Turnover halved, and the book set a record anyway

Five complete verified sessions ran from 21 to 25 September, so the Friday-to-Friday comparison is available without a caveat for the fourth week running.

From the 18 September close to the 25 September close the All-Share (DSEI) rose from 4,602.26 to 4,682.13, about +1.74%. The domestic Tanzania Share Index (TSI) rose from 10,166.24 to 10,419.49, about +2.49%. Banks, Finance & Investment rose from 22,924.73 to 23,714.23, about +3.44%, recovering roughly four fifths of last week’s 4.32% fall. Commercial Services rose from 3,406.60 to 3,563.34, about +4.60%, a second consecutive gain. Industrial & Allied was the only sector index lower, from 5,120.96 to 5,053.49, about −1.32%.

The spread between the best and worst sector index was about 592 basis points, against roughly 1,354 a week ago. Breadth improved on the same arithmetic: across the 28 counters with a verified close on both Fridays, 12 rose, 11 fell and 5 were unchanged, against 5 rose, 15 fell, 8 unchanged. A market where the sectors disagree less and the counters rise more is, on its face, a healthier one than last week’s.

Except that it cost half as much to produce. Equity turnover across the five sessions was TZS 27.471bn, about −48.2% against the previous week’s TZS 53.054bn. That is the third consecutive weekly fall, after −29.3% and −13.1%, and it is much the largest of the three. Prices rose on materially less money changing hands, which is a fact about conviction as well as about direction, and it is the single most useful thing in this week’s tape.

One thing that did improve is who was doing the trading. Last week a single counter supplied 66.0% of the week’s entire turnover. This week the largest counter by turnover changed four times in five sessions: KCB at 32.66% on Monday, CRDB at 70.63% on Tuesday, NMB at 59.84% on Wednesday, CRDB again at 47.34% on Thursday, NMB at 50.31% on Friday. Across the four sessions for which a normalised per-counter daily report sits in the evidence store, 22 to 25 September, CRDB supplied 46.8% and NMB 39.6% of TZS 15.711bn; together 86.4%. The 21 September per-counter report is not in the store, so the five-session per-counter split is not published here. Two banks out of twenty-eight counters is still a narrow market; it is just less narrow than one bank out of twenty-eight.

Every mark reconciles to the verified DSE close. From 18 September to 25 September: Vodacom TSh 1,220 to TSh 1,280 (+4.92%), CRDB TSh 2,810 to TSh 2,940 (+4.63%), NMB TSh 2,070 to TSh 2,130 (+2.90%), Twiga (TPCC) TSh 5,710 to TSh 5,700 (−0.18%), Tanzania Cigarette (TCC) TSh 13,400 to TSh 13,200 (−1.49%). Three of five rose.

On the model book’s unit counts that is about +TSh 5.00m from Vodacom, +TSh 3.35m from NMB and +TSh 2.21m from CRDB, against −TSh 0.89m from TCC and −TSh 0.02m from Twiga. Net equity contribution: about +TSh 9.65m, or +2.87% on a sleeve worth TSh 346.3m. About TSh 0.37m of indicative income accrued across the fixed-income and cash sleeves. Net model market P&L for the week: about +TSh 10.02m.

The hypothetical NAV rose from TSh 607.9m to TSh 618.0m. Contributed capital did not change; it stands at TSh 521.0m, where September’s savings left it on 1 September. So the whole of the rise is return, not deposit. Model market P&L rose from TSh 86.9m to TSh 97.0m, and the money-weighted return since inception on 6 June rose from +17.06% to +19.00%.

The Friday number is not the week’s high. The book marked TSh 620.28m on Tuesday 22 September, a money-weighted +19.46% and the highest reading this ledger has ever carried. It then gave back about TSh 2.32m across the remaining three sessions and closed the week roughly 0.37% below its own record. Two weeks running, the intra-week peak has been on a Wednesday or a Tuesday and the Friday print has been lower; reporting only Friday would be accurate and would keep hiding the same thing.

It is worth putting the record next to the turnover figure and leaving them there. A model book of five Tanzanian equities recorded its best reading in the same week that the money changing hands on the exchange halved. Both statements are true, they are about the same five days, and only one of them is flattering. A number that goes up on thinning volume is a number that has been tested less, not more.

Chapter 18 dated a single-name cap decision to the 1 October allocation. Chapter 19 watched the price take the largest allocation back under 20% without a decision, and kept the date anyway. This week the same allocation rose to 19.26% and the second-largest to 17.26%, so the pair is 36.52% of the book; up for a third consecutive week, and up in every one of those weeks purely because prices moved.

That is the whole argument for writing rules on a date rather than in a mood. Three weeks ago the number looked urgent, two weeks ago it looked resolved, and this week it is quietly higher than both. A book that reacted to each of those readings would have trimmed, un-trimmed and trimmed again, and would have paid for the privilege each time. A book that wrote a date has done nothing three times and still arrives at the decision with better information than it had when it set the date.

So the date holds, unchanged. The single-name cap will be published as its own decision with the October allocation on 1 October 2026: whether a public model book with no single-name limit should have one, what the number is, whether it binds on purchases only or forces trims, and what happens when price drift alone breaches it. Nothing this week changes that, and nothing this week is a signal that a trim is coming. The decision may yet conclude that a concentrated book is the honest expression of a small number of researched views; if it does, it will say so with the reasoning attached.

The first of Chapter 19’s four dated items printed. Bank of Tanzania Treasury bill auction 1207 settled on 23 September: 364-day 6.2852%, 182-day 5.2867%, 91-day 3.3872%, 35-day 2.5215%. Against auction 1206 of 9 September that is about −18.8bp at one year, +36.2bp at six months, −3.5bp at three months and +23.2bp at 35 days.

Chapter 19 asked whether the one-year give-back was a step or noise. It is a step. The 364-day weighted average has now fallen at five consecutive auctions: 7.03% (15 Jul), 6.74% (12 Aug), 6.65% (26 Aug), 6.47% (9 Sep), 6.29% (23 Sep); about 74bp over ten weeks. Meanwhile the 35-day rung has risen at three consecutive auctions, from 2.09% to 2.52%. The short end is getting dearer and the one-year rung cheaper at the same time, which is a flattening from both directions rather than a rally.

That matters to this book, which holds TSh 121.2m of government bonds and TSh 45.6m of bills, and it is exactly the kind of observation that argues about duration. It is also a rotation, and rotations in this book are dated decisions with published reasoning, not Saturday reactions to a Wednesday auction. The place for it is the 1 October allocation, alongside the month’s TSh 7m of savings and the cap. Against August headline inflation of 4.3%, the 364-day still lends roughly 199bp above inflation and the 91-day roughly 91bp below it.

Two items did not print. No bond auction settled in the window; the store’s bond prints still stand at 16 September (5-year 9.53%), and the 10-year’s 10.69% of 2 September remains untested. And no NMB filing explaining last week’s move has appeared: the listed-company filing register, generated 23 September, still shows the issuer’s most recent statement period ending 30 June 2026. A 5.48% fall on 66.0% of a week’s turnover is now two chapters old and still without a published cause. The Ghost will not supply one.

One item enters its fifth week open. A Tanzanian press report of a policy-rate adjustment has still not appeared on the Bank of Tanzania’s own published sheets, which continue to carry the CBR at 6.25% for Q3 2026. The Ghost carries 6.25% and keeps the report as a watch item, not a rate.

Two disclosures about this book’s own arithmetic, because a reader should not have to find them. The bond and bill sleeves accrue at indicative yields rather than being marked to auction, a deliberate simplification stated on the method page; so the fixed-income line here is smoother than a real one would be, and none of the auction moves above are in it. And the commodities sleeve is carried at its contributed value, not marked to the metal: its sleeve index has read 100.0 every session since it opened. The Mining Commission’s world-gold reference fell from $4,383 to $4,291 per troy ounce this week, about −2.10% and a fourth consecutive weekly fall, and none of that fall is in this NAV. A model book that reports what it does not measure is more useful than one that quietly measures nothing.

The week’s largest riser was MUCOBA, from TSh 405 to TSh 450, about +11.11%; the same counter that was last week’s largest faller at −11.96%. Round trips like that, on small counters with thin trading, are a reminder that a weekly percentage is not by itself evidence of anything. AFRIPRISE added +5.41% after last week’s +22.31%, and Maendeleo Bank (MBP) rose +4.37%.

The largest falls: DCB TSh 445 to TSh 420, about −5.62%; MCB TSh 390 to TSh 375, about −3.85%; Nation Media (NMG) TSh 300 to TSh 290, about −3.33%; Swissport (SWIS) TSh 2,640 to TSh 2,560, about −3.03%; Tanga Cement (TCCL) TSh 3,900 to TSh 3,810, about −2.31%.

Those falls are worth reading against the +3.44% banking index. CRDB rose 4.63%, NMB 2.90%, MKCB 1.67% and KCB 0.46%, while DCB fell 5.62% and MCB 3.85%. A sector index that rose three and a half percent contains two constituents that fell four to six. “Banks were up” is a sentence about weights, not about banks.

Tanga Cement fell again, a second consecutive week, while the book’s own cement holding was roughly flat at −0.18%. Two weeks ago this ledger declined to draw a conclusion about owning the wrong cement counter from one dramatic week. The restraint continues to look right, and it continues to be worth saying so when restraint is cheap as well as when it is expensive.

MBP is now in its fifth chapter without a published cause for a counter that has moved roughly +17%, then −7%, then flat, then +1.48%, then +4.37%. Naming an absence five times is duller than naming it once, and that is still the point of a ledger.

Three reasons.

First, there is no drift a decision would correct. Equities 56.0% against a band of 30 to 60, fixed income 31.1%, cash 9.3%, gold 3.6%. Every sleeve is inside its mandate band. The equity sleeve used 0.66 of a percentage point of headroom on price alone and has about 3.96 left; that is a number to keep reading each week, not a number that forces anything this week.

Second, a record is the worst moment to change the rules. The book has just marked its highest reading, on its thinnest week of turnover in a month, with its top two allocations more concentrated than at any point this ledger records. Every one of those three facts is an argument for the cap that is already dated for 1 October, and none of them is an argument for moving that date forward by five days into a week that happened to go well.

Third, symmetry. The book did not trim the largest allocation when it fell 5.16%, or when it rose 8.42%, or when it fell 5.48%. Not trimming it in a week it rose 2.90% is the same rule applied a fourth time. The cap gets written on its date, ahead of the prices that follow it, which is the only order in which a rule constrains anything.

Nothing was bought, nothing was sold. The book is unchanged for a twelfth consecutive week.

Four dated things. The 1 October allocation, which carries the month’s TSh 7m of savings, the single-name cap decision, and now a duration question the bill curve has asked five times. The next Bank of Tanzania bill auction, due on the fortnightly cadence after auction 1207, and whether the 364-day rung falls a sixth time. The next bond auction, still the first real test of 2 September’s 10.69% at ten years. And any NMB disclosure covering the extraordinary turnover of 15 to 18 September, now two chapters unexplained.

Two smaller marks. The official USD/TZS mean on the Bank of Tanzania’s own sheets moved from 2,641.3390 on 18 September to 2,645.4486 on 25 September, about +0.16%, a second consecutive slightly weaker week and marginally larger than last week’s +0.13%; small in isolation, and no longer quite the run of flat weeks it was a month ago. And the Mining Commission’s world-gold reference fell about 2.10% to $4,291, with the domestic buying-centre price at $3,776, about 12.0% below world; the royalty, clearing and margin between the global market and the Tanzanian mine gate. The model book’s gold sleeve is 3.6%, inside its 0 to 8% band, carried at contributed value, and stays where it is.

The discipline for the week is short. When a book makes a record, read the turnover that made it before reading the number. When a sector index moves, look for the constituents that went the other way. And when a threshold keeps drifting back toward you in weeks nobody decided anything, that is the argument for the dated rule, not against it.

; The Ghost. Hypothetical and educational only; a model book, not real money, and never advice.