Semiconductor index closed Friday at 11,137.29, down 3.83%.…

The 4.4% Friday loss is not on the Philadelphia Semiconductor Index tape. Nasdaq’s page for 16 October 2026 shows a last of 11,137.29, down 443.93 points, or 3.83%. That was an eighth straight lower close. Ayushman Ojha’s Investing.com note, on Yahoo Finance the next morning, calls it the longest such streak since October 2008 and a 16% drop since the last advance on 6 October. Bloomberg’s Carmen Reinicke, at 1:44 p.m. New York time on Friday, had the same eighth session but said the index was on track to match the longest losing streak since 2006, with about $32 billion of gross paper gains for shorts over seven days on S3 Partners data. The missing fact is that 2006 and 2008 are not a rounding quarrel, and neither clock is the iShares ETF. SOXX closed Friday at $493.39, down 4.14%, with nine lower closes in the Nasdaq file, because 6 October was a six-cent down day. Wednesday’s 4.3% was 14 October, not Friday.

What just happened, and why the obvious reading is wrong

The obvious reading is a 2008-style break: eight down days, a crisis year, and a Friday percent rounded to 4.4%. The day range was 11,128.08 to 11,480.76, so the 11,137.29 close sat 9.21 points above the low. That is a weak finish, not a break through the session floor.

Adding the net change back to the last print is arithmetic, not a second downloaded close. 11,137.29 plus 443.93 is 11,581.22. Dividing 443.93 by 11,581.22 rounds to the published 3.83%. The summary panel also repeats 11,137.29 in the previous-close field. That is what a shut market does when Friday’s last is the latest close. It is not evidence that Thursday settled at 11,137.29.

The 4.4% lead is not on a fetched Friday page. The large midweek drop is Wednesday. Bloomberg’s takeaways on Subrat Patnaik’s story, updated 14 October at 8:39 p.m. UTC, say the benchmark fell 4.3% that day, that six sessions left the index down 12%, and that this shorter streak was the longest since May 2025. The takeaway puts the loss at $1.6 trillion. The headline on the same story says $1 trillion. JBizNews, on 16 October, said about $2 trillion. Those three value figures, and Wednesday’s 4.3%, are not Friday’s 3.83%.

Traders Daily lists the eight index sessions as 7, 8, 9, 12, 13, 14, 15, and 16 October, with the heavy days at 4.30% on 14 October, 3.83% on Friday, and 3.39% on 8 October. Neil Sethi, at 10:50 p.m. UTC on 16 October, rounds Friday to 3.8% and the week to minus 11.4%, the most since April 2025, and dates the run to 2008. Zerohedge, at 4:02 p.m. New York time, said the eighth day matched 2006.

Who is exposed

The index is a modified market-cap basket of 30 semiconductor companies. Nasdaq says it began on 1 December 1993 at a base value of 200.00 and split two-for-one on 24 July 1995. The live panel also lists a base value of 100.00. The holders are the cash basket, the ETF wrappers, the leveraged product, and the short book.

Ojha says all 30 names fell, with Broadcom, Micron, AMD, Intel, Lam Research, and KLA around 5% lower, and Arm about 7% lower. Traders Daily is finer: Micron 5.6%, AMD 6%, Broadcom 4.5%, Intel 5.8%. Do not average those cuts. Nvidia, the heaviest weight in that Traders Daily account, closed at $214.13, down 1.43%, and was still up about 15% in 2026. That close is one constituent, not this article’s price call. The basket fell 3.83% because the rest of the book fell harder.

Bloomberg’s afternoon story says all but 10 of the 30 names were down more than 10% since the selloff began. Coherent was off nearly 30%. Astera Labs, Macom, and Credo were each off more than 20% over eight days. Ojha says CoreWeave carries roughly $50 billion of debt. Traders Daily, citing the Q2 filing, puts operating lease liabilities near $16.3 billion and total debt near $35.6 billion. Both lines are fetched. They are not one number.

The ETF desk was adding. At 12:02 p.m. New York time, The Kobeissi Letter said SOXL had taken in $3.8 billion so far that week, on track to beat a June record of $3.1 billion, while the fund was down 35% over seven sessions. SMH was at $430 million and SOXX at $574 million. Those midday figures are not finished creations.

Reinicke’s $32 billion is gross paper gains over seven days, not cash taken out. Matthew Unterman, managing director at S3, is paraphrased there: part of the short interest is convertible-bond arbitrage, not a pure directional bet. CoreWeave shorts were on track that afternoon for more than $1 billion of mark-to-market gains, with five-year default swaps above a 50% chance. Later notes put Friday near 12% and five sessions near 25%, with yields above 15%. The CoreWeave year-end piece is a separate calculation.

What the tape actually shows

Two instruments, three clocks, and several loss windows are being sold as one headline. Index points stay in the rows that cite an index page. The chart is only SOXX.

Thursday’s ETF close was $514.71 and Friday’s was $493.39, a gap of $21.32. Dividing 21.32 by 514.71 gives 4.142%, which Nasdaq rounds to 4.14%. The Friday range was $492.83 to $509.60 on 14,673,630 shares, and the close was $0.56 above the low. The 31 March close was $328.66, so the rise to the 22 June high of $655.01 was 99.3%, not a double. Bloomberg’s index line, a double from the end of March plus about $6.5 trillion of market value, does not belong on this ETF.

Ojha says the index last advanced on 6 October, so its eight down days start on 7 October. In the SOXX file, 5 October was $589.51 and 6 October was $589.45, down $0.06. Then $582.82, $563.28, $559.40, $544.85, $543.90, $519.42, $514.71, and $493.39. That is nine lower ETF closes. Wednesday’s ETF drop, $543.90 to $519.42, is 4.50%, not the index’s 4.3%.

A Stocktwits note on Yahoo, at 2:13 a.m. UTC on 16 October, said a red SOXX close that day would tie a nine-day record from 2006. Friday did close red. JBizNews said ETFs were heading toward a ninth session, the longest since May 2006, still an intraday line about funds. Mike Zaccardi, CFA, CMT, posted at 6:50 a.m. New York time on Wednesday that SOXX and SMH were down six sessions through Tuesday. Citing Bluekurtic, he put the three-month rebound rate after similar six-day streaks at 16 of 17 since 2010, median gain 11.4%. That is not a nine-day study. Financial News on 17 October also printed a SOXX close of $493.67, against the Nasdaq row of $493.39, and defined its 11.6% headline as 20 trading days.

What the named firms have said, and what they have not

The fetched wires quote investors and short-interest analysts. They do not carry a Friday statement from Arm, Broadcom, Micron, AMD, Intel, Lam Research, or KLA answering the streak.

Dec Mullarkey of SLC Management told Bloomberg, “The run up in chips was very aggressive so invariably there was going to be some consolidation or pull back.” That is about the prior rally, the same run Ojha put at more than 60% through early October. It does not say 2008 has returned.

Joe Tigay, portfolio manager of the Rational Equity Armor Fund, told Bloomberg, “CoreWeave specifically is going to need a new story. It’s going to need time to digest this new reality.” Ojha’s list for that story is AI spending, financing costs, valuations, hotter US inflation, and reports that China’s CXMT progressed on advanced memory. He dates the first cracks to June. That is not a bank run.

Rob Haworth, senior investment strategist at U.S. Bank Wealth Management in Seattle, told the Globe and Mail, “There is some fear in the market.” He added, “High oil prices, in particular, tend to weigh on the price/earnings multiple of stocks, and so I think you have investors looking to be cautious into the weekend rather than take more risk.” TheStreet has November WTI up 3.5% and settling above $101. Oil up and chips down is a multiple story, not a funding stop. Jim Cramer of CNBC, as reported on the Yahoo page timestamped 16 October at 2:13 a.m. UTC, wrote, “But you have to find buyers of Coreweave bonds, Oracle common stock and SK Hynix, of course.” That line is from before the cash open.

The macro tension is real, and it is not 2008

A 2008-style break would be a broad funding stop. Ojha has the Nasdaq Composite down 2.1% and the S&P 500 down 1.1% on Friday. TheStreet adds the Russell 2000 off 1.7%, the Dow off 0.3%, tech off 2.5%, six of 11 sectors higher, and energy up 2.5% on the day, with five sectors set to finish the week up 2% or more. Traders Daily says the Dow still finished the week up 0.7%.

Volatility is a disagreement, so it stays one. TheStreet says the VIX jumped above 20 for the first time since July. Traders Daily says it ended Friday at 19.99, below 22. 19.99 is not above 20. This desk did not download a Cboe close and will not pick a winner.

Ojha cites hotter-than-expected US inflation and does not give a CPI percent in the fetched paragraphs. TheStreet says Fed Chair Warsh avoided guidance, October rate-hike expectations fell back, and the 10-year yield fell 16 basis points on the week. Reinicke’s 1:44 p.m. line had the S&P 500 “on track” for a 1.7% weekly drop and the Nasdaq 100 “set to shed 5%.” Daily Trader Insights later said the Nasdaq fell 4.5% for the week. Composite versus Nasdaq 100, and intraday versus close, should not be merged. How far the Nasdaq’s week went is a separate question.

Ojha’s rise of more than 60% through early October is not erased by a 16% giveback since 6 October. That is an air pocket in a crowded trade, not a demand break on these pages. Traders Daily locates the trigger in the financing wrapped around the chips. Daily Trader says Coherent lost nearly 30% even with quarterly revenue of $2.05 billion, up 34%.

What happens next

Monday 19 October ends the eight-day count if the SOX close is above 11,137.29. One higher close retires both the 2006 wire and the 2008 wire, because both are claims about consecutive lower closes. A lower close makes it nine. A premarket print is not the test.

Wednesday 21 October is the company check. Daily Trader Insights says Lam Research reports after the close. Lam was in Ojha’s roughly 5% group, and it sells the tools fabs buy. Orders that hold would support an air pocket. A cut to the tool outlook would move Tigay’s “new story” into the equipment backlog. See Intel, AMD in the same chip index, and Broadcom. None of those replaces Lam’s call.

Friday 23 October is the flow check. Kobeissi’s $3.8 billion into SOXL, $574 million into SOXX, and $430 million into SMH were midday figures, with SOXL already down 35% over seven sessions. The finished week says whether that bid survived the close. If creations reverse, the $32 billion short gain is no longer only a paper win on a crowded long. None of these dates is a new index target. This is not financial advice.

Frequently asked questions

Semiconductor index closed Friday at 11,137.29, down 3.83%. Is an eight-day air pocket a 2008-style break?

No. Friday’s official index print is an eighth lower close at 11,137.29, down 3.83%, and the rest of the market did not break with it. Ojha ties those eight days to October 2008. Bloomberg and Zerohedge tied the same length to 2006. An up week for the Dow and several higher sectors argue against a 2008 reading. The damage is the chip book and the financing around it.

Did the semiconductor index lose 4.4% on Friday?

No. Nasdaq shows minus 3.83% to 11,137.29 on 16 October 2026, and Ojha and Traders Daily use that Friday percent. TheStreet rounds the same session to 3.8%. The 4.4% lead does not appear on a fetched page as Friday’s index move. The large midweek drop is Wednesday 14 October, at 4.3% in Bloomberg’s takeaway and 4.30% in Traders Daily. That session is not Friday.

Was the losing streak the longest since 2006, 2008, or some other year?

The fetched pages disagree, and the disagreement is the point. Ojha on 17 October, TheStreet on Friday evening, and the Globe and Mail account say the eighth session is the longest since 2008. Reinicke at 1:44 p.m. New York time, and Zerohedge at 4:02 p.m., say 2006. Bloomberg’s Wednesday takeaway, for a six-day streak, said May 2025. This desk did not retrieve an Investor’s Business Daily page, so a 2018 comparison is not used here.

Is the Friday percent on SOXX the same 3.83%?

No. SOXX is the iShares Semiconductor ETF, and its Nasdaq close on 16 October 2026 was $493.39, down 4.14% from $514.71. The chart in this piece is that ETF. It does not plot 11,137.29. The file also shows nine lower closes, not eight, because 6 October was down six cents. A nine-day ETF streak and an eight-day index streak are different tapes. Do not put the ETF’s 4.14% into the index headline.

How much did shorts actually make?

Bloomberg, using S3 Partners, says about $32 billion of gross paper gains over seven days. Paper means mark-to-market, not cash that has been realized. Unterman told that story that part of the short interest is convertible-bond arbitrage. CoreWeave shorts were on track Friday afternoon for more than $1 billion of mark-to-market gains, with five-year default swaps above a 50% probability in the same piece. The $32 billion is not a Friday profit print for one fund.

What would end the air-pocket reading?

A higher SOX close on Monday 19 October ends the eight-day count by definition. It would not repair the 16% index drop since 6 October, or the ETF’s 24.67% drawdown from the 22 June file high of $655.01. The next fundamental check is Lam Research after the close on 21 October. The next positioning check is whether Friday 23 October’s finished flows still resemble the midday $3.8 billion into the 3x long fund.