Three of the four narrowest ranges on the book are semiconductors: how quiet the design and fab names are

TL;DR

These are the five companies the whole AI narrative rests on, and in this channel they are the quietest instruments on the book.

• Of the six narrowest ranges among the 46 instruments listed 100 days or more, three are in this group: NVDAB 1.286x (3rd narrowest), TSMB 1.322x (4th), AVGOB 1.350x (6th). The only things narrower are two broad index ETFs.
• Group turnover of 0.6234x sits below the book 0.6625x. Three of the five are in the bottom quartile of the book by turnover: AVGOB 92nd, TSMB 85th, AMDB 79th.
• One name is the whole sector. NVDAB holds $43.80M, 12.1% more than the other four combined ($39.08M), and did $38.93M, 3.05 times their total ($12.75M). Remove Nvidia and the remaining four turn over 0.3261x, under half the book figure.
• But "semiconductors are simply narrow here" is wrong. Over an identical 110-day window starting the same day, MRVLB — another fabless chip designer — has a range width of 1.888x, 40% wider than AVGOB 1.350x; and SNDKB, listed the same day as NVDAB, is 2.432x, 1.89 times NVDAB. Narrow is a property of these five names, not of this industry.
• A narrow range does not mean it held up. The AVGOB low is 74.1% of its high, and it is sitting at 6.5% of that range, −24.2% from the high. Narrow says it fell slowly, not that it did not fall.
• Two of the five track ADRs (TSMB, ARMB). An ADR is itself a depositary receipt, so these two tokens are a token over a certificate over a receipt — one layer more than a US common stock.

1. What these five are

bStocks is Binance's in-house tokenized-equity issuer (BTech Holdings Limited, ADGM / FSRA listed certificates, BNB Chain only, traded on Binance only). CoinMarketCap reports 105 instruments; the snapshot behind this piece holds 100 of them, timestamped 2026-10-31T06:06:52Z.

These five cover both the design and the manufacturing end of semiconductors: three fabless designers (Nvidia, Broadcom, AMD), one foundry (TSMC) and one IP licensor (Arm). They are not five representatives of a theme — they are essentially all of the large semiconductor design and fab exposure in this snapshot.

The group holds $82.89M, 6.29% of the book, and did $51.67M, 5.92% of its volume. The volume share is below the market-cap share — this group is traded less than its size would imply.

For reference: the other 95 instruments turn over 0.6651x, the whole book 0.6625x, this group 0.6234x. The gap is small, but the direction is unambiguous — and section 4 shows how that "small" was dragged up by one instrument.

2. One name is the whole sector

One token, Nvidia, holds 12.1% more than Broadcom, AMD, TSMC and Arm combined, and does 3.05 times their total volume. It ranks 8th by market cap and 7th by volume, the only member of this group in either top ten.

Take it out and the remaining four turn over 0.3261x, 49% of the book 0.6625x. In other words, the 0.6234x that makes this group look close to the book average is Nvidia lifting four very low-turnover instruments.

Turnover rank across all 100: NVDAB 48th, ARMB 57th, AMDB 79th, TSMB 85th, AVGOB 92nd. Only Nvidia is anywhere near the median; the other four are all in the back half, three of them in the bottom quartile.

One reading to settle immediately: ARMB turns over 0.765x, the second highest in the group, and holds $1.76M, the smallest market cap in the group, 58th on the book. Turnover is 24h volume divided by market cap, and a small denominator inflates the ratio — the previous piece quantified that once and this piece does not repeat it, but ARMB 0.765x must not be read next to NVDAB 0.889x as "similar activity". ARMB absolute volume of $1.35M is 3.5% of NVDAB.

3. Three of the four narrowest ranges are semiconductors

Range width is the listed high divided by the listed low. The nine narrowest among the 46 instruments listed 100 days or more:

The two narrowest are broad index ETFs, as they should be — an index flattens single-stock movement by construction. What deserves attention is the four single names immediately after them, three of which are members of this group: NVDAB, TSMB, AVGOB. Excluding the two index ETFs, three of the four narrowest single instruments on the book are in this group, and 9th place QCOMB is another chip designer.

The full position of all five members:

A narrow range does not mean it held up, and that is the easiest cell in this table to misread. AVGOB has the 6th narrowest range, and its position in range is 6.5% — it is essentially sitting on its listed low. Its low of $319.48 is 74.1% of its high of $431.15, so getting from the high to the low only takes −25.9%, and it has already travelled −24.2%. Narrow says it fell slowly, not that it did not fall. ARMB is a different shape: range width 1.680x (21st, middling), −38.1% from the high, 5.8% into the range — it fell a lot and it is near the bottom.

For comparison: among those 46 instruments, the median width for common stocks is 1.768x (n=28) and for ADRs 1.459x (n=4). The three common stocks here print 1.565x / 1.350x / 1.286x, all three below the common-stock median; the two ADRs print 1.680x and 1.322x, one above and one below the ADR median.

4. The counter-example: another chip designer listed the same day is 40% wider

The ordering in the previous section reads easily as "semiconductors are simply narrow in this channel". Set it against the instruments that listed on the same day and it falls over immediately.

Counter-example one: MRVLB listed the same day as AVGOB / TSMB / ARMB, over an identical window, and is 40% wider.

Ten instruments listed on 2026-07-13, by range width (group members in bold):

Marvell, the underlying of MRVLB, is the same type of company as Broadcom — a fabless designer selling into the data centre. Same issuer, same listing day, the same 110-day window, and its range width of 1.888x is 1.40 times AVGOB 1.350x. The widest in the cohort, AAOIB at 2.453x, supplies data-centre optics and sits in the same AI compute supply chain.

Counter-example two: SNDKB, listed the same day as NVDAB, is 1.89 times as wide.

Six instruments listed on 2026-06-11, from 2.432x down to 1.286x:

NVDAB is the narrowest of those six and SNDKB is the widest, and both are semiconductor companies. SanDisk makes NAND, Nvidia makes GPUs, both listed the same day over the same 142-day window, and their widths differ 1.89 times. The memory group covers the SNDKB side of this. Dispersion inside one industry is larger here than dispersion between industries.

Counter-example three: turnover is not a sector property either. Several semiconductor-chain instruments outside this group turn over clearly faster:

All four are in the front half of the book by turnover, while four of the five group members are in the back half. But note their market caps: $979.6K to $2.43M, all around or below the book median of $2.39M, so a good part of that high turnover is the denominator effect.

Put the three counter-examples together and the claim narrows to something that survives: not "semiconductors are narrow and quiet", but "the largest semiconductor names in this channel are narrow and quiet". The difference is testable: these five hold most of the semiconductor design and fab market cap on the book, while the smaller names in the same industry (QCOMB, ASMLB, MRVLB, AAOIB) behave differently on both width and turnover.

There is one more ready-made counter-finding along the same line: the two lowest-turnover instruments on the whole book are both semiconductor-chain companies, and neither is in this group — AMATB (Applied Materials, $9.02M / 0.130x, 100th) and COHRB (Coherent, $8.44M / 0.141x, 99th). Together with AVGOB 0.224x (92nd), they say this: in this channel the quiet of semiconductors is not specific to this group, and it is not uniform across the industry either.

5. Putting this group back in the book

First, this group fell less than the book this month. The 30-day median across the 100 instruments is −16.71%, with 82 negative. All five members are negative, but three did better than the median: AVGOB −8.44%, TSMB −11.31%, NVDAB −11.95%. Only AMDB −18.88% and ARMB −31.34% did worse. The book median is roughly twice the decline of three of these members, so "the AI narrative broke" does not hold in this snapshot — what actually broke through is something else.

Second, listing early sets the size. The book by listing month:

NVDAB and AMDB belong to the June cohort (ten instruments holding 52.3% of all book value); AVGOB, TSMB and ARMB to the July cohort. The largest member of this group is also the earliest-listed one — a pattern that keeps repeating in this snapshot.

Third, ADRs are a small class on the book. Six of the 100 are ADRs (TSMB, BABAB, NOKB, PDDB, ARMB, ASMLB), worth $25.28M together, only 1.92% of the book, turning over 0.4472x in aggregate. The two ADRs in this group account for $10.77M of those six, the largest single piece.

Fourth, the kind mix. The book holds 67 common stocks, 8 unclassified, 6 ADRs, 6 ETFs, 10 leveraged ETFs, 2 inverse ETFs and 1 pre-IPO. This group is three common stocks and two ADRs, with no leveraged or inverse products at all — so the narrow ranges here do not depend on "no leverage" as an explanation, because section 4 already showed that MRVLB and AAOIB, equally unleveraged, are much wider.

6. Structure: two ADRs mean one more layer

All of this comes from the issuer record on this site, the same data behind /stocks/bstocks:

• Legal wrapper: tracker certificate, ADGM / FSRA listed, issued by BTech Holdings Limited. The holder gets a certificate that tracks a price, not a share registration.
• Voting: none.
• Dividends: pass-through, but a 30% US withholding deduction comes first, and the remainder is reinvested through a multiplier.
• Redemption: available, KYC required, $5 minimum.
• Proof of reserve: attestation, not a live on-chain proof.
• US persons: blocked.
• Chain: BNB Chain only. BSC deposits only, so a wrong-network transfer loses the asset.
• Issuance model: exchange self-issued; the only venue is Binance, operated by the issuer.
• Grade on this site: B.

Two of those matter extra for this group.

TSMB and ARMB carry one layer more than the other three. TSMC is listed in Taiwan and Arm is a UK company, and what trades in the United States for both is an ADR — a receipt issued by a depositary bank representing some number of foreign shares. So the chain for these two tokens is foreign shares, depositary bank, ADR, ADGM tracker certificate, BNB Chain token: four layers, against three for NVDAB / AVGOB / AMDB, whose underlyings are US-domestic common stock. Each extra layer adds a fee set, an operating rule set and a counterparty, and none of that is in this data.

That 30% records only one deduction. The issuer record states "deduct 30% US withholding, then reinvest through a multiplier". For a US common stock that is the whole story; for an ADR, the receipt may already have been through a source-country step when it received the foreign distribution, and depositary banks typically charge a service fee as well. This data records the US step and says nothing about what happened before it — this piece does not guess at rates, because the snapshot has none.

One related point about voting: most of Arm free float is held by SoftBank, and TSMC governance differs from a typical US company. None of that reaches the token holder, because voting at the certificate layer is "none" — structurally, governance differences between the underlying companies are flattened out entirely by this wrapper.

7. This snapshot was taken on a Saturday

The timestamp 2026-10-31T06:06:52Z is a Saturday morning UTC. US cash equities closed on Friday at 20:00 UTC and do not reopen until Monday — for more than two days after this snapshot, these token prices are the only prices these names have.

One detail worth stating for this group: the reference market for both ADRs is still the US market. ADRs trade in New York, so the "closed" moment for TSMB and ARMB is the same as for the other three, unlike a stock listed directly in Seoul or Taipei. But their underlying companies have their own sessions in Taiwan and the UK, and a gap between an ADR price and the local foreign price exists by construction — a layer entirely invisible in this snapshot.

This has already been quantified once on the same issuer: one stock, four books lays the four tokenized versions of AAPL side by side on a Saturday and concludes that a quote outside trading hours is a market maker opinion, not a market price. The 24-hour moves here (NVDAB −1.97%, AMDB −3.20%) run from Friday 06:06 UTC to Saturday 06:06 UTC and contain one complete US session, so they are not pure closed-market quotes.

8. What this method cannot compute

First, no premium or discount. The snapshot holds only the token price, with no reference price for the underlying stock or ADR. More to the point, the bStocks dividend mechanism is "deduct 30% US withholding, then reinvest through a multiplier" — without knowing the current multiplier, the token unit price and the underlying share price are not comparable in the first place. So no premium figure appears here.

Second, no shares per token. The snapshot carries no conversion ratio. AMDB has 20,585 tokens outstanding at $510.80 each, and neither number converts into a share count without that ratio. An ADR carries a ratio of its own (how many foreign shares one receipt represents), also absent from this data, so TSMB and ARMB sit two unknowns away from any share count.

Third, no cause for the narrow ranges. This piece can say the price range of these five tokens is narrow; it cannot say the underlying stocks moved little, because the snapshot holds no underlying price series. Range width is an observation on the token side, shaped at once by underlying volatility, market-maker behaviour and the liquidity of this channel, and those three cannot be separated.

Fourth, the range-width windows are unequal. These five are 142/110/132/110/110 days. The book comparison uses only the 46 instruments listed 100 days or more, and even there the windows differ by more than a month — the longer something is listed, the more easily it reaches an extreme, which makes the NVDAB 1.286x harder to achieve over the longest window, not easier.

Fifth, market cap is CMC's definition (price times circulating supply), and the book total of $1,318.38M is the sum of these 100 rows; the bStocks row on this site's /stocks page shows an on-chain size of $914M — two figures from a different source and a different moment, which will not tie out. This piece uses the former throughout.

Sixth, 100 of 105. CMC reports 105 instruments; the snapshot holds 100. The missing five appear in no figure here.

Seventh, turnover is 24h volume divided by market cap. A small denominator inflates the ratio, so the market caps of the four high-turnover instruments in section 4 (IBMB, ASMLB, QCOMB, SMHB) all sitting under $2.5M has to be read along with them — their high turnover must not be taken as better liquidity elsewhere in the industry.

Conclusion. The five semiconductor design and fab tokens hold 6.29% of book value and did 5.92% of its volume — in this channel, the five companies the whole AI narrative rests on are traded less than their size would imply. They are also the narrowest price ranges available: among the 46 instruments listed 100 days or more, and excluding the two broad index ETFs, three of the four narrowest single instruments are members of this group.

But that profile does not belong to the semiconductor industry, only to these five names. MRVLB, listed the same day over an identical window and also a data-centre chip designer, is 40% wider; SNDKB, listed the same day as NVDAB, is 1.89 times wider; and the two lowest-turnover instruments on the book, AMATB and COHRB, are both semiconductor-chain companies and neither is in this group. Dispersion inside the industry beats dispersion between industries.

Two readings travel from this. First, a narrow range has to be read next to the position in range — AVGOB has the 6th narrowest width on the book and is sitting at 6.5% of it: it fell slowly, it did not avoid falling. Second, this group having "average turnover close to the book" is one instrument doing the work — remove NVDAB and the remaining four print 0.3261x, under half the book.

Source CoinMarketCap, timestamped 2026-10-31T06:06:52Z, covering 100 of the 105 instruments bStocks reports. Instrument detail and quotes at /stocks/bstocks. Research notes, not investment advice. DYOR.