How a stacked-trend model profited as CRWV slid lower

CRWV traded for less than a year of saved data and finished the window lower, and the angle of interest is how a model built on stacked trend reads pulled a gain from a stock that drifted down. What you are looking at replays those fixed rules across CRWV's stored history; it is a backtest, not a forecast or a recommendation, and with a short live record there is not enough yet to grade it.

No trade leaves the model without a verdict first. It reads CRWV's direction up a ladder of look-backs, the rungs running from quick to slow, and acts only once the bulk of those rungs lean the same way. Rungs leaning higher send it long; rungs leaning lower send it short. The higher lean kept recurring even as CRWV gave ground, leaving 61 long positions among the 79 trades. The book posted a 58.23% win rate across that set, and at one position size the outcomes accumulated to +212.28%.

How the model is built, end to end.

The reading the model depends on is how well CRWV's stacked trend reads agree. When the short, medium, and longer spans disagree, the model treats the move as unsettled and waits; when most of them stack the same way, it reads a direction worth taking. The spans are measured against CRWV's own behaviour, so agreement is judged relative to how this stock typically trends. A stack that points in different directions keeps the model flat; a stack that lines up lets a trade form.

A stack that agrees is a green light to look, not to fire. The model still wants a separate direction check to confirm before it acts, which keeps it from chasing a span that lines up briefly and then breaks. When the stack and the check agree it opens a position with a stop and a target set up front. After that the trade settles in one of three ways, and on CRWV the profit target did the most work, closing slightly more positions than the protective stop or the settling reading.

The illustration above runs the model across CRWV end to end: the stack of trend reads, the agreement they must reach, the confirming check, and the long or short it produces. None of the steps is elaborate; the model's character is its insistence that the stack line up and the check pass before any capital moves. Its lineage belongs in the frame as well, it appeared as one candidate among many in an automated search and survived only by clearing backtest and walk-forward checks, the road from idea to deployment behind every figure shown.

At a glance

These figures come from a backtest of the model on CRWV, scored against fixed acceptance gates, not from a live track record.

Walk-forward verification

A trade walked through

The walked example is a short held about three days. CRWV's trend reads stacked to the downside near 117 dollars, the direction check confirmed, and the model sold; it rode the slide and closed near 111 dollars for +5.03%. It is a tidy example of the model's down-side at work, in once the stack agreed lower, out with a contained gain as the move ran.

What the full trade record shows

Across its 79 CRWV trades the model won 46 and lost 33. The exits were close to balanced at the top: 31 reached the profit target, 30 were stopped out, and 18 closed when agreement settled back.

A near-even split between targets and stops fits a model trading a volatile, sliding stock: roughly as many positions reach their target as get knocked out, while the rest are closed as the stack stops agreeing. The book wins by a small edge of targets over stops rather than by avoiding losses.

The biggest winner was a short that ran for +29.73% over about seven days in November 2025, entered near 115 dollars and closed near 81 as the settling reading carried it past a hard target. The model's best work came on the short side, where an agreed down-stack tracked CRWV's decline cleanly.

The worst trade was a long stopped out for -15.39% inside a single day in late February 2026, entered near 99 dollars; that loss anchors the equity curve's 20.28% drawdown, the deepest of the three gates that bite.

No single trade carries the record, and with CRWV ending the window lower, the lesson of the book is that the model's edge was its two-sided reach, profiting from a falling stock while a passive hold lost, even as the drawdown showed how much its long lean can cost.

How does this compare to just holding CRWV

Over the same window the model was tested on, simply buying CRWV and holding it would have lost money. Setting the two side by side shows whether the rule earned its keep, and here it clearly did, turning a profit on a stock that fell, helped by its ability to short. The tiles below put numbers on the gap.

How well does the model reproduce its tape?

Walk-forward verification checks whether the saved rule path reproduces the expected signal behaviour on held-out timestamps it was not built on. It is a consistency and replay-integrity test, not proof the model will make money live. A clean reproduction means the deployed rules behave like the studied rules; it says nothing about whether CRWV will keep moving the way it did across the short test window.

In live trading the model has been quiet so far, with too few signals to line up against the backtest. Until more live trades accumulate, the backtest is the only evidence there is, and it should be read as exactly that, a study of how the rules behaved on saved history.

When this approach fails

The model's sharpest costs come straight from its long lean. Its deepest trade was a long near 99 dollars that CRWV dropped through fast, stopping out for -15.39% in a single day in late February 2026, and the equity curve's worst stretch reached 20.28%. Because it bought into a stock that was trending down, an up-stack that proved wrong cut quickly, and on a name this volatile the move could run against the position further than the reads expected before the stop did its job.

Three things are worth watching if this ever trades at size. The first is drawdown: the worst stretch in the backtest was about 20.28%, the deepest of the gate misses, so losses can land hard and fast. The second is the long lean on a stock that fell, which leaves the model exposed when an up-stack reverses. The third is the distance between live and backtested behaviour, sharper here because the test window is short, and the first sign the agreement reads are no longer marking the same moves.

Risk and honest limits

On this run the model's automated checks logged a caution rather than a clean pass. It is one more reason to read everything here as backtest evidence about the rules, not a verdict on the stock.

Lifecycle