From soft launch to settled desk: SkyAnalyst's 2026 so far

This is the year so far, counted the only way we count: settled trades, TP1-baseline R, every one of them published. Through July 2, 2026, from the January 12 inception, the system has banked +29.15R YTD across 139 trades at a 60.4% win rate. A $100,000 simulated account at 2 percent risk per trade sits at $158,308.55 on the static path, or $173,776.63 on the compounded path. Those are the headline numbers, but the number that matters more is the shape behind them. 2026 did not arrive as a finished desk. It arrived as a soft launch in the middle of January, ramped through a staggered first quarter, consolidated around a primary execution model at the end of March, and settled into a steady cadence by mid-year. This recap is the story of that arc, told through the trades that drew it. The trade-by-trade ledger sits below in the index; this is the year around it.

Act one — the soft launch: January into February

The record starts on January 12, not January 1, and the distinction is the honest one. The desk opened with a single instrument live and the rest of the roster still in internal validation, so January is a short month by design rather than a thin month by accident. The first published entry was a NAS100 pullback long on the fifteenth, written up in full in its [January case study](/en/blog/nas100-long-pullback-primary-01-15-2026). Volume was deliberately low while the engine proved it could size, hold, and exit under real fills.

February is where the sample begins. Half the roster came live in production, activity multiplied, and for the first time multiple instrument-traders competed on overlapping setups and reasoned to different conclusions. That is also where the year's first real drawdown printed, a cluster of index longs that stopped inside the first two weeks. It was the expected cost of running a young system into live tape, and it is torn down in the loss section below.

Act two — full force and the discipline pivot: March into April

By March every instrument-trader was live and the trade count stepped up hard. March carried the most entries of any month in the first quarter and the widest spread of outcomes, which is exactly what a full roster in a mixed regime looks like. The [early-March NAS100 breakout continuation](/en/blog/nas100-long-breakout-continuation-03-04-2026) is a clean example of the setup that would become the desk's bread and butter for the rest of the year.

The pivot came at the end of the quarter. Across the regime shifts of the first quarter, one execution engine held the most consistent discipline, and toward the end of March the desk consolidated around it as the primary execution model. This was not a marketing decision; it was a ledger decision. From that point the win rate stopped swinging and settled into the high-50s to low-60s band it has held since. April was the first month that read like a system rather than a launch: volume stable, discipline visible, R-multiple positive.

Act three — the settled desk: May into June

The back half of the window is the part that looks like a track record instead of a ramp. May was the desk's strongest month of the year by volume and by R, and June followed it with the highest single-month R of the year. The setups repeated, the sizing stayed flat, and the equity curve stopped looking like a startup and started looking like a desk. The [late-June NAS100 short](/en/blog/nas100-short-bear-flag-breakdown-06-22-2026) that booked the year's best single result, +3.26R, and the [US30 support-reclaim long](/en/blog/us30-long-pullback-support-reclaim-06-24-2026) a few sessions later are the shape of the settled desk: the same handful of reads, executed the same way, over and over.

Pattern of the week

The pattern that defines 2026 is not a setup. It is a system-design choice that shows up in the numbers: the desk trades a basket, not a conviction. Six instrument-traders ran all year, each gated by the same confluence and risk thresholds, and the year's return is the sum of six independent edges rather than one hero call. That is why a minus 1.14R stretch on USDJPY never threatened the year: NAS100, EURUSD, US30, and US500 were each carrying their own positive line at the same time.

The second pattern is confluence discipline under regime change. The first quarter moved through a trending open, a choppy middle, and a volatility expansion, and the entries that survived were consistently the ones that waited for structure to confirm before sizing. The losers, concentrated early, were the ones that entered on a premise the tape had not yet printed. That single distinction, confirmation before size, is the thread that runs from the February drawdown through the June winners.

The third pattern is restraint. The desk averaged roughly five to six trades a week across the year and never manufactured activity to fill a calendar. Weeks with nothing that cleared the gate produced nothing, and the gate was never lowered to generate content. A system with an audience has a standing temptation to trade for the feed; this one does not have that setting, and the flat weeks are part of why the positive ones held.

Decision highlights

The end-of-March consolidation around a single primary execution model was the year's defining call. It was made on the ledger, not on preference: across the first quarter's regime shifts, one engine held the most consistent discipline, so it earned the majority of the execution load. The win rate steadied from that point forward.

Keeping USDJPY small was a discipline win disguised as an omission. The pair never cleared the desk's confluence bar cleanly, so it drew only 6 entries all year and cost only 1.14R. A system that chased activity would have forced size into a read that was not there; this one let a weak instrument stay a small line item.

Holding the basket flat through the February drawdown, rather than cutting the instruments that were losing, is why the year compounded. The early index stops were treated as independent events, sizing stayed level, and the same instruments that stopped out in February went on to carry the back half of the year.

Loss worth learning from

The year's most instructive stretch is not a single trade; it is the February index drawdown, and it teaches more than any one loss could. In the first two weeks of the month a cluster of US30 longs stopped for a flat 1R each, and a NAS100 short followed them into the red on the thirteenth. Coming weeks into a young live system, it was the moment the record could have wobbled.

It did not, and the reason is the lesson. Every one of those stops was a flat 1R, placed at real structure and sized off the same fixed risk as every winner. There was no revenge sizing after the cluster, no fear discount on the next entries, and no quiet decision to bench the instruments that were losing. The desk treated each trade as statistically independent of the last, and the same US30 and NAS100 traders that printed the February trough went on to book +6.27R and +8.71R for the year. A drawdown absorbed at flat size is a cost; a drawdown that triggers a sizing spiral is a wound. This one stayed a cost.

From the desk

A $100,000 simulated account at 2 percent risk per trade sits at $158,308.55 on the static path through July 2, 2026, and at $173,776.63 on the compounded path. The compounded figure runs about $15,468 ahead of static, and that gap is the receipt for disciplined sizing: sizing each trade off a growing balance across 139 trades, rather than a fixed one, is what turns the same +29.15R into a larger dollar line. The distance between the two paths is not luck; it is the arithmetic of consistent risk applied to a positive edge.

From here, the mandate does not change. The desk keeps trading the basket, keeps sizing flat, keeps publishing every settled trade, and keeps the flat weeks flat. The record is now mirrored independently on MyFxbook from mid-year forward, so the curve is auditable outside our own numbers. The next monthly recap will extend this line by one month; the next system tune will be argued in public before it ships; and the next case study will show the reasoning behind the next trade, win or lose. Six months in, the year reads exactly like the process that produced it: unhurried, diversified, and fully on the record.

— The SkyAnalyst Team

What we're tuning

Two threads dominated the tuning queue this year. The first was confirmation discipline: the concentration of losers in premises that had not yet printed pushed a standing bias toward entries that wait for structure to confirm before sizing. That bias is visible in the back-half win rate, and it stays the desk's first filter into the third quarter.

The second is a stop-and-re-entry cooldown, floated in the weekly losses reports, that would downgrade a same-instrument, same-direction re-entry inside a short window of a stop. It has already met a test case it failed, a winner it would have flagged, so it is not shipping until it clears a high bar on the full-year back-test. We publish why a rule ships or dies rather than quietly changing the engine, and this one is still on trial.