Start with the Fed, because it set the tone. The FOMC held at 3.50%–3.75% on Wednesday, July 29, on a 9–3 vote, with three officials dissenting in favor of a 25 basis point hike — the first time three have dissented in the same direction since September 2016. The Chair offered no forward guidance at all. The bond market answered immediately: the 30-year pushed to 5.21%–5.23%, a 19-year high; the 10-year ran to roughly 4.66%–4.69%, the highest since January 2025; and the 2-year fell to about 4.24%. That’s a bear steepener, with 2s10s out around +42 basis points. The odd part sits underneath — odds of a September hike actually fell from roughly 80% to 60%–65% even as the long end blew out. The front end priced patience while the back end priced something else entirely.

There are two defensible readings, and the split isn’t bull versus bear — it’s credibility versus sequencing. One view holds that a hawkish tone unaccompanied by a hawkish action is worth very little, and that the only reliable way to bring long rates down is to move short rates up. The other holds that this was intentional: a chair who has argued for years that a less transparent Fed is a healthier one is doing precisely what he said he would, allowing the long end to absorb some of the tightening and leaving distance between restating an inflation mandate and acting on it. The options market is worth watching here mostly because it just got caught. Going into the meeting, traders were loading up on TLT calls — a bet on falling yields — and had it exactly backwards. The flow has now flipped hard: puts outnumbered calls by more than seven to one in 10-year futures, and the five most active TLT contracts were all puts. The counterweight is straightforward — real, inflation-adjusted Treasury yields are the highest they have been in years, and buyers of past yield spikes have generally been paid for the patience.

The other pole of the week was genuinely strong. Amazon surged 14% on earnings, a day after Microsoft gained 15%. AWS grew 37%, its best in eighteen quarters, against Azure’s 43%, with Microsoft guiding for further acceleration and disclosing that paid Copilot seats jumped from 20 million to 30 million in a single quarter. Amazon’s total revenue grew 20% year over year — roughly $35 billion of incremental sales in the quarter alone — with operating margin at a record 13.7%, and AWS now representing about 60% of company operating income. The number that changed the conversation was the payback math: management indicated a data center investment recovers its cost in under three years on average, against depreciable lives measured in decades. Behind that sits a $496 billion AWS backlog plus roughly $225 billion in Trainium-related commitments. There was a strategic tell as well — the company is now signaling it intends to build its own frontier model, which would leave it holding custom silicon, cloud infrastructure and a model of its own, the same combination that belongs to the one hyperscaler whose stock has led for eighteen months. The companies that could not put a number on the return went the other way: a social platform that raised its spending floor while missing on earnings, and a hardware giant whose guidance is now being squeezed by memory prices.

The skeptical case deserves equal airtime. These results arrived from very different starting points — both winners had been dead money for the better part of a year, so the bar was set low, while the disappointments were already carrying elevated expectations. More to the point, nothing in this quarter resolves whether AI eventually funds itself. The buildout today is financed by equity issuance, debt, and a lattice of vendor arrangements, not by AI revenue. Estimates for total industry capex have climbed from roughly $800 billion at the start of the year to $900–950 billion now, with serious arguments for $1.6 trillion next year. Free cash flow is turning negative just as the ten-year climbs, which makes each successive financing round more expensive and lifts the hurdle every project has to clear. The backlogs being advertised are contracts, not cash received. And an increasing share of the financing is being done at the project level, secured against specific assets rather than the parent balance sheet — a structure that limits recourse, but also means what’s being underwritten is an asset return, not a platform return.

The rebuttal is equally concrete. The companies that just reported are generating record free cash flow, which is exactly what reduces the need to visit the capital markets at all, and the returns are finally showing up in reported numbers rather than in projections. The lending market for this paper has shown no sign of closing, and the beneficiaries reach well past semiconductors — into power generation, electrical equipment, industrials and services, several of which quietly had excellent weeks. The valuation argument lives in memory: Micron (MU), a leading supplier, trades near six times forward earnings with a roughly 14% free cash flow yield, sitting on about $100 billion of take-or-pay commitments through 2030 and $22 billion of cash already received, with a December anniversary that opens the door to buybacks. Against that, the same name is finishing the week down 10% and roughly $100 below its intraday high. Both are true at once.

The earnings scorecard deserves an asterisk most summaries left off. With 61% of the S&P 500 reported, 77% have beaten on the top line and 86% on the bottom line, both well above their five- and ten-year averages. Blended earnings growth is tracking at 47.4%, up from 38.0% a week ago and 23.2% at quarter-end, which would be the strongest quarter since Q2 2021, with blended revenue growth of 14.1%, the best since Q4 2021. But the headline is heavily distorted by two companies: one search giant’s GAAP result included a $98 billion gain, and one retailer’s included $53.4 billion of non-operating other income tied largely to an AI equity stake. Strip those out and blended growth falls to 28.8%, with the average earnings surprise dropping from 31.4% to 9.2%. Still an excellent quarter by any historical standard — just not a record-breaking one. That gap may explain a lot about the sour reactions to good prints: the tape appears to be pricing the adjusted number, not the headline.

None of the worst sessions this week came with a downgrade to growth. What they came with was forced selling. Margin calls have been firing across the system — 1.2 million Korean retail accounts liquidated two weeks ago, and this week an AI-focused hedge fund running extreme leverage on semiconductor and infrastructure names blew up and handed its equity book to a large market maker. Add six weeks of heavy stock supply that still has to be absorbed, and an August calendar that is historically among the thinnest liquidity stretches of the year. Sentiment is already washed out — AAII bulls at 29.6% against 42.3% bears, a spread of −12.8; the Fear & Greed gauge at 38; the VIX closed at 20.66 on FOMC day and drifted back near 16 by Friday. Yet trend-following models have quietly rebuilt long exposure, with the Nasdaq model at its highest since October. That combination is what makes the next headline count for more than it should, and there are two on the way: the quarterly refunding announcement on August 5 and the July jobs report on August 7.

Currencies produced the week’s other real event. Dollar-yen had been grinding toward 164 with the market openly discounting Japan’s intervention warnings, until Thursday’s New York session, when the Ministry of Finance moved and the yen ripped as much as 3.3% to the 158 handle. The more consequential detail was Washington’s posture — a US rate check on the pair and a Treasury Secretary describing the yen as “very undervalued.” It didn’t hold: by Friday the pair was back above 160, most of the move surrendered inside a day, even as the BOJ held at 1.00% on an 8–1 vote with one dissent for a hike and warned that core inflation likely runs above 2% from September. Korea produced the more interesting outcome, with Friday bringing an unprecedented trilateral joint intervention by Seoul, Washington and Tokyo. Dollar-won went from an intraday 1,440.6 to 1,418 before settling near 1,424 — a nine-month high for the won and its best month since March 2009, helped by expected ADR listing inflows, exporter repatriation for tax payments, and a pause in foreign equity selling. That changes the shape of the risk rather than its direction: an explicitly US-backed floor caps the left tail, which has been the single biggest deterrent to unhedged foreign money in Korea. The yen just demonstrated the limit of that logic — intervention gains get handed back when the rate differential hasn’t moved. The won now behaves like a managed level rather than a free float, with the defended zone appearing to sit somewhere between 1,420 and 1,440, and the durable catalyst sitting in Tokyo rather than Seoul.

One piece of breadth is being misread. The equal-weight S&P is up roughly 11.4% year to date against about 8.3% for the cap-weighted index, and the cause is subtraction rather than addition. Concentration finally cut the other way: the top ten names are around 40% of index market value and semiconductors alone are roughly 18% of the S&P, against about 3% for most of the index’s modern history — so when that cohort takes a 5% day, the cap-weighted index absorbs it at full weight and equal-weight barely registers it. Add a long end that is lifting the discount rate on long-duration growth faster than on the cash-generative middle of the index. What this is not is a small-cap signal. Every equal-weight member is still an S&P 500 company with investment-grade access to credit, while roughly 40% of the Russell 2000 is unprofitable and carries a materially higher share of floating-rate and near-dated debt. A real broadening would show small caps leading with the long end falling; right now we have two trades driven by different forces that happen to point the same way. Elsewhere, the US–Iran story remains the same on-again, off-again headline generator it’s been all month, and crude finished the week lower.

I’ve been studying charts for 32 years now, as you can see week in and week out when I give you levels that’s where price tends to settle for a fight between the bulls and bears.

SPX closed the week just above its 20 and 50-day moving average. Next supports 7,300 & 7,237 June low, FIB levels — horizontal lines — and the 100 & 200 DMA. The resistance can be the dotted trendlines and all-time high. I am noticing we made a short term lower high, but not a lower low. (My concern is the 10-year yield is rising.)

NASDAQ, I see a lower low, closed right around its last FIB level for the year, below the 20- and 50-day MA. Next supports can be found at the FIB levels — horizontal lines — and the 100 & 200 DMA. Resistance can be found at trendlines, last high and all-time high.

DOW found support right at the 50 DMA, closed the 20 DMA. Next support 50 DMA, followed by FIB levels; horizontal lines. Resistance; last high, all-time high, then 53,400 and again at 53,500 trendline.

RUT broke its all-time high a month ago, resistance trendline dating back from 2000 on May 5 around 2,850. The IWM ETF penetrated its resistance trendline and came back to close below that level, 296 area. I am following the small cap story to get an understanding regarding risk on or off. RUT did close below its 50 DMA. Next supports, trendlines and FIB #’s; horizontal lines. (If the 10-year yield is rising, that should support the bears.)

VIX found resistance at its last high from June 26.

CL — a few weeks ago I wrote CL is approaching a very interesting level at $62.16, which is a 68% FIB # and there is a trendline support at that level as well. The RSI came off oversold levels. Crude found a support at $67; I have some blue support line there, and for now that was the low. Next resistance can be found at trendlines, MA and FIB levels.

Gold found some support, closed above its 20-day MA but still below its 50 DMA. (3,516.1 is a 38% FIB retracement and 3,350 is a trendline going back 3 years.) Resistance can be found at moving averages and FIB horizontal lines.

Probably one of the most important instruments to watch — 3-year chart. Sometimes you have to look at the bigger picture to see the bigger picture. High yields and interest rates are bad for earnings growth, primarily because they dramatically increase corporate borrowing costs and interest expenses. What the Fed will do is the question on the table. Look for my next levels using the purple lines and moving averages. (Note: I am watching yields closely due to all the debt the hyperscalers are issuing as of late.)

3-year chart. Sometimes you have to look at the bigger picture to see the bigger picture. And now the trendlines can help you assess real risk better.

The front end of the curve went down after the Fed meeting; the 2-year yield went down this week and the DXY did the same.

The daily chart hit the level I gave almost exactly — $101.797. (Note the correlation between yields and the dollar, and the inverse correlation between the dollar and gold.)

A few weeks ago I posted a 3-year weekly chart; the last support dates back to 8/5/2024 @ $50,534. We broke the February low. I do see that BTC has been playing the FIB levels nicely. Past indications of level don’t automatically mean they will hold again. I will be keeping an eye out at the $29,682 level.

We closed just below the 200 DMA. A few weeks ago I wrote “we need to remember software was hit hard this year; if this market turns down then IGV is susceptible to further downside.” That seems to have happened — IGV corrected from 108 to 85. Next support and resistance are FIB #’s and MA.

My levels weren’t working for Silver; it’s mostly support and resistance lines per prior days and weeks. The prior all-time high for silver was in 2011 around $50.68; the 61% FIB retracement is around $47.31. If those areas don’t hold, next major support is around $27.

SOX closed below its 50 & 20 DMA. Next support MA & FIB levels. It seems the levels are working. Note: be cautious, this sector is volatile.

This week I am showing ORCL. Around February time I posted a video that ORCL is approaching a major trendline around the $136 and a 68% FIB level. The stock shot up to the $247 area and came all the way back. We broke the upward trendline (light blue dotted line); FIB level did not hold at $132.14 and $121.76. Next support after that level around the $98 and then $74. This is probably one of the major stocks to watch as it’s fully invested in the AI buildout — I watch the credit spreads, and so far they kept going up, which probably explains why the stock was making new lows.

NVDA found support a bit below its 200 DMA three weeks ago and again this week. There are a few support zones along the way; next support $181 area.

Monday (August 3) — Before the Open: MAR, TSN, HESM, TWST
After the Close: PLTR, GRAB, ON, SNAP, STRL, VRTX, BWXT, POWL, FANG, CLPT, ARE, OKE, CLX, WHR, VOV, WGS, TKO, WMB

Tuesday (August 4) — Before the Open: SPOT, PFE, CAT, MRK, CI, DOCN, HUT, BP, WIX, ET, ADM, HSBC, TSER, NRG, DJK, W, KMB, DD, CMI, MPC, LDOS, ROK, IDXX, BNTX, GWV, APO, TDG, LHS
After the Close: AMD, ANET, ZETA, ALAB, OPEN, TMDX, KTOS, UPST, BKNG, PINS, TOST, LCID, AMPX, AMGN, LUMN, CPNG, GILD, PRCT, PSKY, DVN, WYNN, EMR, APPS, HL, LSCC, EOG

Wednesday (August 5) — Before the Open: UBER, LLY, SHOP, NVO, DIS, CRCL, EOSE, FOX, FUBO, CVS, XRX, RIOT, KHC, OSS, GLXY, SN, SEDG, PSX, GFS, BAM, DT, NICE, EVGO, IRM
After the Close: IONQ, API, SOUN, MELI, SNDK, O, DUOL, WDC, SMR, ELF, AXON, JOBY, OXY, DASH, COKE, XYZ, FIG, UUU, LEU, SYM, RDW, EBAY, FLNC, BROS, HUBS, AMSC, HNST, ETSY, ALB, ROOT, TLN, ABCL, RUN, VIAV, FSLY, PAYC, MSI, Z, TTMI, DAVE, BYND, CDE, BMBL, EXPE, TPL, BLBD, AEVA, MCK, SITM, RIG, CHYM, RELY, SEI, ALL, ARRY, MNTK

Thursday (August 6) — Before the Open: CELH, QBTS, OSCR, CEG, DDOG, U, FISV, FOUR, WBD, ZTS, HWM, PTN, COP, HTZ, LNG, NVAX, UMAC, GCT, BKSY, ASPN, CNQ, PZZA, VISN, KDP, NTLA, PH, KVUE, PDYN, RL, VITL
After the Close: ABNB, TTD, RGTI, NET, MP, MARA, LYFT, DKNG, RCAT, AAOI, ROKU, SERV, CLSK, TEAM, TWLO, OUST, INOD, SEZL, RKT, MAIN, CART, TXRH, INDI, DOCS, DBX, ONTO, SG, SYNA, AKAM, AFL, PBR, WPM, PBR.A, PSIX, FROG, FIVN, KRMN, MITK, WLDN, RSG

Friday (August 7) — Before the Open: OKLO, VST, TTWO, UA, ACMR, WEN