Seven Weeks, Two Tapes.

Steady Edge Trading spent seven weeks heads-down building The Forge. The tape kept trading. Here is what it did, closed and counted, right up to the point where this week’s Sunday Macro Brief takes over.

Start with the number that tells you almost nothing. Across the seven weeks the Dow finished up 1.75 percent and the S&P 500 up 1.80 percent. The Nasdaq 100 finished down 1.31 percent. Seven weeks, and all three ended close to where they started.

Now the number that tells you everything. The Nasdaq 100 fell 8.83 percent into its July 29 low and rallied 10.64 percent off it into August 13. The Dow fell 1.98 percent and rallied 5.34 percent to its own high on August 5, or 4.35 percent measured to that same August 13 date. On the way down the Nasdaq fell more than four times as far as the Dow. Traders call that dispersion, which in plain English just means one index swung much harder than the other. A net number under two percent describes none of what a trader had to sit through.

Two things ran underneath all seven weeks, and neither was the Federal Reserve. One was oil. The other was chips.

Source: Yahoo Finance cash index closes (^DJI, ^GSPC, ^NDX, ^VIX). Weekly percentages measured close to close. Net measured from the July 10 close to the August 28 close. The VIX column shows each week's close, and the Net row its final reading against 15.03 at the start. Dealer gamma and IV Rank are not shown for these weeks because those readings are point-in-time and are not recoverable after the fact.

Crude ran from 71.41 dollars on July 10 to 92.19 on July 23, up 29 percent in nine sessions, while the Strait of Hormuz standoff stayed unresolved. Rising energy costs feed straight into the inflation outlook, and an inflation scare is hardest on the companies whose value sits furthest in the future. That is the tech trade.

Semiconductors took it worst. The Philadelphia Semiconductor Index fell 9.97 percent in the week of July 13 alone, and 4.78 percent of that came on the Monday, four sessions before any AI headline hit. Then on Friday July 17 the Chinese startup Moonshot AI released a model called Kimi K3 at the World Artificial Intelligence Conference in Shanghai, priced at roughly a third of the leading American model for the same output. The assumption that United States firms stay ahead by outspending everyone on computing power stopped looking safe. The selling was already running before Kimi K3 landed. The model gave it a second reason, not its first.

By July 29 the semiconductor index had lost 28.61 percent from its June 22 high. That is a bear market, which in plain English just means a fall of a fifth or more off the peak. The Dow does not hold those names. The Nasdaq 100 is built on them. That single sentence is the dispersion story, and most of the table above follows from that one fact.

Two scheduled events had landed in that same first week and neither one mattered. June CPI printed 3.5 percent on July 14, and the monthly reading fell 0.4 percent, its largest one-month drop since April 2020. Fed Chair Kevin Warsh gave his first semiannual testimony to Congress on July 14 and 15. The tape looked straight past both, because the selling was never about the data calendar at all.

July 29 was the worst day of the seven weeks and the low of the whole stretch. The Dow closed down 1,153.18 points, off 2.19 percent. The Nasdaq 100 closed at 27,192.31 and the semiconductor index at 10,447.49, both their lows. All three bottomed on the same afternoon.

Warsh's Federal Open Market Committee held the federal funds target range at 3.50 to 3.75 percent. A hold is usually the quiet outcome. Not this one. The vote was 9 to 3, and all three dissenters, Beth Hammack, Neel Kashkari and Lorie Logan, wanted a quarter-point increase. The statement pointed at supply shocks in energy, which is the polite way of saying crude had just run to 92 dollars.

The next session reversed the decline. Microsoft reported after the close on July 29 and rose 15.51 percent the following day on cloud growth strong enough to blunt the argument that cheap models had made American AI spending look wasteful. Taiwan Semiconductor added 7.64 percent the same day. The Nasdaq 100 rose 3.36 percent and the Dow added back 613.92 points, then 276.97 more on Friday, recovering 77 percent of the drop in two sessions. A week containing the worst day of the stretch still closed green.

The week of August 3 was the best of the seven. Dow up 2.96 percent, S&P up 3.58 percent, Nasdaq 100 up 5.12 percent. Three sessions did the work: the Dow added 693.38, then 907.47, then 263.24 points, closing August 5 at 54,349.12, its highest close of the stretch.

Watch the oil tape alongside it. Crude peaked at 92.19 on July 23 and fell to 75.22 by August 5, down 18 percent, with the two largest down days landing on July 27 and August 4. The energy supply shock the Fed pointed at on July 29 was already unwinding by the time it said so. The semiconductor index rallied 20.80 percent off its July 29 low into August 17.

Now the part worth studying. In the CFTC report dated August 4, leveraged funds held a net short position of 100,640 contracts in Nasdaq-100 futures, the largest of the period. Leveraged funds are the fast money, hedge funds and managed futures rather than pensions. Net short means they held that many more short contracts than long. They sat at maximum short on a day the Nasdaq 100 rose 3.32 percent.

Over the next three reports they bought back 56,324 contracts, cutting the short to 44,316 by August 25. That is the mechanical shape of a short squeeze, which in plain English just means sellers get forced to buy at higher prices and their buying pushes price higher still. Positioning did not cause the move. But 56,324 contracts of short covering is buying that has to happen, and it landed on a market that was already rising.

Source: CFTC Traders in Financial Futures, futures only, in contracts, net of long minus short. These are the e-mini and consolidated books, not the micro contracts. Positions are as of the prior Tuesday and are released Friday at 3:30 PM ET, so the reading always lags the tape. "Most Short" is that contract's largest net short reading across the reports cited here, some of which predate July 13, shown with its as-of date; "Bought Back" is the change from that reading to August 25. The Commitments of Traders report, or COT, is a positioning context tool, not a trade signal.

The Dow tells the quiet version: fast money went net long in late July and again in mid-August, and was back to a small short by the 25th. Asset managers, the slower money, cut their S&P net long from 992,729 contracts in late June to 929,159 by July 21, then rebuilt to 953,228 by August 25. Neither is a signal alone. Together they describe a market that de-risked into the break and re-engaged into the recovery that followed.

The week of August 10 is where the recovery topped out. The Dow slipped 0.56 percent while the S&P added 0.36 and the Nasdaq 100 another 1.09, and on August 13 the Nasdaq put in its highest close of the stretch at 30,084.50. The semiconductor index kept climbing for two more sessions and peaked on August 17. That is the high-water mark that everything after it is measured from.

One footnote that says more than it looks. Friday August 7 brought the July jobs report, and payrolls fell by 23,000 against an LSEG consensus of 80,000 added. The Dow gained 151.83 points that day. The week had already made its move without the data.

The July minutes were released Wednesday August 19 and read hawkish. The Dow fell 703.84 points on Thursday, recovered 517.80 on Friday and still closed the week down 0.85 percent at 53,277.01. The Nasdaq 100 lost 2.45 percent to 29,308.86, about 776 points below its August 13 high close. Crude had crept back to 87.06 dollars. Same news, and the tech-heavy index took it close to three times as hard. In July the gap had been wider still.

Then the last week did something worth noting. The three indexes drifted up, the Dow 0.53 percent, the S&P 0.49 percent and the Nasdaq 100 0.43 percent, while the semiconductor index fell another 2.31 percent. The engine that had driven both the break and the recovery was still falling, and the indexes rose without it.

Volatility never confirmed a fear event. The VIX ended the seven weeks at 14.43 against 15.03 at the start. The VXN, the same measure for the Nasdaq 100, ended at 19.92 against 24.89, its lowest reading of the whole stretch. Expected movement in tech was lower at the end than at the beginning, after all of that. Crude closed the window at 83.40 dollars, still well above where it started but 8.79 dollars off the July 23 high.

Nothing here required a forecast. The FOMC date was published in advance and the direction of the reaction was not knowable. Neither was a Chinese lab shipping a cheap model on a Friday in Shanghai, or Microsoft printing a number that undid three weeks of doubt in one session.

What was available the whole time was structure: trend, the levels that had held and the ones that had not, and a defined answer to where the idea stops being right. All of that was readable on July 28 without knowing what the Fed would do on July 29.

That is the argument for a rules-based system, and it is why the system SET teaches is built to read the same way on any liquid, trending instrument. TREPP stands for Trend, RSI, EMA, Pullback, Probability. Nothing in that is specific to the Dow, the Nasdaq or futures. The Nasdaq's 8.83 percent and the Dow's 1.98 percent were one set of conditions at two amplitudes, and a process that only survives on one of them is not a process, it is a preference. Which instrument is set up is a decision in its own right, and size belongs to the account, not to the conviction.

The honest counterpoint: a trading journal will also tell a trader whether they followed their rules, and a journal costs nothing. True, and it is the right place to start for anyone not ready to spend money. What a journal cannot do is put the same decision in front of a trader a hundred times in a week with the outcome still unknown.