Weekly Recap: The Week the Pullback Paid, Then Charged Tuition

The week had four distinct days and one recurring idea. Monday opened with a stop and recovered it by dinner. Tuesday ground out two index wins. Wednesday put four winners on the board in thirty minutes. Thursday took three of those chips back in a single morning, and then the desk went flat into the holiday weekend. The ledger closed at +1.48R net on eleven trades, seven wins against four losses, a 63.6% week that earned its result early and defended it late. Through Jul 5, 2026, the system has banked +27.29R from its Jan 12 inception. A $100,000 simulated account risking 2% per trade sits at $154,585 on the static ledger, and the weekly number above is measured the same conservative way every week is: TP1-baseline, where winners are credited at the first target and every loss is a full -1R. What follows is the whole week: every trade in the index below, the per-instrument breakdown, the win of the week, and an honest teardown of the Thursday morning that made the final number smaller than the Wednesday desk thought it would be.

Act one: the grind

Monday and Tuesday were the week at its most ordinary. A NAS100 washout long stopped out at the open of the week for -1R, and the system's response was characteristically without memory: it kept scoring tapes. A US30 pullback long off the VWAP and Fibonacci confluence recovered the loss within hours, and Tuesday added two more index wins, a NAS100 long off session support and a US30 continuation off post-data support. Four trades in, the week stood at +1.54R and nothing dramatic had happened, which is what most profitable weeks look like from the inside.

Act two: the burst

Wednesday, July 1, was thirty minutes of everything working. Between 14:38 and 15:08 UTC the desk opened four positions on four instruments: a bear-flag short on NAS100, a pullback continuation long on US30, a fib-exhaustion short on EURUSD, and a breakout-retest long on GBPUSD. Long the Dow while short the Nasdaq, long the pound while short the euro, and every one of them a winner. The day banked +2.95R, with two trades running their full ladders overnight and the Nasdaq short landing as the win of the week at +1.29R.

Act three: the tuition bill

Thursday reversed the mood in fifty-two minutes. Three longs went on between 14:24 and 15:16 UTC, a US500 retest of the prior day's high, a EURUSD dip into fib support, and a US30 pullback into the 50 to 61.8% retrace, and all three stopped out. Each loss was the budgeted -1R and nothing more, but together they were 3R of give-back on what was structurally one idea: buy the morning pullback. The week ended there. Friday belonged to the holiday, and the desk carried nothing into it.

Pattern of the week

The recurring setup was the pullback to a pre-named level, worn six different ways. Monday's washout bounce, the VWAP and Fibonacci confluences on US30, the session-support long on NAS100, Wednesday's breakout retest on cable and fib exhaustion on the euro, even the bear-flag short, which is a pullback read in the other direction. In every case the system named the level first and let the market decide whether to deliver it.

The honest half of the observation is that the same shape produced the losses. Thursday's three stops were pullback entries too, structurally identical to the ones that paid all week. The pattern did not stop working because it was wrong; it stopped working because Thursday's tape stopped honoring retracement levels, and the only way to know that in real time is to pay a defined price for the information. The week's edge came from the same place its losses did, which is exactly what a real edge looks like.

Decision highlights

The clearest judgment of the week was Wednesday's refusal to hold a house view. Within thirty minutes the book was long the Dow and short the Nasdaq, long the pound and short the euro, because four separate analyses read four separate tapes and none of them was asked to agree with the others. All four paid, and the point would survive even if they had not: the system prices instruments, not moods.

The second was the NAS100 short's sizing decision. The instrument's Trend read was bearish while the Macro read leaned bull, and instead of resolving the argument the system priced it, docking the setup score, restricting the playbook to VWAP mean-reversion, and cutting allowable risk to a 0.5 to 0.75% band. The half-believed trade became the win of the week, and the sizing discipline is what made it safe to take at all.

The third was Thursday morning, and it cuts the other way. The system entered three pullback longs inside fifty-two minutes, each individually qualified, collectively a single 3R bet on one thesis about one morning's tape. Every stop was honored and the damage was exactly what the risk framework budgeted, but the clustering itself is a judgment worth interrogating, and we do, further down.

Loss worth learning from

Thursday's give-back deserves its own section, not because the losses were unusual but because their arrangement was.

What was right

Each entry was individually defensible: a prior-day-high retest on US500, a 61.8% fib support on EURUSD, a 50 to 61.8% retrace on US30, all textbook pullback locations with defined invalidations. Every stop was placed where the thesis died and every stop was honored for exactly -1R. There was no averaging down, no widened stops, no revenge entry after the first loss. The risk framework did its entire job.

What was wrong

The book bought the same idea three times. Three longs, three pullback entries, three index-or-dollar-adjacent expressions of one belief: that Thursday morning's dip was a retracement rather than a turn. When that belief failed, it failed everywhere at once, and a morning that could have cost -1R cost -3R. Individually qualified is not the same as collectively diversified, and the fifty-two minute window made the correlation nearly total.

What we would do again

Take each trade, at that price, with that stop. The read on any single chart was reasonable, and a system that refuses every pullback after two losses is curve-fitting to its own morning. The part worth changing is not the entries; it is the aggregate exposure math while they stack, which is where the tuning note below picks up.

From the desk

From the desk

A week that ends green after a -3R morning is not a story about luck. It is arithmetic doing what it was designed to do: seven wins collected at conservative TP1 accounting, four losses capped at exactly -1R each, and enough breadth across instruments that no single tape could take the week down alone.

The running ledger tells the same story at scale. From the Jan 12 inception through Jul 5, the system has banked +27.29R across 136 trades at a 60.3% win rate. On a $100,000 simulated account risking 2% per trade, that is $154,585 held statically, and $167,370 if each trade's risk is compounded on the growing balance instead. The gap between those two figures is the quiet argument for disciplined sizing: the same reads, the same stops, the same R, and the difference comes entirely from how consistently the risk math is applied. We will take that kind of edge over a hot week every time, including this one, which was both.

What we're tuning

The tuning question the week leaves behind is portfolio-level correlation during entry clusters. The confluence gate evaluates each instrument alone, and on Thursday it approved three same-direction pullback entries inside an hour without anything in the stack pricing the fact that they shared a thesis. We are examining a session-level exposure check that would treat stacked same-direction index entries as one growing position for sizing purposes, so the third expression of an idea pays a higher admission price than the first. No change ships until it survives backtesting; the note here is the commitment to look.