A rule nobody could break, and a deposit the ledger has not taken

Five complete verified sessions ran from 28 September to 2 October, so the Friday-to-Friday comparison is available without a caveat for a fifth week running.

From the 25 September close to the 2 October close the All-Share (DSEI) fell from 4,682.13 to 4,631.77, about −1.08%. The domestic Tanzania Share Index (TSI) fell from 10,419.49 to 10,255.60, about −1.57%. Banks, Finance & Investment fell from 23,714.23 to 23,188.19, about −2.22%, giving back roughly two thirds of last week’s 3.44% gain. Commercial Services fell from 3,563.34 to 3,507.72, about −1.56%, ending two consecutive gains. Industrial & Allied was the only sector index higher, from 5,053.49 to 5,075.25, about +0.43%; it was the only one lower a week ago.

The spread between the best and worst sector index was about 265 basis points, against roughly 592 a week ago and 1,354 the week before that. Third consecutive narrowing. The sectors are disagreeing less each week, and this week they agreed on the way down, which is worth knowing before “breadth is improving” gets written about a narrowing spread again.

The sessions ran −0.50%, −0.18%, +0.08%, −0.31%, −0.18%; four of five lower, and none of them dramatic. Breadth across those five sessions was 27 advances against 48 declines, against 42 and 37 the week before. On the Friday-to-Friday arithmetic, across the 28 counters with a verified close on both dates, 9 rose, 11 fell and 8 were unchanged, against 12 rose, 11 fell, 5 unchanged.

And it cost the same to produce. Equity turnover across the five sessions was TZS 27.874bn, about +1.5% against the previous week’s TZS 27.471bn. That ends three consecutive weekly falls of −13.1%, −29.3% and −48.2%, but ending a decline is not the same as reversing one: this is the same money as last week, and last week was the thinnest in a month. Last week prices rose on half the money. This week prices fell on all of the same money. Flat turnover is not a flat market.

Where the money came from went the other way from last week, and sharply. CRDB was the largest counter by turnover in all five sessions: 55.24% on Monday, 46.57% on Tuesday, 62.49% on Wednesday, 48.87% on Thursday and 65.44% on Friday. A week ago the leader changed four times in five sessions, and this ledger called that an improvement. It lasted one week. One counter out of twenty-eight supplying roughly half of every session’s turnover is the market structure this book actually operates in, and a week of rotation at the top of the turnover table does not change it.

Every mark reconciles to the verified DSE close. From 25 September to 2 October: NMB TSh 2,130 to TSh 2,040 (−4.23%), Vodacom TSh 1,280 to TSh 1,260 (−1.56%), Twiga (TPCC) TSh 5,700 to TSh 5,670 (−0.53%), CRDB TSh 2,940 to TSh 2,950 (+0.34%), Tanzania Cigarette (TCC) TSh 13,200 to TSh 13,350 (+1.14%). Two of five rose.

On the model book’s unit counts that is about −TSh 5.03m from NMB, −TSh 1.67m from Vodacom and −TSh 0.06m from Twiga, against +TSh 0.67m from TCC and +TSh 0.17m from CRDB. Net equity contribution: about −TSh 5.92m, or −1.71% on a sleeve worth TSh 346.31m at the start of the week. 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 5.55m.

That arithmetic is the week in one line. A single holding cost more than the other four returned, and the ballast earned back about 6% of the loss. It is worth being exact about what that means, because it is where model books usually flatter themselves: a fixed-income sleeve does not rescue a bad equity week. It declines to add to it. On TSh 192.3m of bonds, bills and an income fund accruing at indicative yields, a week is worth roughly a third of a million shillings, and no honest arrangement of those numbers turns that into a hedge.

The hypothetical NAV fell from TSh 617.96m to TSh 612.41m. Contributed capital did not change in the book of record; it stands at TSh 521.0m, and the next section explains why that sentence needs a footnote this week. Model market P&L fell from TSh 96.96m to TSh 91.41m, and the money-weighted return since inception on 6 June fell from +19.00% to +17.89%.

In the Friday-to-Friday series this ledger has published since 17 July, −0.90% is the largest weekly fall on record. The only other lower week was 21 August at −0.14%, so a run of six higher weeks ends here. The book’s record reading is still TSh 620.28m on 22 September, and Friday closed about 1.27% below it.

The Friday number is again not the week’s high. The book marked TSh 616.57m on Wednesday 30 September and finished at its weekly low on Friday. Three weeks running, the intra-week peak has been on a Tuesday or a Wednesday and the Friday print has been lower; the difference this week is that the direction of the week makes the habit easy to report rather than awkward. A ledger that only flags the pattern in weeks the pattern flatters it is not a ledger.

Chapter 21 published the single-name cap on 1 October, after three chapters had dated it there: 20% of the book, binding on new savings only, never forcing a sale, with a drift breach reported weekly rather than traded.

In its first week the largest model-book equity allocation fell 4.23% and its weight went from 19.26% to 18.61%. The top two fell from 36.52% to 35.76%. So the cap did nothing, and could not have done anything: its trigger is new money, and the next new money is dated 1 November.

That is not a complaint about the rule; it is the honest reading of its first week. The cap was not vindicated by a week in which the number moved away from it, any more than it would have been refuted by a week in which the number moved toward it. A rule earns its keep the first time it stops something. This week there was nothing to stop, and a book that treated a quiet week as evidence for its newest rule would be marking its own homework seven days after setting it.

What the week does put a small amount of weight on is the other half of the Chapter 21 design: that price drift gets reported rather than traded. The drift this week was downward, which is the easy direction to leave alone. The real test of that clause arrives the first week drift runs upward and the report says so while the book does nothing about it. Readers should hold this ledger to that, and no conclusion about the cap is being drawn before then.

Two disclosures about this book’s own bookkeeping, because a reader should not have to find them.

First, October’s savings are not yet in the book of record. Chapter 21 allocated TSh 7.0m on 1 October: 5.0m to government bonds, 1.0m to the collective income fund, 0.5m to cash and 0.5m to the gold sleeve. As at the 2 October mark the canonical tracker still records three contributions, dated 1 July, 1 August and 1 September, and contributed capital of TSh 521.0m. The four destination sleeves are short by exactly 5.0m, 1.0m, 0.5m and 0.5m. Every live figure on these pages, the TSh 612.41m NAV included, is therefore the book before the deposit lands, and will step up by TSh 7.0m of capital when it does.

Second, Chapter 21 paired two numbers from different days. It reported contributed capital of TSh 528.0m alongside a money-weighted return of +18.70%. The +18.70% is the 30 September mark, struck before the deposit; the TSh 528.0m assumes it. Pairing them was an error, and it is this ledger’s error to name on the next page it publishes rather than to quietly re-mark.

Here is the reconciled arithmetic, which is also the lesson. Model market P&L on 2 October is TSh 91,411,890. That figure is identical whether or not the October deposit is recorded, because a deposit adds the same amount to the book’s value and to its capital, and P&L is the gap between them. The money-weighted return on 2 October is +17.888% with three contributions recorded and +17.886% with four: a difference of two thousandths of a percentage point, because money arriving on the last day of a 118-day window is weighted at roughly one day of it.

So when the tracker records the October flow, the NAV will rise by TSh 7.0m and the reported return will not move. That is what “contributions are capital, not performance” means in arithmetic rather than in principle, and it is the clearest demonstration of it this book has had. If a reader takes one thing from this chapter, take that: a published return that rises when money is deposited is being measured wrongly.

Until the entry is made, the figures to read on these pages are market P&L of TSh 91.41m and a money-weighted +17.89%, which are the same on either side of it. The correction is dated here, and the capital figure will reconcile at the next mark.

Of the dated items the last two chapters left open, one moved at an unexpected rung, two did not print, and one cannot be checked at all.

No Treasury bill auction settled in the window. Auction 1207 of 23 September remains the latest listed on the Bank of Tanzania’s own index, checked 2 October: 364-day 6.2852%, 182-day 5.2867%, 91-day 3.3872%, 35-day 2.5215%. So the decision Chapter 21 took on the evidence of five consecutive falls at one year gets no sixth data point this week. The next auction falls due on the fortnightly cadence, and it is the first thing that could argue with that decision.

The bond store moved, at a rung nobody was watching. Its latest readings now carry a 15-year yield of 10.79% dated 30 September, where the most recent 15-year auction in the series is 10.39% from 13 May; roughly 40 basis points higher at the long end. Two caveats, both material. The 30 September reading sits in the curve’s latest table without a matching per-auction row in the bond series, so the Ghost publishes it as the store’s current 15-year reading and not as a verified auction print. And it is not the rung that was being watched: the 10-year’s 10.69% of 2 September is still untested, for a third chapter.

Against August headline inflation of 4.3%, the government’s own curve still lends roughly 649 basis points above inflation at fifteen years, 639 at ten and 523 at five, while the 364-day bill lends about 199 above and the 91-day about 91 below. The policy rate is 6.25% for the third quarter. A Tanzanian press report of an adjustment has now gone six weeks without appearing on the Bank’s own published sheets; the Ghost carries 6.25% and keeps the report as a watch item, not a rate.

The item that cannot be checked. The largest model-book equity allocation fell 4.23% this week, after a 5.48% fall two chapters ago that still has no published cause. The DSE listed-company filing register in the evidence store was generated on 23 September and has not been regenerated since, so the store can neither confirm nor exclude a filing in the ten days that followed. The Ghost will not supply a cause, and it will not report an absence it has not actually looked for. The register’s own staleness is the finding this week.

Two standing disclosures about this book’s arithmetic. The bond and bill sleeves accrue at indicative yields rather than being marked to auction, a deliberate simplification stated on the method page, so none of the curve movement above is in the fixed-income line. And the commodities sleeve is carried at its contributed value, not marked to the metal: the Mining Commission’s world gold reference fell from $4,291 to $4,191 per troy ounce this week, about −2.33% and a fifth consecutive weekly fall, with a single session on 29 September accounting for −3.21% of it. 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 Nation Media (NMG), from TSh 290 to TSh 315, about +8.62%. NICO added +4.43% to TSh 3,770, Tanga Cement (TCCL) +3.67% to TSh 3,950, and the exchange’s own listed counter DSE +2.42% to TSh 6,350.

The largest falls: Maendeleo Bank (MBP) TSh 2,150 to TSh 1,990, about −7.44%; DCB TSh 420 to TSh 400, about −4.76%; AFRIPRISE TSh 780 to TSh 745, about −4.49%; TOL TSh 1,810 to TSh 1,730, about −4.42%; and the book’s own largest holding at −4.23%.

Read those against the −2.22% banking index. CRDB rose 0.34% and MKCB 0.55%, while MBP fell 7.44%, DCB 4.76%, NMB 4.23%, MCB 1.33% and KCB 0.45%. Last week this ledger noted a banking index that rose 3.44% while containing two constituents that fell four to six percent. This week is the mirror image: an index that fell 2.22% containing two constituents that rose. “Banks were down” is a sentence about weights, not about banks; the same sentence as last week, in the other direction, seven days later.

Tanga Cement rose 3.67%, its first higher week in three, while the book’s own cement holding fell 0.53%. Two chapters ago this ledger declined to conclude anything about owning the wrong cement counter from one dramatic week in Tanga’s favour. It declines again now the comparison has turned back the same way. A view that changes every Saturday was never a view.

MUCOBA closed unchanged at TSh 450 for a second week, after +11.11% and −11.96% in the two before. AFRIPRISE gave back 4.49% after +22.31% and +5.41%. Small counters with thin trading produce large weekly percentages and very little information, and that is worth repeating in the weeks it is boring rather than only in the weeks it stings.

MBP enters its sixth chapter without a published cause, having moved roughly +17%, then −7%, flat, +1.48%, +4.37% and now −7.44%. Naming an absence six times is duller than naming it once, and it is still the point of keeping a ledger.

One corroborating reading, dated and at medium confidence: on 30 September, foreign participation was 35.26% of buy-side turnover, TZS 3,407.50m, in a session where one counter carried 62.49% of the tape. Two concentrations at once, pulling in the same direction.

Three reasons.

First, there is no drift a decision would correct. Equities 55.58% against a band of 30 to 60, fixed income 31.40%, cash 9.42%, gold 3.59%. Every sleeve is inside its mandate band, and the sleeve that had been closest to its ceiling moved away from it this week without being touched. There is nothing here that a transaction improves.

Second, this week’s decision was already taken, two days ago. Chapter 21 committed the month’s savings and published the cap on 1 October. A book that allocates on the first of the month and then revisits it on the third has not got a monthly process; it has got a mood with a calendar attached. The next allocation is dated 1 November, and the only thing that should move it forward is evidence, not a soft week.

Third, symmetry, again. This book did not trim the largest allocation when it rose 2.90%, and it does not add to it because it fell 4.23%. A fall of that size in a holding that is 18.61% of the book, with no published cause available, is a reason to read the filings when the register is regenerated. It is not a reason to transact on a Saturday on the strength of not knowing.

Nothing was trimmed and nothing was sold. One count does end here: Chapter 20 recorded a twelfth consecutive week unchanged, and that run stops, not because anything was sold but because Chapter 21 committed new savings on 1 October. This week changes no existing holding.

Four dated things. The next Bank of Tanzania bill auction, due on the fortnightly cadence after 1207, and whether the 364-day rung falls a sixth time; that is the first evidence that can argue with Chapter 21’s duration answer. A regeneration of the listed-company filing register, which would finally say whether anything was disclosed in the ten days after 23 September. The next bond auction at ten years, still the first real test of 2 September’s 10.69%. And the 1 November allocation, which is the first occasion on which the single-name cap can bind on anything at all.

Two smaller marks. The official USD/TZS mean on the Bank of Tanzania’s own sheets moved from 2,645.4486 on 25 September to 2,641.4186 on 2 October, about −0.15%; the shilling’s first firmer week after two weaker ones, with the provisional sheet for 3 October reading 2,639.2494. And the Mining Commission’s world-gold reference fell about 2.33% to $4,191, with the domestic buying-centre price at $3,688, 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 3.59%, inside its 0 to 8% band, carried at contributed value, and stays where it is.

The discipline for the week is short. When a new rule’s first week is quiet, do not mistake quiet for proof. When a sector index moves, look for the constituents that went the other way; in either direction, because the lesson is not about banks going up. And when a book’s own reported return would not move if money were added to it, that is the return being measured correctly, not a flaw to be fixed.

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