced three record bookings quarters — a claim the July report accepted on the CEO's word alone. | Primary-source (quarterly income statement), single-vendor, arithmetic verified |
| N2 | FY27 consensus EPS is $6.1432 — the first quantified test of the "2027 inflection" claim, and it falls short of the bull framing. That is +12.3% on the FY26 guide-mid of $5.47. The July report's bull case was a 2027–2028 acceleration of total growth toward the 26% cloud rate. +12% EPS growth is a good compounder; it is not that inflection. Counter-force, and it is a real one: FY26 estimates moved UP in this window — Zacks FY26 consensus rose +$0.14 to $5.47 with five analysts revising up in 60 days, which is what drove the Sept 1 Zacks upgrade to #2 (Buy). So the near year is being marked up even as the multiple compresses. Separately, this row is pitfall-vendor-forward-eps-is-the-wrong-fiscal-year firing live for the third pass running — Yahoo's forwardPE 32.38 / forwardEps $6.1432 are FY2027; every multiple in §2 is computed on the FY26 guide instead. |
FY26 revision corroborated (Yahoo epsCurrentYear $5.4976 + Zacks $5.47). FY27 $6.14 single-source (Yahoo); no FY27 revision magnitude was obtainable. Used as direction only. |
| N3 | DA Davidson downgrades Buy → Neutral on 2026-09-28, price target at $210. First downgrade since the Q2 re-rating and the proximate cause of today's gap-down. The rationale is explicitly and only valuation — stock +70% off April lows vs IGV +42%, trading "nearly in line with peers showing 20%+ growth at 29x 2027 FCF", so risk/reward is "more neutral". The firm expressly reaffirmed MANH's WMS/TMS category leadership and the cloud-conversion + AI thesis. A downgrade that keeps its target and restates the thesis is a Price force wearing a Structural costume — it is recorded against the Price rows, and it moved no Structural row. | Corroborated (rating feed + named-analyst news) |
| N7 | Manhattan Active® Editions (Sept 22) is a packaging change, not a feature release — and it is the only genuine business event of the window. Three tiers (Essentials / Enterprise / Enterprise Premier) across ActiveWarehouse, Transportation, Store and Order, pitched at downmarket and mid-market customers — "accessible at every stage of growth" — with upgrades between tiers explicitly requiring no re-implementation, retraining, data migration or operational disruption. Direction is genuinely two-sided: it reinforces #1 and #4 (the switching-cost and embedded-workflow mechanism now has a formal land-and-expand ladder attached to it, and a lower-friction on-ramp widens the installed base the AI agents monetise), while introducing an unpriced blended-ASP and gross-margin risk — a downmarket tier sold into smaller accounts is structurally lower-margin than the enterprise base. No pricing, no TAM figure, no ARR or RPO impact was disclosed, so neither side can be sized. Notably the tape reaction was −1.1%, reversing from +2% intraday. Also in the window: the Boscov's win (Sept 1, 51 stores, Active Order) — a real logo, no contract value disclosed. | Primary-source (company PR, widely syndicated). Commercial impact entirely unquantified — moved no status. |
| N8 | Short interest rose 2.18M → 2.95M shares, +35% month-on-month — now 6.6% of float, 4.08 days to cover. A meaningful build against a name whose July report described the short side as negligible. | Single-source (aggregator) |
| N9 | A fundamental bear datapoint worth carrying forward: Q2 billings $285.4M, with a four-quarter average growth rate of just 6.5%. That sits below reported revenue growth of 9.3% and far below RPO growth of +23%. Billings, revenue and RPO are telling three different stories and the gap is not yet explained. RPO can outrun billings on long-duration contracts — which is the benign reading and is consistent with the cloud transition — but 6.5% billings is also the single most direct measure of current demand in the file, and it is the slowest number MANH produces. Sourced from an algorithmic screen (StockStory, 9/25) rather than the company; it was not verified against a filing and moves no status, but it is the most useful question this window produced for the Q3 call. | Single-source, algorithmic screen. Unverified — flagged as a Q3 test, not used. |
| N4 | Sell-side dispersion is the widest in this analysis chain: $180 low to $260 high (Baird raised to $260 on 8/26; Barclays to $239 on 9/8) against a $220.70 mean. A 44% spread across 10 analysts on a name with an unchanged guide means the street is disagreeing about the multiple, not the numbers — which is exactly what this report concludes independently. | Corroborating |
| N5 | The rate regime flipped inside the window. FOMC raised 25bp on 9/16 to 3.75–4.00%; ~60% odds of another on 10/28; 10yr ~4.96%. The Aug 27 band was set days after a rate-cut-hope rally the baseline itself flagged as a basket move. This is the single largest force on the Price rows and the stated reason the fair-value multiple assumption was cut. | Macro, corroborated |
| N6 | The Q3 print lands inside a guide-punished, not miss-punished, season. This desk's standing observation for the current season is that several software names beat the quarter and fell 8–20% on forward guidance. MANH's Q3 will carry the first commentary shaping FY27 expectations against a consensus that already embeds +12.3%. The asymmetry around 10/27 is therefore wider than the quarter itself implies, and it is the direct justification for DEFER over WATCH. | Repo standing observation |
3. How the Close Calls Were Decided
(a) The $150–210 entry zone — it was never a zone, and it is now narrowed 5-fold
The zone does not survive, but not for the reason the question implies. entry: 150-210 was
never an analytic judgment. The baseline's actual structure was a three-tranche ladder —
$195–210 (40%) / $175–195 (35%) / $150–170 (25%) — with different sizes and different rationales
per rung. The frontmatter field collapsed the union of all three into one string, and the
Watchlist row then rendered that union as "✅ IN $150-210 — top of band."
That is a frontmatter flattening artifact, and it produces a live false positive. A 40%-wide
band will report a name as "in zone" across essentially any price it trades at, which is the
mirror image of pitfall-stale-entry-zone-suppresses-a-name: that note records a zone set
beyond reach silently suppressing a name; this is a zone set too wide silently promoting one.
The Watchlist row already half-diagnosed it — "its zone lives only in report frontmatter" — but
treated the symptom (invisibility to scans) rather than the cause (a three-rung ladder flattened
into a single band). MANH has been reading as in-zone-and-actionable at $206, $210 and $223 on a
verdict that said, in terms, do not initiate. Conviction [8.5] + a green zone marker + WATCH is
not a coherent row.
Resolved three ways:
1. The band is re-derived from scratch against the new $180–230 fair value: T1 $180–195 / T2 $162–180 / T3 $140–160.
2. Frontmatter entry: carries T1 only — 180-195 — because T1 is the only rung that should fire an "in zone" scan. T2 and T3 live in the body where their conditions can be read. Width goes from 40% → 8.3%.
3. At $199.63, MANH is 2.4% ABOVE the top of T1 — it is NOT in zone. The honest reading of the last month is that the name fell 10.8% and got close to actionable, not that it became actionable.
(b) Conviction [8.5] against a WATCH verdict — both were wrong, in the same direction
The framework's separation of business conviction from purchase attractiveness is correct and is not the problem. An 8.5 business at a bad price is properly a WATCH. The defect is narrower and it is real:
[8.5] was raised from [8.0] on July 29 on four specific pieces of Q2 evidence — three record bookings quarters, RPO at 2.1x coverage, Active Agents at 10% penetration with 100% conversion, and win rates >70%. Two months and two passes later, not one of those four has been re-tested. Rows #22 and #24 are UNTESTED for the 2nd and 3rd consecutive pass; #6 (win rate) for the 3rd;
14 (buyback pace) for the 3rd. The command's own rule applies: *"the conviction score should
reflect the volatility of your own analysis, not just the company's."* A score standing on evidence that has sat unverified across two passes has decayed whether or not the business has.
Against that, two rows moved genuinely against the thesis this pass — #12, where the margin series turns out to be four consecutive declines sitting 24bp from a break trigger (and the baseline mis-stated the available data), and N2, where the first quantified FY27 number (+12.3%) undershoots the "acceleration toward the 26% cloud rate" that was the July bull case.
Three rows moved in the baseline's favour and are why this is one notch and not two. The strongest is the one that argues hardest against cutting at all: FY26 estimates were revised UP inside the drawdown — Zacks consensus +$0.14 to $5.47, five analysts, sixty days. A stock falling 11% while its near-year estimates rise is being re-priced, not re-rated downward on fundamentals, and that is a direct argument for holding conviction where it was. It is joined by N1 (the margin drag is sized, identifiable sales-capacity spend, not structural cost creep) and #26 (the predicted heavy insider selling at $220+ did not appear at all — though that row is discounted here, because it is unverified negative evidence and no insider has bought in a year either).
That is why the cut is one notch and not more, and why it is explicitly not a call that the business deteriorated. It is a markdown of how fresh the evidence under an 8.5 actually is.
Net: [8.5] → [8.0]. And the verdict moves too, in the opposite direction from what a conviction cut usually implies: WATCH → DEFER.
DEFER is the correct verb under CLAUDE.md because this report is not decision-ready until a
specific dated event. MANH is 29 days from the Q3 FY26 print, inside a guide-punished season
(N6), with the first realistic entry rung 2.4% below spot, one break trigger 24bp from firing, and
five claims — including all four that earned the conviction raise — untestable until that print.
Calling this WATCH implies a considered "no" on the merits. The truthful statement is that
there is no call to make here before October 27, and the ledger says so explicitly rather than
dressing a deferral up as a decision. recheck is set to 10/28, the day after.
So: neither number alone was the error. Conviction was one notch too high (stale evidence), and WATCH was too passive a label for a name a month from the event that settles it. [8.0] + DEFER is the consistent pair.
(c) Why the fair value was cut on a name that already fell 11%
This is the uncomfortable one, and it is the inverse of the QLYS case in
pitfall-stale-fair-value-is-most-costly-on-winners. There, re-valuing a winner raised the
ceiling 21% and still returned TRIM. Here, re-valuing a faller lowered the ceiling 8% at the
midpoint — which reads like chasing the price down, and would be exactly that if it were done on
price action.
It is not. Applying that note's rule 3 — change the inputs before changing the assumptions, and report both separately:
- Inputs: FY26 guide unchanged. Balance sheet unchanged. CAGRs unchanged. FY26 consensus was revised UP $0.14 to $5.47 by five analysts inside this very window, and FY27 consensus is a new input at +12.3%. Every input that moved, moved in the stock's favour. On inputs alone the band goes up, not down.
- Assumption: the 36–46x target multiple range. That range was set on August 27, days after a rally the baseline itself identified as a rate-cut-hope basket move. On September 16 the Fed hiked instead, with another ~60% likely on October 28 and the 10yr at ~4.96%.
The band fell because the discount rate rose, and for no other reason. Said plainly: the business did not get worse — estimates went up — the price of a dollar of future earnings did. That is a legitimate reason to move a multiple assumption and an illegitimate reason to move conviction, which is why #18 is SUPERSEDED while the Structural and Trend rows are almost entirely CARRIED.
Two independent parties reached the same place from outside this desk in the same week, which is worth recording as corroboration of the level rather than of the reasoning: Simply Wall St's narrative fair value is $180 — precisely this report's new floor — and DA Davidson's $210 sits inside the new band while the firm reaffirmed the business thesis. The three of us disagree about almost everything except that the multiple, not the company, is the live question.
(d) Corroboration check — what is resting on one source
Three conclusions here rest thinly and are flagged rather than elevated:
- N2 (FY27 +12.3%) is a single vendor consensus field, and it is the same field family this repo has caught mislabelling fiscal years. It is used as a direction (the inflection is not showing up in 2027 estimates) and never as a fact; no Structural or Trend row was moved on it. A transcript or an IR-deck read would corroborate it and was not obtained.
- N3 (the downgrade) is one firm's opinion. It moved no claim; it explains today's tape.
- #12's basis resolution (GAAP as-reported rather than ex-restructuring) is a judgment call made by this report, not a disclosure by the company. It is the conservative reading and it is stated so the next pass can overturn it cheaply.
4. Thesis Persistence and Conviction Delta
Thesis persistence: 10 of 14 Structural + Trend rows CARRIED = 71% (baseline claimed 100%).
The drop needs honest decomposition. Three of the four non-carried rows are UNTESTED, not broken — #6, #11, #14 all await the Q3 print. Only #12 (margin) genuinely drifted. Measured on rows that could actually be tested, persistence is 10 of 11 = 91%.
But the 71% is the more useful number, and it is the honest correction to the baseline's "100% persistence." The baseline computed 100% by counting untestable rows as survivors. A claim that could not be tested did not survive a test; it simply was not tested. This report counts it the other way and recommends the next pass do the same.
Conviction: 8.5 → 8.0. Driving rows, named:
| Direction | Row | Weight |
|---|---|---|
| ↓ | #12 margin four consecutive declines, 22.24% vs a 22% trigger; baseline mis-stated the data on record | Heaviest single input |
| ↓ | N2 FY27 consensus +12.3% — first quantified read on the 2027 inflection, and it undershoots the July bull case | Moderate; single-source, used as direction only |
| ↓ | #6, #14, #22, #24 the four rows that earned the 8.0→8.5 raise, untested across 2–3 consecutive passes | Moderate — this is analysis volatility, not company deterioration |
| ↑ | N2 (counter-force) FY26 consensus revised up $0.14 to $5.47 by five analysts inside the drawdown | Real positive force |
| ↑ | N1 margin drag is sized, identifiable S&M growth spend (+53.7%), R&D flat | Real positive force |
| ↑ | #26 predicted heavy insider selling at $220+ did not materialise | Positive, but discounted — it rests on unverified negative evidence (§6), and there has been no insider buying in 12 months either |
| ↑ | N3 the only downgrade of the window held its target and expressly reaffirmed the thesis | Mild — a valuation call, not a thesis call |
| ↓ | N8 short interest +35% m/m to 6.6% of float | Mild |
| ? | N7 Editions: wider on-ramp vs. downmarket ASP dilution, both unsized | Deliberately zero weight — recorded, not used |
| ? | N9 billings four-quarter average ~6.5% vs RPO +23% | Zero weight this pass (unverified); potentially heavy at Q3 |
| → | #1–5, #7–10, #13, #15, #23 the entire moat and balance-sheet core | Unmoved; this is why it is 8.0 and not lower |
8.0 is still a high score and it is meant to be. Nothing in this pass touched the moat, the cash generation, or the balance sheet. The cut is about the freshness of the evidence under the score and one genuine margin drift — not about the quality of the business.
5. Updated Verdict
Action: DEFER — no call until the Q3 FY26 print (~Oct 27, 2026). Do not initiate at $199.63.
A portfolio-specific passage was removed from the public build.
Fair value: $180–230 (midpoint ~$205) — cut from $195–250 on the multiple assumption, not the business.
| Basis | Range |
|---|---|
| FY26 non-GAAP guide-mid $5.47 | 33x – 42x |
| FY27 consensus $6.14 | 29x – 37x |
| Spot $199.63 sits at | 36.5x FY26 / 32.5x FY27 — the 53rd percentile of the band |
Entry ladder (replaces $195–210 / $175–195 / $150–170)
| Tranche | Zone | Size | Rationale |
|---|---|---|---|
| 1 | $180–195 | 40% | Bottom third of the band, 2–10% below the $205 midpoint. The first rung that pays a real discount rather than the midpoint. Currently 2.4% below spot — not in zone. |
| 2 | $162–180 | 35% | Below the fair-value floor; 12–21% under the midpoint. The market discarding the Q2 re-rating without a thesis break — the best risk/reward on the list. |
| 3 | $140–160 | 25% | A broad SaaS multiple reset or a fired break trigger. Accumulate only if none of the five triggers have fired. |
Frontmatter entry: carries T1 only (180-195) — deliberately, so scans fire on the rung that
is actually actionable. Do not re-flatten the ladder into the field.
Trim: 37x fwd ≈ $228
⚠️ Read the basis before changing this number. The ceiling judgment is $230 ≈ 42x the FY26
non-GAAP guide-mid of $5.47. It is written as 37x because site.py derives EPS as
price / forwardPE, and Yahoo's forwardPE (32.38) is on the FY2027 EPS of $6.14. 37 ×
($199.63/32.38) = $228 ✓. Writing "42x fwd" would render $259; the retracted baseline value
of "46x fwd" renders $284 against an intended $252. Per
pitfall-multiple-trim-inherits-the-broken-vendor-field.
A portfolio-specific passage was removed from the public build.
Break triggers
Unchanged in substance; one is now nearly live and the basis is pinned:
- Operating margin below 22% for two consecutive quarters — ⚠️ GAAP as-reported basis. Q1 FY26 23.01%, Q2 FY26 22.24%. One weak quarter from firing. (Ex-restructuring the Q2 figure is 25.02%; if Q3's Q&A establishes that management's own framing is ex-items, revisit this basis.)
- Cloud subscription growth below 18% for two consecutive quarters
- RPO growth below 15% yoy, or any sequential RPO decline
- Win-rate commentary falling below ~65%
- Active Agents penetration stalling below ~20% of installed base by Q4 FY26, or the 100% conversion rate breaking materially
Upgrade conditions — what would move this to ACCUMULATE
- Q3 op margin back above 23.5% GAAP as-reported with S&M growth decelerating — proves N1's "growth spend, not structural" reading.
- A fourth consecutive bookings record plus RPO holding ≥20% yoy growth.
- Price into $180–195 with none of the five triggers fired — T1 fills on its own terms.
- FY27 guidance (at Q4, Jan/Feb) above the $6.14 consensus, which would restore the inflection claim that N2 currently undercuts.
Any two of 1, 2 and 4 together would support restoring [8.5].
What to watch, in order
- Q3 FY26 print — Oct 27, after close (date confirmed, not estimated). Consensus Q3 EPS ~$1.46. It tests eleven rows at once: the four conviction rows, the margin trigger, RPO, cloud mix and the share count. Everything here defers to it.
- The operating margin line, on a GAAP as-reported basis, against 22%.
- Billings (N9). Ask the call to reconcile ~6.5% four-quarter-average billings growth against +23% RPO and +9.3% revenue. If billings stay mid-single-digit while RPO decelerates at all, the benign long-duration explanation fails and this becomes the leading indicator, ahead of every trigger in the list.
- Whether $180–195 gets tested before the print. If it does, T1 is live on a name with an unbroken thesis — the single best outcome available from here.
- The 10/28 FOMC. A second hike pressures the multiple assumption again; a hold would argue the 33–42x band was cut one notch too hard.
- Any Editions disclosure with a number attached — pricing, tier mix, mid-market logo count, or a gross-margin comment. Right now N7 is a real strategic move with zero quantification on either side of the ledger.
- Any insider buying at sub-$200. There has been none in twelve months; a purchase here would be the first genuinely bullish Form 4 signal in this entire analysis chain.
6. What This Pass Did NOT Test
- The five UNTESTED rows — #6 (win rate), #11 (cloud mix), #14 (buyback pace), #22 (RPO), #24 (Active Agents penetration/conversion). Three are now deferred for a third consecutive pass, and they are precisely the four claims that earned the July conviction raise. They carry forward unverified, not confirmed-intact.
- No primary-filing pull. As with the last two passes, this relied on Yahoo Finance MCP and
fin.py. The baseline explicitly recommended that "the next differential pass (post-Q3) should pull the 10-Q directly for the margin and share-count rows given how much weight they carry" — that recommendation stands and is now more urgent, because the margin row moved to DRIFTED this pass on vendor-sourced quarterly data. The Q3 pass must read the 10-Q. - No earnings-call transcript read. The #12 basis question (GAAP as-reported vs ex-restructuring) would likely be settled by management's own framing on the Q2 call. It was not obtained. This is the highest-value cheap test available before October 27.
- N2 rests on one vendor consensus field of a kind this repo has caught mislabelling fiscal years. It moved no Structural or Trend row and is used only as a direction. A roic.ai or IR cross-read would corroborate it.
- N3 is single-source (one downgrade). It moved no claim.
- Editions (N7) is entirely unquantified, and it is the one Structural-adjacent event of the window. The launch itself is primary-sourced, but no pricing, tier mix, TAM figure or margin commentary exists. Both the bull read (wider on-ramp for the AI-agent monetisation base) and the bear read (downmarket ASP and gross-margin dilution) are unsized, so they were recorded and deliberately not used to move #1 or #4. This is the largest genuinely open question the pass leaves behind.
- N9 (billings ~6.5%) is an unverified screen output and is the single highest-value thing to check against the 10-Q. If it is right, it is a more sensitive demand gauge than four of the five break triggers.
- The EDGAR full-text filing sweep failed (HTTP 503). Consequently "no 8-K, no pre-announcement" and "no Form 4 since Aug 11" are negative findings from aggregators only, not verified against the primary filing index. Negative evidence from a single aggregator is the weakest class of finding in this report, and #26 — which is counted as a positive force in the conviction table — rests on it. Re-verify directly at the Q3 pass before that row is leaned on again.
- A third-party reference to "governance, litigation and margin structure risks" (Simply Wall St, 9/24) was not run down. The margin piece is covered by #12; the governance and litigation references are unexamined and could touch rows this report did not open.
- Today's tape is a thin print. The 9/28 session traded ~187k shares against a ~798k three-month average, and quoted closes differ ($199.04 vs $199.63). The −3.1% is directionally corroborated by the named downgrade, but the exact level should not be treated as settled; no conclusion here depends on it.
- Competitive win/loss data — nothing landed on SAP EWM, and nothing dated inside the window landed on Blue Yonder beyond its own 2026 Gartner WMS Magic Quadrant Leader placement, which is a reminder that MANH's Gartner leadership is not exclusive. Vectors #5/#6/#7 are carried on absence of evidence, which is weaker than confirmation.
Sources: .mcp/fin.py MANH --news · Yahoo Finance MCP (get_stock_info, get_financial_statement quarterly_income_stmt + quarterly_balance_sheet, get_holder_info insider_transactions, get_recommendations upgrades_downgrades, get_historical_stock_prices) · direct yfinance field read for forwardEps/forwardPE/target fields · baseline reports analyze-from-before-2026-08-27.md, analyze-2026-07-29.md, analyze-2026-06-04.md · Knowledge/Playbook/pitfall-multiple-trim-inherits-the-broken-vendor-field.md · pitfall-vendor-forward-eps-is-the-wrong-fiscal-year.md · pitfall-stale-entry-zone-suppresses-a-name.md · pitfall-stale-fair-value-is-most-costly-on-winners.md