The signal does not move — HOLD on all three horizons, as in August — but almost everything underneath it does, and one thing that should have happened in August did not.
The price has risen 10.5% from $26.58 to $29.38, which closes part of the gap to the $35.00 offer: the discount to the deal price has narrowed from 24.1% to 16.1%. Two of August’s three gate findings are retired on the evidence rather than carried: Gate 1 (Financial Distress) is now clear — the current ratio is 1.24, EBITDA covers interest 4.2 times, and August’s trigger was a lease-inclusive debt figure read as distress on a company that charters its fleet — and Gate 2 (Earnings Event) is clear because it was always date-driven: Q2 printed on 19 August and Q3 is not due until roughly mid-November. What remains is the one that matters, and it is now dated: Gate 5 (Binary / Regulatory Event) continues from August, where it was carried under the unnumbered label “Binary / Regulatory Event” — it has not newly fired, it has acquired a deadline of 27 September.
One ladder change worth naming: short_entry_confirmed moves from true to false. August recorded a confirmed short entry; this report does not, because Gate 5 now caps every horizon and the Catalyst group fails on the point that 27 September decides the thesis rather than confirming it. The conviction ladder itself is unchanged at one group of three, Half-Size.
August left a mandatory step undone. ZIM was a Donatien Pick returning HOLD on all three horizons, which under the house rule takes it off the active pick list automatically. That was never adjudicated. It is adjudicated here: ZIM moves to Stopped — still visible on the watchlist, and re-activating on any future BUY. This was found by a deterministic pre-flight check rather than by reading, which is the honest way to say it.
ZIM is an Israeli container liner — the world’s tenth largest by capacity, and a small one, at roughly 1.5% of global slots against Hapag-Lloyd’s ~7% and MSC’s ~20%. It does not try to be everywhere. It runs a deliberately asset-light model, chartering most of its ~145 vessels rather than owning them, and concentrates on niche trades where speed and reliability earn a premium — expedited Asia–US East Coast being the best known. That model is why ZIM’s earnings swing more violently than any peer’s: chartered capacity is a fixed cost that cannot be shed when freight rates fall, and an operating leverage that pays spectacularly when they rise. In the 2021–22 boom it earned more than its entire market value; in 2023 it lost $2.7bn. Roughly 4,700 employees, headquartered in Haifa, with the State of Israel holding a golden share over the company’s strategic assets — a detail that is usually a footnote and is, this year, the whole story.
One triggered gate, one caution. Gate 5 caps all three horizons at HOLD. Gates 1 and 2 are retired from August rather than silently dropped, each with its reason above. No Do-Not-Buy trigger fires — and Trigger 2, the valuation trigger, is explicitly not fired: see Gate 3 for why a 25.6× trailing multiple on collapsed earnings is not the same thing as an expensive stock.
40 — medium, the same score as August, though not for the same reasons. Three components moved underneath an unchanged total and a reader is entitled to see them: the moat score rises 28 → 45, the industry benchmark was swapped from a unit-cost read to EV/EBITDA against listed liners (scoring 30 → 70) — both August figures come from that report’s calibration, where they sat nested inside a quality_detail object rather than at the top level, and the competitive share trajectory is revised from “losing” to “stable”. The last of those is a reversal of August’s judgement, not a new observation, and it is argued below rather than asserted. What follows describes ZIM standalone. If the deal closes none of it will matter to a holder; if it breaks, all of it will.
Trailing twelve-month revenue is $6,439m, down 24.9% year on year against $8,575m — though the latest quarter was up 8.9% on Q2 2025, which is the recovery the deal is being struck into rather than a trend. Operating income is $387m, a 6.0% operating margin. Net income is $139m, a 2.15% net margin, on a return on equity of 3.58% and a return on assets of 2.12%. Those are not good numbers, and they are not meant to be read as a bad year in an otherwise steady business — they are a mid-point in a range that has no middle.
The quarterly earnings per share of the last six quarters, in order: $2.45, $0.19, $1.02, $0.32, −$0.71, $0.53. Operating income over the same run: $454m, $137m, $254m, $3m, −$14m, $144m. A business that can print $454m and −$14m of operating income within four quarters is not a business you can value on a multiple of one year’s earnings, which is the single most important fact in §4.
Container shipping is a commodity with a cartel-shaped cost structure and no pricing power at the individual-carrier level. ZIM carries roughly 1.5% of global slot capacity. Its direct competitors are its acquirer and everyone larger: MSC at around 20%, Maersk and CMA CGM in the low teens, COSCO and Hapag-Lloyd at roughly 7% each. Scale in this industry is not a vanity metric — it buys network density, slot-cost advantage and negotiating power with terminals, and ZIM has the least of all three among listed carriers.
Share trajectory: stable but structurally disadvantaged — revised from August’s “losing”. ZIM has held its niche rather than lost it: the expedited Asia–US East Coast service still commands a premium, and the transpacific strength described in §5 is in the lane it is built around. But it has not gained, and the industry has spent a decade consolidating away from operators of its size. That is precisely why the company is being bought rather than buying. The revision is a judgement call and the honest caveat is that a company mid-acquisition does not fight for share the way an independent one does — “stable” here means “not visibly losing”, which is a lower bar.
Chartering rather than owning is a genuine choice with a genuine trade. It keeps capital employed low and lets the fleet flex faster than an owner’s can. It also converts vessel capacity into a fixed obligation priced at whatever the charter market demanded on the day the contract was signed — which is why the 2021–22 boom produced earnings larger than the company’s market value and the 2023 bust produced a $2.7bn loss. The balance sheet reflects the same choice: the gap between EBITDA of $1,939m and operating income of $387m is $1,552m, of which $1,287m is depreciation and amortisation — the bulk of that right-of-use charges on the charters rather than wear on owned steel. The remaining $265m sits below EBIT in non-operating items.
Moat score 45, up from 28. There is no moat in the durable sense — no switching cost, no network effect the larger carriers do not have more of, no cost advantage. What ZIM has is a reputation for reliability on specific lanes and an Israeli-flag franchise that is strategically protected by its own government. The second of those is worth something, and it is the reason this acquisition is difficult.
Comfortable, once the leases are read correctly. Current ratio 1.24, quick ratio 1.15, cash $1,638m, net debt $3,675m at 1.9× EBITDA, interest covered 4.2×. Free cash flow of $991m is a 28.0% yield on the market capitalisation and 13.7% on enterprise value — the sort of figure that looks like a screaming buy signal and, on a cyclical at an uncertain point in its cycle with a pending takeover, means considerably less than it appears to.
On the dividend. The data feed reports a 7.88% yield. Ignore it. The last declaration was $0.88 on 9 March 2026, paid on 26 March, and there has been nothing since — two quarters missed, which is what one expects while an all-cash merger is pending. Measured properly by ex-dividend date the trailing twelve months paid $1.19 — $0.31 ex 1 December 2025 and $0.88 ex 20 March 2026 — a 4.05% yield, and the forward expectation is not that. A quoted yield on a company being acquired for cash is an artefact of the calculation window, not income anyone should plan around.
68 — Attractive, on a warranted ratio of 0.84. The basis is price-to-book, not earnings, and that choice needs defending before any of the numbers mean anything.
The framework’s default anchor is a warranted forward earnings multiple. It should not be struck here, and the reason is the shape of the earnings rather than the availability of a number. Four sources give four figures for the same year:
get_analyst_estimates has FY2026 at $4.31 on two analysts, a range of $3.88–$4.90A correction, because an earlier draft of this report said something false. It stated that get_analyst_estimates returned no earnings values at all. It does return them — the figures above. That draft was wrong because my pull filtered on the wrong field names and I read the empty result as an empty endpoint rather than as a mistake in the query. The conclusion below did not depend on it and does not change; the stated reason did, and it was the weaker argument in any case.
The real objection is not the spread. It is that a single forward year cannot value this business. Take the framework’s own $4.31: that is a forward P/E of 6.8×, which looks like a gift. The same source has FY2027 at −$2.34 and FY2028 at −$2.95. A six-times multiple on what consensus treats as the last good year of a cycle, with losses forecast on both sides of it, is the oldest trap in cyclical investing — and §3 has already made the case: a company that printed $454m and −$14m of quarterly operating income inside four quarters cannot be valued on one year’s earnings, whichever year you pick.
So the anchor moves to the asset lens. This system already does exactly this for mining companies, where earnings are the wrong lens and the anchor is price-to-net-asset-value. A chartered-fleet container liner across a cycle this violent is the same problem wearing different clothes. This is applying an existing rule to an analogous case, not inventing one for the occasion.
Book value per share is $31.73. At $29.38 the shares trade at 0.926× book. The warranted multiple is the contracted price over the same book: $35.00 ÷ $31.73 = 1.103×. The ratio is 0.926 ÷ 1.103 = 0.84, which falls in the Attractive band (0.80–1.00, scoring 65–77). Fair value is therefore $35.00 — not a discounted-cash-flow estimate but the price a counterparty has contracted to pay.
State the judgement plainly: I am treating the deal price as warranted value. That is defensible for a binding all-cash agreement already approved by shareholders, and it is what a holder actually receives if the deal completes. It is also a choice, and a reader should see the alternatives. On a standalone discounted-cash-flow or a mid-cycle return-on-equity basis the warranted multiple would be lower than 1.10 — trailing return on equity of 3.58% against any sensible cost of equity argues for well under book — and on that reading the shares are not cheap at all. The completion probability that reconciles those two views is not handled in the anchor; it is handled in the scenarios at §11 and in Gate 5. The anchor says what you get if it closes. It does not say whether it closes.
Non-operating items are a net drag of −$196.2m on the trailing year — 141.6% of net income — so the reported trailing P/E of 25.6× overstates the earnings multiple rather than flattering it. This is the reverse of the pattern this framework usually has to correct. Stripping the drag: operating income of $387.4m taxed at the effective 27.5% gives clean earnings of $2.33 per share and a clean trailing P/E of 12.6×.
Treat that as a ceiling, not a fair value. Most of the drag is interest, and interest is a genuine, contractual cost for an operator that finances a chartered fleet — $457m of it on the trailing year. The 12.6× describes what ZIM would earn if it had no financing cost, which is not a company that exists. It is shown because the framework requires the clean figure whenever non-operating items exceed 15%, and because the direction matters: this business is cheaper on operations than its headline multiple suggests, and still not obviously cheap.
The reverse-DCF implied-growth read is not published, because it cannot be computed. A reverse DCF solves for the growth rate the current price implies, and it needs a credible forward earnings base to start from. There is no such base here — that is the whole of the problem set out above. Producing an implied-growth number from any one of $4.31, $3.15, $0.29 or $9.18 would be arithmetic dressed as analysis — and solving for growth from a base that the same consensus has turning negative the following year would produce a figure with no meaning at all.
The anchor carries 40% of the valuation score. The rest is a blend, and here it is honest to say that most of it is compromised by the same problem.
The framework’s grades tool returns 0 strong-buy, 0 buy, 3 hold, 3 sell — a Hold consensus on six analysts, 0% bullish. Yahoo’s target feed, on four analysts, gives a mean of $25.80 against a high of $38.00 and a low of $16.50. That mean is below the current price of $29.38 and well below the $35.00 offer.
Two readings, and I lean to the first. Either the targets are standalone valuations that have not been marked to a live cash offer — a mean of $25.80 sits almost exactly between a $35 completion and an $18 break, which is suspiciously like a probability-weighted number nobody labelled as one — or the analysts genuinely think the deal fails. On four analysts I would not build a thesis on either. What I will not do is quietly drop the lens because it disagrees: 0% bullish is a real datapoint and it is why the valuation confidence here is 45%, the lowest in this report.
A source note, because it bit a report last month: the FMP price-target endpoint returns high, low, consensus and median all equal to $20.00, which is a degenerate response rather than a tight consensus. The Yahoo figures above are used instead.
48 — neutral, up from 38 in August, and the improvement is real but it mostly cannot reach the shareholder.
ZIM’s earnings are a leveraged bet on spot container rates, and the current reading is genuinely firm at the front and softening behind it. Drewry’s World Container Index was $4,465 per 40ft on 3 September, flat week on week. Underneath that composite the two legs disagree: Shanghai–Los Angeles rose 5% to $7,185 and Shanghai–New York rose 3% to $9,587, with the transpacific at a 22-month high, while Asia–Europe fell. Drewry expected a modest decline the following week.
The transpacific strength matters more than usual for this name, because expedited Asia–US East Coast is exactly the lane ZIM is built around. But three things cut the other way, and they belong in the same paragraph rather than a footnote: carriers are ramping transits back through Suez, which returns effective capacity to the market and is the single most reliable way to end a rate spike; Asia–Europe is already softening; and Brent above $100 raises bunker costs directly, with carriers having announced emergency fuel surcharges from August.
Which dominates? On the evidence I have, I cannot tell, and I would rather say so than pick. Route disruption has historically raised freight rates by far more than fuel costs subtract — the 2024 Red Sea diversions produced ZIM’s best year outside the 2021–22 boom — $2.15bn of net income, which the company itself called its best results ever excluding the extraordinary COVID period, against $4.65bn in 2021 — but that episode removed capacity for a year, and returning Suez capacity is the opposite trade. A flat composite index with one leg at a 22-month high and the other falling is exactly what an unresolved tug-of-war looks like. The score moves to neutral, not to positive.
The macro report of 9 September rates the Iran/Hormuz energy shock its most dominant driver, at CRITICAL (5, raised from 4): an active tanker war, Iran signalling a new Hormuz exclusion zone on 7 September, and Brent through $100. For most of the watchlist that is a cost shock. For a container liner it is ambiguous in the way described above, and for this container liner it is close to irrelevant — which brings us to the point of this section.
A deal-pinned equity does not transmit its drivers. If Hapag-Lloyd pays $35.00 in cash, a holder receives $35.00 whether freight rates double or halve in the interim. Improving fundamentals do not raise the consideration; they raise the price of the thing being bought, which the buyer has already fixed. The freight tape only reaches the shareholder through one channel: it sets what ZIM is worth if the deal breaks. That is a genuine and underappreciated asymmetry — the break scenario is better today than it would have been in a soft freight market — and it is why the bear case at §11 is set at $18.00 rather than at the $15.50 the shares traded at before any of this began. It is not a reason to own the shares for their drivers.
45 — Neutral. The 9 September macro report rates Industrials at Neutral on all three horizons, and ZIM does not carry its own watchlist forecast in that report, so the sector read is the starting point rather than a name-specific signal.
The dominant regime is energy-shock stagflation, intensified — a Hormuz supply shock with Brent above $100 landing on core PCE stuck at 3.34%, with the Fed debate now hike-versus-hold rather than how-fast-to-cut. For a global trade name that cuts both ways with unusual symmetry: stagflation compresses the goods volumes that fill containers, while supply disruption is what lifts the rates those containers earn. Neither dominates in the data at this moment.
What genuinely applies to ZIM is narrower. Israeli sovereign politics is the macro exposure that matters, and it is not an interest-rate story at all. The transaction requires the State of Israel to accept foreign ownership of a strategic national asset during an active regional conflict — and the same conflict that is lifting Brent is the reason the defence establishment cares about who controls sixteen Israel-flagged vessels. The macro regime and the deal risk are not independent variables here; they share a cause.
A rate cut or a hike will move this share price by very little compared with a single announcement from the Israeli finance ministry. That is what a Neutral score with a 55% confidence is trying to express: the sector read is defensible, and it is close to beside the point.
Source: MacroDriver-state-20260909 — Industrials (XLI) N/N/N; ZIM carries no watchlist forecast · Macro report 9 September 2026
47 — neutral. The technicals look better than they are, because on a deal-pinned equity they are not measuring what they usually measure.
The daily frame is in a strong uptrend: price above its 20-day ($27.89), 50-day ($26.15) and 200-day ($24.77) averages, RSI at 61.1, MACD positive at 0.75 against a 0.64 signal, and a resistance breakout. Weekly and monthly are both uptrends. Ordinarily that is momentum. Here it is mostly convergence: the share price is being pulled toward $35.00 as the market’s confidence in completion rises, and pulled back when it falls. The 8 September session is the tell — the shares jumped 5.2% from $28.58 to $30.08 on 4.42m shares against a 1.50m average, roughly 2.9× normal volume, on news that Hapag-Lloyd was working with Israel on structural changes. That is not a trend. That is a probability being repriced.
The two intraday frames disagree with the daily: hourly is weakening with RSI at 39.3 and MACD turned negative, and the 15-minute frame is in a downtrend with RSI 36.7 and a support breakdown. The composite confluence still reads bullish on the weight of the higher frames. I would not put much on either intraday read at 0.53 and 0.74 volume ratios.
Risk-reward score 30, unfavourable. The nearest daily support is $23.29, which is $6.09 away, or 6.3× the daily ATR of $0.97. The framework’s comfort line is 2.5 ATR and this is two and a half times beyond it. Upside to the offer is +19.1%; downside to my break estimate of $18.00 is −38.7%. As a raw ratio that is about 0.5 to 1, which is a bad trade — and it is only redeemed by probability, which is precisely the thing a risk-reward score is not allowed to assume.
On stops, honestly: a stop-loss is close to useless on this name. The risk is not that the price drifts down through a level; it is that a ministry announcement gaps it 30% overnight. A stop at $23.29 would not fill at $23.29. Publishing one would imply a protection that does not exist, so §12 does not publish one as a price to rely on.
On the framework’s scored windows ZIM is at the very top of the relative-strength table, and the figure is almost meaningless. Over one month it is +16.5% against the S&P 500 at −1.4% and Industrials (XLI) at −6.9%; over three months +16.0% against +5.1% and +1.3%. That is +10.9 points versus the market and +14.7 versus the sector on the three-month window, outperforming both benchmarks on both timeframes. Over twelve months it is +122.6%, against +16.9% and +13.5%.
Attributing any of that to relative strength would be a category error. The shares rose because a takeover was agreed at a 126% premium to the unaffected price and because the market’s odds on completion improved. It is relative strength in arithmetic and deal news in substance. It does not lift the timing score of 47 — relative strength is not a term in the timing formula, which is built from the multi-timeframe trend, risk-reward, macro, sentiment and catalysts — and in this particular case I would not read it as information at all. Computed this run from raw closes; price returns, not total returns.
Sentiment 30, and it is poor. The grades consensus is Hold on six analysts with zero bullish — three hold, three sell — and the four-analyst mean target of $25.80 sits below the market price. Whatever one makes of the coverage depth, nobody is recommending this.
The catalyst calendar is crowded and it is dated. Three high-impact events fall inside seventeen days: US CPI on 11 September, the FOMC on 16 September carrying a 48–65% implied hike, and — the one that actually decides this name — the 27 September deadline for Hapag-Lloyd and FIMI to submit their revised proposal to the Israeli authorities. That is the framework’s clustered case, so the clustering score is 40, in the “noisy” band for two or three events inside a fortnight, rather than the calm end. To be clear what that does: clustering modulates position sizing, and below 30 it would dock timing confidence. It is not the catalyst term in the timing formula, so it does not change the 47 above.
The catalyst sub-signal itself scores 50 — genuinely neutral, because 27 September is a binary that can resolve in either direction, and a dated binary is neither an opportunity nor a warning until it resolves.
| Date | Event | Impact | Forecast | Previous | Relevant? | Why |
|---|---|---|---|---|---|---|
| 27 Sep 2026 | Revised Hapag-Lloyd / FIMI proposal due to the Israeli authorities | CRITICAL | Submitted, not rejected outright | 30-day extension granted | ✅ DECISIVE | The event that determines whether a holder receives $35.00 in cash or owns a sub-scale container liner. Nothing else on this calendar is close. |
| 16 Sep 2026 | FOMC decision | HIGH | 48–65% implied hike | 3.50–3.75% | Low | Moves the discount rate and the dollar. For a cash-takeover target it is close to noise — the consideration is contractual, not discounted. |
| 11 Sep 2026 | US CPI (August) | HIGH | +0.4% MoM, 3.4% YoY | 3.3% YoY | Low | Feeds the FOMC above. Relevant to the tape ZIM trades in, not to ZIM. |
| ~mid-Nov 2026 | ZIM Q3 2026 results | MEDIUM | n/a | Q2: rev $1,780.7m, net $63.5m | Medium | Outside the 14-day earnings gate. Matters chiefly as evidence for the break case. Date approximate from the reporting pattern — the earnings-calendar feed returned empty. |
| Late 2026 | Expected completion, if approved | CRITICAL | n/a | n/a | High | The company's own guidance, contingent on 27 September and the remaining regulatory clearances. |
| Date | Event | Actual | Forecast | Surprise | Impact |
|---|---|---|---|---|---|
| 9 Sep 2026 | All eight reviewing authorities filed positions with the Government Companies Authority | Six of eight opposed | n/a | Defence, Economy, Agriculture, Transportation and the Shipping and Ports Authority against; Finance had not formally filed; the National Security Council alone inclined to support | The most decision-relevant input available on the day this report was priced — and it is why the base weight moved DOWN from the draft's 65% to 52% |
| 8 Sep 2026 | Hapag-Lloyd reported working with Israel on structural changes | +5.2% ($28.58 → $30.08) | n/a | 4.42m shares vs a 1.50m average — 2.9× normal | Deal odds repriced upward |
| 19 Aug 2026 | Q2 2026 results + 6-K | Revenue $1,780.7m, net income $63.5m | n/a | Recovery from Q1's $86.0m loss | Filed as a first-half loss with the buyout pending |
| Early Sep 2026 | Fresh Israeli pushback reported against the $4.2bn structure | Shares fell | n/a | n/a | Deal odds repriced downward |
The stock is trading the deal, not the results. Three of the four moves above are political; the one set of actual results barely registered.
| Timeframe | Trend | Direction | RSI | MACD | Key S/R | Breakout | Vol |
|---|---|---|---|---|---|---|---|
| Monthly | Uptrend | ▲ | 60.7 | 1.44 / 0.09 | S 15.90 · R 30.15 | Resistance breakout | 0.13× |
| Weekly | Uptrend | ▲ | 64.2 | 1.25 / 1.05 | S 23.29 · R 29.97 | Resistance breakout | 0.97× |
| Daily | Strong uptrend | ▲ | 61.1 | 0.75 / 0.64 | S 23.29 · R 29.56 | Resistance breakout | 0.92× |
| Hourly | Weakening | ▼ | 39.3 | −0.06 / 0.01 | — | Resistance breakout | 0.53× |
| 15-minute | Downtrend | ▼ | 36.7 | −0.07 / −0.02 | — | Support breakdown | 0.74× |
| Confluence: Bullish · MTF Score 62 | |||||||
Three of five frames are in uptrends and they are the three that carry weight. But the composite deserves a larger caveat than usual: these frames are measuring a price anchored to a contractual $35.00 that moves on political news, so the technical structure describes sentiment about a deal rather than supply and demand for a business. The monthly row rests on a 0.13× volume ratio on a partial month and should carry almost no weight. The number to take from this section is not a trend label — it is the 6.3-ATR distance to support at §7.
ZIM daily closes, 10 March to 9 September 2026, with the 50-day average. The series ends at the 9 September close of $29.38 — the last completed session — because the 10 September bar was still open when this report was built. The flat ceiling below $30.40 is the market’s standing discount to a $35.00 cash offer, not resistance in the usual sense.
Sourced: $37.50 is not a round number chosen for symmetry — it is the price an Israeli investor group led by Haim Sakal offered in May 2026, a $4.5bn all-cash proposal pitched $300m above Hapag-Lloyd’s and carrying a $250m employee bonus. Be clear that this is not a live auction. The Sakal approach arrived after shareholders had already approved the Hapag-Lloyd agreement, ZIM’s board reconfirmed that agreement as binding, and it is now four months old. It survives here as evidence that a credible domestic buyer valued the company above $35 with a structure Israel could accept — which is exactly the objection Hapag-Lloyd is now trying to engineer around. Trigger: the Israeli authorities reject the 27 September revision and a domestic bidder returns, or Hapag-Lloyd raises to clear the political price. Why only 10%, when the Israeli objection is precisely about foreign control? That argument cuts toward Sakal, not away from it — a domestic buyer is what six of the eight reviewing bodies say they want, so a break ought to make a domestic bid more likely, not less. It stays at 10% because the obstacle is not the argument, it is the agreement: the board has already chosen once, in the knowledge of a higher offer, and reconfirmed the Hapag-Lloyd contract as binding. This weight is held at August’s level rather than cut — the reduction in completion odds falls entirely on the base case, where the evidence actually moved.
The deal closes at the agreed price. This is a contractual number, not an estimate: $35.00 per share in cash, roughly $4.2bn of equity, unanimously approved by the board and approved by shareholders in late April 2026. The path runs through the 27 September revised proposal — FIMI owning the carved-out New ZIM Israel, the foreign-shareholder notification threshold cut from 24% to 10%, a pledge not to list outside Israel, all sixteen Israel-flagged vessels at the state’s disposal, and expanded Asian route access — then the remaining regulatory clearances, with completion guided to late 2026. Trigger: Israeli approval of the revised structure. Return if it lands: +19.1% from $29.38, plus whatever the timing is worth, and no more — a cash offer has no upside beyond its own price.
The deal breaks and ZIM is a sub-scale container liner again. The shares traded at $15.50 on 8 August 2025 before any of this began, and that is the honest starting point for a break. I set the bear at $18.00 rather than $15.50 for two reasons, both of which should be stated so a reader can disagree. First, the freight tape is materially better: Drewry’s World Container Index was $2,424 per 40ft on 7 August 2025 and is $4,465 on 3 September 2026 — +84% — with the transpacific at a 22-month high. Second, the balance sheet holds $1.64bn of cash against $991m of trailing free cash flow. $18.00 is 0.57× book. Trigger: the Israeli authorities reject the revised structure on 27 September, or the golden-share negotiation fails outright. The mechanism matters more than the number: this would not be a drift, it would be a gap on an announcement, which is why §12 does not pretend a stop-loss protects against it. Against that, consensus already has ZIM at a loss in 2027, so the break case is not a soft landing.
Probability-weighted 12-month value: $28.79 (0.10 × $37.50 + 0.52 × $35.00 + 0.38 × $18.00), or −2.0% from $29.38.
These weights moved, and the honest framing is that one of them was an anomaly. An earlier draft of this report carried 65% on completion and produced $31.00, which put it more optimistic than the market. That draft was struck without the most decision-relevant fact available on the day: on 9 September, all eight Israeli reviewing authorities filed their positions and six opposed the structure. Note what the bear weight does here — the August report carried bear at 40%. So 38% is not a lurch; it is a return to where this report already was, and the draft’s 25% was the outlier. The base case absorbs the whole correction, to 52%.
This is not a claim to an edge, and it should not be read as one. The 9 September filings were public before the 9 September close that sets this report’s $29.38 basis, so the market’s own odds already contain them. Moving toward the market on information the market has is not insight; it is the correction of a weight that had been struck as though the filings had not happened.
The check that matters. If the only two real outcomes are $35.00 and $18.00, the market’s price of $29.38 implies a completion probability of 66.9% — ($29.38 − $18.00) ÷ ($35.00 − $18.00). This report’s $28.79 implies 63.5% on the same two poles. Three points apart is agreement within noise, which is the honest place for a framework with no edge on Israeli cabinet politics to sit. Anyone who reads those politics differently can substitute their own probability and get their own fair value in one line.
Sit with what a −2.0% expected return means. It says the spread is roughly fair compensation for the risk — that you are not being paid to take the political bet, merely priced for it.
And note what kind of distribution this is. It is not wide because the business is uncertain — it is bimodal, because a binary political decision sits seventeen days away. There is no meaningful outcome between $18 and $35, which is why the weighted average describes no world that will actually happen, and why the downside arrives as a gap rather than a drift.
Forecast: One of three entry groups is met — Technical, on price structure alone. The ladder is not permission to buy: Gate 5 withholds that. It says what size would be warranted if the gate cleared, and one group of three says Half-Size. The Fundamental group fails on a point worth being explicit about: the price IS below fair value, but that fair value is a merger consideration rather than anything the business supports, and §4 says so. The Catalyst group fails because 27 September is a risk event that can resolve either way, not a confirmation.
Forecast: By the next update on 28 September the 27 September deadline will have passed, and I expect the revised proposal to have been submitted rather than the deal abandoned — Hapag-Lloyd has already conceded on the golden share, the notification threshold, the listing venue and the sixteen vessels, which is not the behaviour of a buyer preparing to walk. Submission is not approval. My expectation is that ZIM is still HOLD at the next update, with the discount to $35.00 narrower than 16.1% and the Israeli decision still outstanding. I would be wrong if the authorities reject outright, in which case the shares should trade toward the high teens within days rather than weeks.
You would be risking about 38.7% to capture 19.1%, on a political decision seventeen days away.
What you are risking: $11.38 per share to the $18.00 break case, to capture $5.62 to the $35.00 offer. That is roughly 0.5 to 1, and it is only redeemed by probability — you would need better than 67% confidence in completion simply to break even against the market's own implied odds. There is no stop that protects you, because the downside arrives as an overnight gap on a ministry announcement rather than a drift through a level.
What you would be buying: not a shipping company. A binding cash contract at $35.00, and the risk that the State of Israel refuses to let it complete. The freight tape, the 3.7× EV/EBITDA and the 28% free-cash-flow yield are all real and all irrelevant unless the deal breaks — at which point they become the whole case, alongside a consensus that has 2027 at a loss.
You would be giving up 19.1% of contracted upside to avoid 38.7% of political downside.
Selling at $29.38 banks a 10.5% gain since the August report and 122.6% over twelve months, and it removes an exposure that no amount of analysis of ZIM's business can resolve — the decision sits with eight Israeli reviewing authorities of which six filed against the original structure on 9 September, and it is not a question this framework has any edge on.
The honest counter: Hapag-Lloyd has already conceded on the golden share, the notification threshold, the listing venue and all sixteen vessels. That is a buyer working hard to close, not one preparing to walk, and selling now forfeits the 16.1% for a risk that may well be priced.
Half-Size, and only if you are doing something other than investing. A position in ZIM today is a merger-arbitrage trade: you are underwriting a political decision in Israel for a 19.1% gross return, with 38.7% of downside if it goes the other way and no way to hedge the gap. That can be a perfectly rational thing to do. It is not what this framework scores, and it is not what a Donatien Pick is for.
The clustering score of 40 argues for reducing size by 30–50% against a normal position on event density alone — three high-impact events inside seventeen days. Combined with a Half-Size ladder that puts any entry at a quarter to a third of a full position, and it should be sized as money you can lose outright rather than money you are compounding.
For an existing holder — and this is the more relevant case, since ZIM has been on this watchlist since before the deal — the question is whether to keep underwriting a binary you did not originally sign up for. The answer this report gives is HOLD, not because the arb is unattractive but because the framework will not call a political coin-toss an investment. If the 27 September step passes and the discount narrows, that is the moment to take the return rather than wait for the last two dollars.
{
"version": "v6",
"brand": "",
"ticker": "ZIM",
"company": "ZIM Integrated Shipping Services Ltd.",
"currency": "USD",
"exchange": "NYSE",
"exchange_ticker": "NYSE:ZIM",
"api_ticker": "ZIM",
"isin": "IL0065100930",
"finder_ticker": "ZIM",
"finder_exchange": "US NYSE",
"date": "2026-09-10",
"analysis_status": "stopped",
"macro_report_date": "2026-09-09",
"primary_signal": "HOLD",
"signal_short": "HOLD",
"signal_medium": "HOLD",
"signal_long": "HOLD",
"base_signal_short": "UNDEFINED",
"base_signal_medium": "UNDEFINED",
"base_signal_long": "UNDEFINED",
"composite_short": 51,
"composite_medium": 52,
"composite_long": 49,
"quality_score": 40,
"valuation_score": 68,
"timing_score": 47,
"driver_score": 48,
"driver_score_short": 46,
"driver_score_medium": 48,
"driver_score_long": 50,
"driver_name": "Container freight rates",
"driver_label": "Neutral",
"economic_alignment_conviction": 45,
"economic_alignment_stance": "Neutral",
"economic_alignment_pressure": "neutral",
"economic_alignment_source": "MacroDriver-state-20260909 (Industrials XLI N/N/N; ZIM carries no watchlist forecast)",
"overall_confidence": 45,
"price_at_rating": 29.38,
"price_basis": "9 September 2026 close, the last completed session; the NYSE was open when this report was built",
"lifecycle_stage": "industrials marine shipping mature",
"moat_score": 45,
"competitive_share_trajectory": "stable",
"competitive_threat_level": "elevated",
"industry_benchmark_name": "EV/EBITDA (listed container liners)",
"industry_benchmark_value": 3.72,
"industry_benchmark_score": 70,
"eps_trailing": 1.15,
"trailing_pe": 25.55,
"price_to_book": 0.926,
"book_value_per_share": 31.73,
"ev_ebitda": 3.72,
"fcf_yield_on_mktcap": 28.0,
"fcf_yield_on_ev": 13.7,
"fcf_yield": 28.0,
"nonop_pct_of_net_income": -141.6,
"earnings_quality_note": "Net income is DEPRESSED by non-operating items, not inflated - the opposite of the mega-cap pattern this framework usually corrects. TTM non-operating is a NET DRAG of -196.2m against operating income of 387.4m, so the trailing P/E of 25.55x OVERSTATES the earnings multiple rather than flattering it. No clean_pe is published: stripping the drag would mean stripping interest, and interest is a genuine cost for a chartered-fleet operator. EV/EBITDA of 3.72x is the meaningful figure.",
"val_multiple_basis": "P/B (asset lens - a single forward year cannot value this cycle)",
"actual_multiple": 0.926,
"warranted_multiple": 1.103,
"warranted_ratio": 0.84,
"val_band": "attractive",
"val_anchor_note": "JUDGEMENT, stated as one in section 4. Warranted P/B = the CONTRACTED price 35.00 / book 31.73 = 1.103. Defensible for a binding all-cash agreement already approved by shareholders; it says what a holder receives IF the deal closes and says nothing about whether it closes, which is handled in Gate 5 and the scenarios. On a standalone DCF or mid-cycle ROE basis the warranted multiple would be LOWER than 1.10 and the shares would not be cheap. WHY THE FORWARD-EARNINGS ANCHOR WAS NOT USED: not because no number exists - an earlier draft said get_analyst_estimates returned no EPS and that was FALSE, caused by a pull filtered on the wrong field names. It returns FY2026 4.31 (n=2), FY2027 -2.34, FY2028 -2.95. The real objection is that a single forward year cannot value this business: 4.31 is a 6.8x forward multiple on what consensus treats as the last good year of a cycle, with losses forecast either side. Four sources give 4.31 / 3.15 / 0.29 / 9.18 for the SAME year. The anchor moves to the asset lens on the established miners precedent.",
"dnb_note": "Trigger 2 (valuation) explicitly does NOT fire. The trailing 25.55x is a TROUGH artefact - earnings collapsed while the price is pinned by a cash contract - not valuation excess. The P/B anchor reads 0.84, Attractive. Trigger 3 (persistent negative earnings revisions) is CLEARED on framework data: get_analyst_estimates returns FY2026 4.31, FY2027 -2.34, FY2028 -2.95, FY2029 -0.21. That is a declining estimate CURVE across a cycle, not a REVISIONS trend, and it is carried as the Gate 6 caution instead. Trigger 4 (insider selling) cleared on a 10-article news scan with no disposal story.",
"completion_probability_note": "Bull 10 / base 52 / bear 38 gives a probability-weighted 28.79, which implies 63.5% completion at the 35.00 pole against the market's 66.9% at 29.38 - about 3.4 points apart, which is agreement within noise. History, because the weights were revised twice: the round-1 draft carried 65/10/25 for 31.00 and was more OPTIMISTIC than the market, having been struck without the 9 September filings, when all EIGHT Israeli reviewing authorities filed and SIX opposed. Round 1 cut it to 55/7/38 for 28.71. Round 2 restored bull to 10% - the Sakal structure is DOMESTIC ownership, which is what the objecting bodies want, so a break should make a domestic bid MORE likely, not less - and put the whole reduction on the base at 52%. The 38% bear is a RETURN to August's own 40%, not a new alarm. This is not a claim to an edge: the 9 September filings were public before the close that prices this report.",
"israeli_review_state": "9 September 2026: all eight reviewing authorities filed with the Government Companies Authority; six opposed. AGAINST - Defence Ministry (Minister Israel Katz publicly opposed), Economy, Agriculture, Transportation, and the Shipping and Ports Authority (second review recommending rejection). NOT FORMALLY FILED - Finance Ministry (Accountant General's division reported against, Budget Division still considering). INCLINED TO SUPPORT - National Security Council, with reservations. ZIM's workers' union opposes.",
"fair_value_est": 35.0,
"target_price": 35.0,
"scenario_bull_target": 37.5,
"scenario_base_target": 35.0,
"scenario_bear_target": 18.0,
"scenario_probabilities": {
"bull": 10,
"base": 52,
"bear": 38
},
"probability_weighted_target": 28.79,
"deal_price": 35.0,
"deal_discount_pct": 16.1,
"deal_return_pct": 19.1,
"spread_convention": "HEADLINE = 16.1% BELOW the offer ((35.00-29.38)/35.00). The +19.1% is the RETURN if it closes (35.00/29.38-1). Same gap, two conventions - both are labelled wherever they appear.",
"market_implied_completion_pct": 66.9,
"mtf_confluence": 62,
"risk_reward_score": 30,
"relative_strength_vs_spy": 10.9,
"relative_strength_vs_sector": 14.7,
"relative_strength_note": "COMPUTED this run from raw Yahoo closes. Fields carry the 3-MONTH spread, the framework's scored window (SKILL L809). 1-month: ZIM +16.5 vs SPY -1.4 vs XLI -6.9. 3-month: +16.0 vs +5.1 vs +1.3, i.e. +10.9 and +14.7. 12-month: +122.6 vs +16.9 vs +13.5. PRICE returns, not total returns. Top band of the L817 table on both scored windows - and close to meaningless: the shares rose because a takeover was agreed at a 126% premium and its odds improved. RS is not a term in the timing_score formula (L845).",
"catalyst_clustering_score": 40,
"catalyst_clustering_note": "Three high-impact events inside 17 days - US CPI 11 Sep, FOMC 16 Sep, and the 27 Sep revised-proposal deadline. The framework's noisy band (20-49) for 2-3 events inside a fortnight. Modulates position sizing, and below 30 would dock confidence; it is NOT the catalysts term in the timing formula, so it does not move timing_score.",
"dynamic_macro_weight": 0.15,
"sector_median_note": "Listed container liners trade at roughly 4-6x EV/EBITDA through the cycle; ZIM is at 3.72x. ESTIMATED from the peer set, not sourced from a data feed this run.",
"own_history_percentile_52w": 62,
"implied_growth_pct": null,
"implied_growth_rate": null,
"peg_computed": null,
"clean_pe": 12.62,
"eps_forward": null,
"clean_pe_note": "12.62x = price 29.38 / clean EPS 2.327, where clean EPS is operating income 387.4m taxed at the effective rate (27.5% on the TTM sum; 27.0% on the filed figures, which moves the multiple to 12.53x - immaterial, and it is explicitly a ceiling) over 120.658m diluted shares. Published because non-operating items are -141.6% of net income, past the 15% threshold. It is a CEILING, not a fair value: most of the drag is the 457.1m of interest a chartered-fleet operator genuinely owes, so 12.62x describes a company with no financing cost, which does not exist. Reported trailing is 25.55x - the drag makes the headline multiple OVERSTATE, not flatter.",
"analyst_consensus_target": 25.8,
"analyst_target_high": 38.0,
"analyst_target_low": 16.5,
"analyst_target_upside_pct": -12.2,
"analyst_coverage_count": 4,
"analyst_grades_consensus": "Hold",
"analyst_bullish_pct": 0.0,
"analyst_grades_n": 6,
"analyst_grades_detail": "get_grades_consensus: 0 strongBuy / 0 buy / 3 hold / 3 sell = Hold, n=6, 0% bullish. Yahoo's target feed is a DIFFERENT sample at n=4 (0/1/1/0/2) with a mean of 25.80 - the two counts differ by source, and both are stated wherever they are used. get_price_target_consensus returned a DEGENERATE response (high=low=consensus=median=20.00) and is not used.",
"analyst_dispersion_note": "High 38.00 against low 16.50 is a 2.3x spread, past the framework's 2x trigger, on only 4 analysts. The mean of 25.80 sits BELOW the market price and between the deal and break poles, which looks like an unlabelled probability-weighted number. Valuation confidence docked to 45%.",
"gates_triggered": [
"Gate 5 - Binary / Regulatory Event (Israeli state approval outstanding; 27 Sep revised-proposal deadline)"
],
"gates_caution": [
"Gate 6 - Structural: sub-scale in a consolidating industry, and thin consensus has FY2027 at a LOSS standalone"
],
"gates_retired_this_run": "Gate 1 Financial Distress was an August CAUTION and is now CLEAR: current ratio 1.243, EBITDA/interest 4.2x computed from the income statement, FCF 991m, cash 1,638m, net debt 1.9x EBITDA. August's trigger was a lease-inclusive debt figure read as distress on a company that charters its fleet. Gate 2 Earnings Event was August-TRIGGERED and is now CLEAR: date-driven and self-clearing, Q2 printed 19 Aug and Q3 is ~mid-Nov.",
"hard_gate_state": "caution",
"do_not_buy_triggers": [],
"entry_groups_met": 1,
"entry_conviction": "Half-Size",
"short_entry_confirmed": false,
"short_hold_reason": "Gate 5 caps all three horizons at HOLD; the short horizon is not separately capped.",
"exit_groups_live": 0,
"exit_action": "Hold",
"stop_loss": null,
"stop_loss_note": "DELIBERATELY not published as a price. Nearest support 23.29 is 6.3x the daily ATR of 0.9691 away, and the risk is an overnight gap on a ministry announcement rather than a drift through a level. A stop at 23.29 would not fill at 23.29. Position size is the only real risk control here.",
"next_update_date": "2026-09-28",
"next_update_basis": "EVENT-PINNED, not a default: the 27 September deadline for the revised Hapag-Lloyd/FIMI proposal to the Israeli authorities, +1 day. This supersedes the working note carried into this run, which had only 'expected close late 2026' and no dated catalyst.",
"threshold_fragility": "Three live fragilities. (1) The whole report turns on a single political decision with a hard date 17 days out, and the scenario set is BIMODAL - there is no meaningful outcome between 18.00 and 35.00, so the probability-weighted 28.79 describes a world that cannot occur, and it now sits BELOW the traded price of 29.38. (2) The valuation anchor is a JUDGEMENT that the contracted price is warranted value; on a standalone DCF or mid-cycle ROE basis the warranted multiple would be lower and the 0.84 ratio would not be Attractive. Valuation confidence is 45% for that reason. (3) The base signal is UNDEFINED: quality 40 (medium) + valuation 68 (attractive) + timing 47 (neutral) is a cell the Decision Matrix does not define. Gate 5 caps at HOLD regardless, so the published signal does not depend on the hole - but a different gate outcome would leave the matrix unable to answer.",
"framework_gap_raised": "SKILL L1268-1282: the Decision Matrix defines medium-quality + attractive-valuation for IMPROVING and WEAK timing but not for NEUTRAL. ZIM lands exactly there. Second framework gap found this session, after the catalyst-clustering-vs-catalyst-term ambiguity.",
"user_horizon": null,
"user_allocation_pct": null,
"portfolio_role": null,
"delta_vs_prior": "vs calibration-ZIM-20260806-0421.json. valuation 70 -> 68. timing 53 -> 47. driver 38 -> 48. economic alignment 40 -> 45. price 26.58 -> 29.38. warranted_ratio set to 0.84 (absent in the prior report). val_band na -> attractive. short_entry_confirmed true -> false. analysis_status donatien-pick -> stopped. next_update 2026-08-20 -> 2026-09-28. bear target 19 -> 18. relative strength vs SPY strongly positive (takeover-gap artefact) -> 10.9. relative strength vs sector positive (takeover-gap artefact) -> 14.7. moat 28 -> 45. industry benchmark score 30 -> 70. competitive share trajectory losing -> stable. overall confidence 35 -> 45. probability-weighted target set to 28.79 (absent in the prior report). scenario weights {'bull': 10, 'base': 50, 'bear': 40} -> {'bull': 10, 'base': 52, 'bear': 38}. gates_triggered 2 -> 1 (added Gate 5 - Binary / Regulatory Event (Israeli state approval outstanding; 27 Sep revised-proposal deadline); removed Binary / Regulatory Event; Earnings Event Risk). gates_caution 1 -> 1 (added Gate 6 - Structural: sub-scale in a consolidating industry, and thin consensus has FY2027 at a LOSS standalone; removed Financial Distress (numeric trigger; IFRS-16 lease artefact)). Carried without change: signal_short, signal_medium, signal_long, quality, entry_conviction, entry_groups_met, exit_action, hard_gate_state, base target, bull target, exit_groups_live. NOTE: Gate 5 is CONTINUOUS from August under the unnumbered label 'Binary / Regulatory Event'; what is new is the 27 September date. Gates 1 and 2 are RETIRED on evidence, reasons in gates_retired_this_run. analysis_status donatien-pick -> stopped executes the auto-Stop August left unadjudicated. short_entry_confirmed true -> false: Gate 5 caps every horizon and the Catalyst group fails, so August's confirmed short entry is gone even though the ladder count and label read the same at 1 of 3 Half-Size. Two entries above are BASIS changes rather than movements and must not be diffed as performance: relative strength vs SPY and vs sector go from August's PROSE ('strongly positive (takeover-gap artefact)') to computed numbers, and moat / industry benchmark score sat NESTED in quality_detail in August rather than at the top level - build_delta now looks one level into *_detail objects, which is why they read 28 -> 45 and 30 -> 70 rather than 'absent in the prior report'. The scenario weights end at bull 10 / base 52 / bear 38 after two revisions this run; note the 38% bear is a RETURN to August's own 40%.",
"post_audit_fixes_round1": "Round 1 from a fresh auditor: 2 BLOCKER, 7 MAJOR, 3 MINOR. B1: section 12 rendered a red 'Exit action: Exit' badge on a HOLD report, because the Thesis Invalidation group was marked live with its first condition 'met' - a condition that refutes itself, since the decision window OPENS on 27 September and nothing has been rejected. Worse, the entry-ladder void keys off the calibration exit_action while the badge keys off the spec, so an un-struck 'Half-Size' entry pill sat directly above a red Exit. Group set to not_live, condition to unmet. B2: the section-15 status block gave a tick to get_analyst_estimates while describing it as returning nothing - a defect in the provenance block itself. M1 is the serious one and it is mine: the report asserted in FOUR places that get_analyst_estimates returns no EPS values. It does. My pull filtered on the wrong field names (estimatedEpsAvg instead of epsAvg) and I read my own emptied output as an empty endpoint. FY2026 4.31 (n=2), FY2027 -2.34, FY2028 -2.95. The P/B anchor SURVIVES but its stated justification was false and is re-based on the argument section 3 already made: 4.31 is a 6.8x forward multiple on what consensus treats as the last good year of a cycle with losses either side, and a single forward year cannot value this business. M2: 'TTM revenue up 8.9% year on year' was Yahoo's Q2-over-Q2 growth attached to a TTM base - TTM revenue is DOWN 24.9%, 6439.1 against 8575.1, and it was the lead sentence of the Quality pillar. M3: 'unchanged from August because nothing about the business has changed' was falsified by the report's own calibration - three components moved under the unchanged 40. M4: '2024 produced the largest earnings in ZIM's history' contradicted section 3 two sections earlier and the company's own release; FY2024 was 2.15bn against FY2021's 4.65bn. M5: the header rendered 'Next update: ' with NO VALUE on a report whose entire thesis is a dated event - spec.signals was missing next_update_date. M6: the folder's index.html redirect and canonical link still pointed at the August report, so the bare URL served a superseded document. M7: the Israeli opposition was named as three ministries, one of them wrong, and the report never mentioned that on 9 September - the day before publication and the day of its own price basis - ALL EIGHT reviewing authorities filed and SIX opposed. That omission ran in the direction that flattered completion odds, and it was the single most decision-relevant input available. The scenario weights move on it: base 65 -> 55, bull 10 -> 7, bear 25 -> 38, probability-weighted 31.00 -> 28.71, which at the time read 63.0% implied against the market's 66.9%. Round 2 revised this again - bull restored to 10%, base to 52% - giving 28.79 and 63.5%. Those round-2 figures are the published ones; the numbers in this sentence are the state as round 1 left it. M8: the ex-date dividend TTM was 1.25 / 4.25%, wrongly including a 0.06 that went ex eight days outside the window; it is 1.19 / 4.05%. Fixed in section 3 AND section 15 - the first pass fixed only section 3, which the verification caught. M9: 1,552m was labelled depreciation and amortisation in two places; D&A is 1,286.8m and the remaining 265m is non-operating income below EBIT. M10: the like-for-like implied completion figure.",
"audit_route_note": "Round 1: a FRESH general-purpose agent (never a fork). Verdict FAIL (2 blockers, 7 majors, 3 minors). Round 2: SendMessage to the SAME auditor \u2014 artifact patched, not rebuilt; the auditor read sections 3, 4 and 11 as units and then swept section 14, which nobody had. Verdict FAIL (2 blockers, 4 majors, 4 minors). Round 3: SendMessage to the SAME auditor \u2014 artifact patched, not rebuilt; the auditor confirmed section 14 clean as a unit and then found the round-2 fixes that landed in the tooling but not in the document. Verdict FAIL (0 blockers, 3 majors, 2 minors). Round 4: SendMessage to the SAME auditor \u2014 artifact patched, not rebuilt; section 14 and the audit trail swept as a SET across all three axes, with the headline arithmetic re-derived from scratch. Verdict PASS (0 blockers, 0 majors, 3 minors).",
"post_audit_fixes_round2": "Round 2: 2 BLOCKER, 4 MAJOR, 4 MINOR - and both blockers were section 14, which no fix pass had touched. R2-A: the section-14 one-paragraph summary was entirely pre-fix. It restated the M1 falsehood as settled fact three sections after section 4 retracts it, so a reader reaching the closing paragraph got the correction and then the falsehood, with the falsehood last. It also carried all four superseded scenario weights. Rewritten. R2-B: the section-14 EMBED diverged from the calibration on FIVE narrative fields - val_anchor_note and dnb_note still carried the retracted claim, clean_pe_note was pre-fix, and completion_probability_note and israeli_review_state existed on disk and not in the embed at all. My own embed_divergence check returned 'none' because it watched only the audit-trail fields; the numeric fields HAD updated, which is exactly why a narrow check passed - it was comparing the fields that moved and not the ones that did not. Two fixes: the check now compares EVERY narrative field, proven by a new self-test case, 5/5; and the five notes moved from the calibration file into the SPEC, which is their author, so the two-author bug is closed structurally rather than by patching. R2-C threshold_fragility named the superseded 31.00. R2-D exit_groups_live was 1 against a not_live group and a Hold badge. R2-E the disowned 'could not be struck' framing survived in the section-4 subtitle, the pillar caption, val_multiple_basis and the section-15 haircut - the BODY was re-based and its furniture was not. R2-F delta_vs_prior recorded none of the Quality component moves; moat_score, industry_benchmark_score, competitive_share_trajectory, overall_confidence, exit_groups_live and probability_weighted_target are now in DELTA_FIELDS so it is mechanical rather than remembered. R2-G section 8 counted three moves against a four-row table. R2-H the WCI comparison I reported as added in round 2 was NOT in the artifact - a false claim in my own round message, now actually added at 2,424 on 7 Aug 2025 against 4,465 on 3 Sep 2026, +84%. R2-I the route note stuttered. R2-J the PEG paragraph re-argued the spread after the section disowned it as the objection. And on the auditor's push-back: the bull cut 10 -> 7 was unexplained and ran the wrong way, since the Sakal structure is DOMESTIC ownership, which is what the objecting bodies want - a break should make a domestic bid MORE likely. Bull restored to 10%, base takes the whole reduction to 52%, bear 38% held at August's own level. Probability-weighted 28.79, -2.0%, implying 63.5% against the market's 66.9% - agreement within noise, and section 11 now says explicitly that this is not a claim to an edge, since the 9 September filings were public before the close that prices the report.",
"post_audit_fixes_round3": "Round 3: 0 blockers, 3 majors, 2 minors - and every one was the same pattern, a round-2 fix that landed in the TOOLING but not in the ARTIFACT. R3-1: build_delta's nested *_detail lookup was written and proven, and delta_vs_prior was never regenerated through it, so the published field still asserted 'absent in the prior report' FIVE times - four of them false, because August carries val_band, both relative-strength fields, moat and industry_benchmark_score nested in quality_detail / valuation_detail / timing_detail. Regenerated: val_band na -> attractive, relative strength prose -> 10.9 and 14.7 flagged as a BASIS change, moat 28 -> 45, industry benchmark 30 -> 70. This is the audit brief's rule that a fix must be verified in the RENDERED artifact and not in the source, and I had verified the code. R3-2: section 8's Impact cell still said the base weight moved to 55% when it is 52%, contradicting section 11, section 14 and the embed. R3-3: completion_probability_note was a full round behind - every figure in it superseded - and it renders inside section 14, the block that produced both of the previous round's blockers. R3-4: post_audit_fixes_round1 carried a present-tense claim, 'the report NOW reads more cautious', inside a historical record that a later round had falsified; past-tensed. R3-5: scenario_probabilities added to DELTA_FIELDS, on the report whose weights are its single most-revised quantity. AND the two false-positive checks are fixed on the auditor's analysis, which drew the right line: rewording is the right tax when a check reads free prose that was genuinely ambiguous, and the WRONG tax when a check regexes text a generator produced in a known shape. LADDER-DELTA is an f.err whose false positive was STRUCTURAL - build_delta emits a 'Carried without change: ... entry_conviction, entry_groups_met, exit_action' tail by design, so on any report where one ladder field moves while another holds, a HELD_WORD is GUARANTEED within 90 characters. It now scans only the free prose preceding that generated tail. DELTA-DIRECTION is an f.warn behaving as designed, but its symmetric 70-character window was blind to sentence boundaries, which is how 'weaker' (describing the entry position) landed near 'moat' (which rose); it is now scoped to the containing sentence, and the failure it hunts is always intra-sentence so no signal is lost. Both have regression cases proving the false-positive shape no longer fires: 74/74.",
"post_audit_fixes_round4": "No fixes required. Round 4 returned PASS at 0 blockers and 0 majors. This round changed only four calibration fields and one table cell: delta_vs_prior (regenerated through the nested-lookup fix), completion_probability_note (rewritten to the current 10/52/38 with its two-revision history), post_audit_fixes_round1 (past-tensed so a historical record stops making a live claim) and section 8's impact cell (55% -> 52%). Three items are LOGGED AND ACCEPTED rather than fixed, on the auditor's explicit instruction not to reopen for them. (1) 'warranted_ratio set to 0.84 (absent in the prior report)' is true but under-informative beside its neighbour 'val_band na -> attractive': August WITHDREW the anchor rather than lacking the field, so a next run could diff it as new when the truth is that an anchor was struck where one had been declined. (2) completion_probability_note uses 'round 1' in two senses in adjacent sentences - the draft audited IN round 1 and the fix made BY round 1. (3) A SKILL-level gap, not a ZIM defect: timing_score is not reproducible from the published terms - section 7 gives four of the five (MTF 62, risk-reward 30, sentiment 30, catalysts 50, dynamic_macro_weight 0.15) but never the macro sub-signal itself, so a reader cannot check the pillar. That belongs in the framework, not in this report. Also carried: the FW-MATRIX-SKIP framework gap, where quality 40 medium + valuation 68 attractive + timing 47 neutral is a cell the Decision Matrix does not define; Gate 5 caps at HOLD regardless, so the published signal never depended on it."
}
In one paragraph. ZIM is a HOLD on all three horizons, unchanged from August, and the reason is a hard gate rather than a scorecard. The company is subject to a binding all-cash merger agreement with Hapag-Lloyd at $35.00 per share, approved by its board and by shareholders in late April 2026, and the shares at $29.38 sit 16.1% below that offer — a +19.1% return if it completes, and nothing beyond it, because a cash offer has no upside past its own price. What is not secured is the State of Israel: six of the eight reviewing authorities filed against the original structure on 9 September, and Hapag-Lloyd and FIMI must file a revised proposal by 27 September 2026 under a 30-day extension. Gate 5 caps every horizon at HOLD, and that cap is load-bearing — on the valuation score alone this would read as a buy. The valuation anchor moves to price-to-book, at 0.84 of the contracted price, because a single forward year cannot value a business whose quarterly operating income ran from $454m to −$14m inside four quarters — the four available FY2026 estimates span $4.31, $3.15, $0.29 and $9.18, and the best of them is a 6.8× multiple on what consensus treats as the last good year of a cycle. That basis change is stated in §4 as the judgement it is. Scenarios are deliberately bimodal — $35.00 if it closes at 52%, $18.00 if it breaks at 38%, $37.50 at 10% on the four-month-old Sakal proposal — weighting to $28.79, or −2.0%, against a market price that implies 66.9% on the same two poles. The 38% bear is August’s own weight, not a new alarm; what changed is that on 9 September all eight Israeli reviewing authorities filed and six opposed, which an earlier draft of this report had not accounted for. There is no meaningful outcome between $18 and $35, so the average describes no world that will happen. Two August gate findings are retired on evidence and recorded rather than dropped: financial distress, which was a lease-inclusive debt figure read as a solvency problem on a company that charters its fleet, and the earnings-event gate, which was always date-driven and has self-cleared. Finally, one thing August left undone is done here: a Donatien Pick returning HOLD on every horizon must come off the active pick list, so ZIM moves to Stopped — still visible on the watchlist, re-activating on any future BUY. That omission was caught by a deterministic pre-flight check rather than by reading it, which is the only reason it was caught at all.
Confidence haircuts, and why.
Valuation 45%, the lowest score in this report and deservedly so. The default anchor was not used because a single forward year cannot value a business whose quarterly operating income ran from $454m to −$14m inside four quarters — and the four available FY2026 figures span $0.29 to $9.18; the replacement anchor rests on a judgement that the deal price is warranted value; PEG is not computable; and the analyst lens — 0% bullish, a mean target below the market price — argues against the conclusion. A 68 with 45% confidence is the most honest pair I can produce.
Timing 40%. Three of five timeframes are in uptrends, but they are measuring a price anchored to a contract and moved by political headlines. Technical structure on a deal-pinned equity is close to a category error, and a confluence score that looks bullish should not be read as one.
Drivers 45%. The freight tape is genuinely mixed — transpacific at a 22-month high against a softening Asia–Europe and returning Suez capacity — and I could not determine whether the rate benefit or the bunker cost dominates. More to the point, a deal-pinned equity does not transmit its drivers to the holder at all unless the deal breaks.
Quality 65%. The highest confidence here, because the standalone business is well documented and unambiguous: sub-scale, no moat, extreme operating leverage, and six quarters of earnings running from −$0.71 to $2.45 per share.
Economic alignment 55%. The sector read is defensible and largely beside the point — the macro exposure that decides this name is Israeli sovereign politics, which no sector forecast captures.