The week two shorts carried what the index longs gave back.

The week opened with a stop. Monday's first evaluation, an EURUSD short into the New York morning, closed at -1R by design at 14:34 UTC. Seven minutes later the NAS100 desk bought a pullback into broken resistance and clawed back +0.49R of it. That pairing — a clean loss and a modest win inside the same quarter hour — set the tone for a week the desk finished at +1.19R across eight trades, five of them winners. The shape of the week is worth more than its total. Every one of the eight entries was some form of retracement: pullbacks to broken resistance, shorts into failed reclaims, buys at prior highs. The ones that paid shared a property the ones that lost did not — the level under test had already been broken and defended once before we arrived. That single distinction cost us exactly 2R on US500 and earned it back everywhere else. Through Jul 13, 2026, the system stands at +28.48R YTD across 144 published trades at a 60.42% win rate. On a $100,000 simulated account risking 2% per trade, that is $156,978.99 — the running evidence that weeks like this one, unremarkable on their own, are what the compounding is made of.

Act I — Monday pays for its own lesson

The EURUSD short at 14:34 UTC was the week's first trade and its first loss. A New York AM VWAP rejection graded C+, it never found the follow-through the read required and the stop did its job at -1R. Seven minutes later, at 14:41, the NAS100 desk entered a pullback long into broken resistance — the same species of trade the rest of the week would keep offering — and took +0.49R at TP1. The desk ended Monday down half an R, with the useful part of the day being the contrast itself: the losing entry tested a level that had merely paused; the winning one tested a level that had already broken.

Act II — Two shorts, two full runs

Tuesday and Wednesday were the week's engine. At 14:33 UTC Tuesday the GBPUSD desk shorted a bearish retracement inside the NY overlap; the ledger banked +1.07R at TP1 and the position worked all the way through TP3. Wednesday at 14:10 the US30 desk shorted a failed reclaim bounce and produced a near-copy of the result: +0.96R banked at TP1, TP3 tagged. Two different instruments, one identical logic — price had broken, attempted to reclaim, failed, and the retracement into that failure was the entry. By Wednesday's close the week sat at +1.52R cumulative and the equity curve's high-water mark, $103,032.57.

Act III — The index-long split

Thursday and Friday sent every index long to the same test and graded them differently. The US500 desk bought a pullback at yesterday's high on Thursday (14:06 UTC) and a pullback to the prior day high at the 7,550 zone on Friday (14:15) — the second read graded B, better than the first — and both stopped at -1R. Between them, the EURUSD desk recovered +0.67R on a pullback long into a breakout zone, and Friday afternoon the US30 desk closed the week with its second winner, a +1.01R pullback continuation long graded B+, the week's highest-graded trade. The curve finished at $102,393.68: +1.19R, five wins, three losses.

Pattern of the week

The week had one pattern and eight expressions of it: the retracement entry. Every trade — long or short, FX or index — entered on a pullback into prior structure. What separated the column of winners from the column of losers was not direction, instrument, or grade. It was whether the level being retested had already proven itself.

The five winners all entered where structure had already broken and been defended: broken resistance on NAS100, a failed reclaim on US30, a completed breakdown on GBPUSD, a breakout zone on EURUSD, a confirmed continuation on US30 again. The three losers all entered where the level's authority was still hypothetical — Monday's EURUSD short into a VWAP touch that had not rejected yet, and two US500 longs at prior highs that had never once been defended from above. Proven levels paid +3.19R combined (TP1 baseline). Aspirational ones cost -3R. That is the whole week in two sentences, and it is not a coincidence we intend to ignore.

Decision highlights

The GBPUSD short entered Tuesday at 14:33 UTC and the ledger banked TP1's +1.07R, but the position was allowed to keep working through TP2 and TP3 rather than being flattened into the first bounce. The read — a bearish retracement inside the NY overlap after a completed breakdown — stayed valid the entire way down, and the exit discipline matched the entry discipline.

The US30 desk shorted Wednesday's failed reclaim and bought Friday's pullback continuation: opposite directions, forty-eight hours apart, both winners. There is no contradiction in that pair. It is the same evaluation loop reading two different tapes, without carrying Wednesday's bearish conclusion into Friday's bullish structure.

After Thursday's US500 pullback long stopped at -1R, Friday's evaluation graded a nearly identical setup at the 7,550 zone a B — cleaner structure, same thesis — and took it again. It stopped again. Fixed sizing meant the second opinion cost exactly what the first did, and the willingness to re-engage a valid read without revenge-sizing it is the judgment worth keeping, even on a week where the read was wrong twice.

Loss worth learning from

The week's teachable loss is really a pair. On Thursday at 14:06 UTC the US500 desk bought a pullback at yesterday's high, framed as an opening-range breakout continuation. It stopped at -1R. On Friday at 14:15 it bought a pullback to the prior day high at the 7,550 zone — a cleaner version of the same idea, graded B where Thursday's was C+ — and stopped again.

The common failure was not the grade and not the timing. Both entries asked a prior session high to act as support the first time it was ever tested from above. Compare that with the week's winners, every one of which entered at a level that had already been broken and defended at least once. A prior high with no defense history is a line on a chart, not a floor; the market probed straight through both of them. The stops were correct, the sizing was correct, and the combined cost was contained to -2R — but the entry criterion that let both trades through is the same criterion, and it fired twice in two days. That regularity is what moves it from bad luck to feedback.

From the desk

A week like this one is easy to under-read. Plus 1.19R across eight trades is not a headline; it is a Tuesday-through-Friday of the system doing what it is designed to do, including the part where it pays for two wrong opinions at exactly -1R each and keeps evaluating.

The number worth sitting with is the running one. Through Jul 13, 2026, the ledger shows +28.48R since the Jan 12 inception. On a $100,000 simulated account risking a fixed 2% per trade, that is $156,978.99 static. Let the same sequence of trades compound — each position sized at 2% of the account as it actually grew — and the figure is $171,192.60. The $14,213 gap between those two numbers was not produced by better trades. It was produced by the same trades under disciplined sizing, which is the least glamorous edge in trading and the most reliable one we own.

That is why weeks like this get published in full: the losses at their real size, the winners with their TP levels disclosed, the tuning candidates named before they are adopted. If you are evaluating this system, the honest pitch is not the +1.07R short — it is that the desk logged eight evaluations this week and will log eight more next week, and the arithmetic of small verified edges is doing the rest.

— The SkyAnalyst Team

What we're tuning

One adjustment earns a review from this window: how the confluence math weights defense history on index pullback longs. Both US500 stops entered at prior highs that had never been retested; the week's five winners all entered at levels with at least one prior defense. If the evaluation loop had required that single property, the week's worst two trades would not have been taken and nothing else would have changed.

Two trades is a sample, not a verdict, so nothing structural moves this week. The candidate change goes on watch: if untested-prior-high entries keep underperforming defended-level entries at this spread, the threshold for the former rises. That is how the system tunes — from published evidence, at the speed of evidence.