Why a disorder-regime model traded AAOI and missed the run

AAOI was one of the great runs of the period, a more than twelvefold climb, and a disorder-regime model that only acts when it judges the tape orderly captured a fraction of it and stood aside for the rest. This is a backtest, a fixed rule set replayed over saved AAOI history, not a forecast or a recommendation, and the live sample is still too short to judge.

Disorder is what the model measures. It gauges how chaotic a monitored AAOI reading has become against its own past and buys only when that chaos collapses to an extreme, a window it treats as an unusually readable, orderly regime. Selling short is not in its vocabulary; every one of its 56 positions was a long. It was right on 57% of them and reached +164.97% by accumulating each result at one steady size.

How the model is built, end to end.

The model watches one monitored AAOI reading and asks a single question about it: how disordered is it right now compared with its own recent past? It turns that into a score of how far from normal the disorder sits. A middle reading keeps the model flat. Only when disorder drops to a statistical extreme, the reading settling into an unusually orderly pattern, does a trade become possible.

When the disorder reading collapses to that orderly extreme, the model opens a long and only a long; it has no short side. The position carries a protective stop and a profit target, but most trades end when the orderly regime breaks down and the reading normalizes, closing the trade on its own signal. The model is built to lean on calm, readable stretches and to stand aside when the reading turns chaotic again.

The illustration follows the model end to end on AAOI: the monitored reading, the measure of its disorder, the test for an orderly extreme, and the long order that goes out. Each step is plain; the model's character is in trading only when it judges the tape readable, and never selling short.

At a glance

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

Walk-forward verification

A trade walked through

A clean example is a long from December 2025. The disorder reading collapsed to an orderly extreme near 26 dollars, the model bought, and the position ran for five days to its profit target near 35 dollars. The trade returned +30.68%, one of the model's biggest, and it shows the design working exactly as intended: buy the orderly stretch, ride it to the target.

What the full trade record shows

Across its 56 AAOI trades the model went long every time, winning 32 and losing 24. Most positions closed on their own signal: 44 of the 56 exited when the reading normalized, with 5 stop-outs, 4 take-profits, and 3 time-limit closes.

That is a one-directional, signal-driven system: it buys when its disorder reading collapses to an orderly extreme and steps aside when order breaks down, rarely waiting for a hard target or stop.

The three biggest winners were all longs that reached their target: +30.78% in January 2026, +30.68% in December 2025, and +30.39% over three weeks in September 2024 from a low base. The slowest winner still added +18.19% before the holding limit closed it.

The worst trade was a long held for three weeks while AAOI fell, drifting to its protective stop for -17.85%; the disorder reading stayed favorable long after the price had turned.

The model captured a handful of clean advances, but with AAOI up more than twelvefold over the window, the lesson of the record is everything it left on the table by waiting for order.

How does this compare to just holding AAOI

Held outright, AAOI was a runaway. The model underperformed a plain hold by a wide margin, and that gap is the honest center of the article: a rule that waits for a calm, readable tape cannot keep pace with a near-vertical melt-up. The tiles below size the gap.

How well does the model reproduce its tape?

Walk-forward verification checks whether the saved rule path reproduces the expected signal behavior 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 does not say AAOI's disorder will keep marking the same entries.

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.

When this approach fails

The model's failures come from reading order in the wrong place. Its central risk is that the disorder signal turns favorable while the price is already rolling over, so the model buys into a decline and holds, as it did for three weeks into the -17.85% loss in late 2025. Because it only goes long, it has no answer to a sustained downtrend except to stand aside, and a stock as violent as AAOI can fall faster than the reading updates. The flip side of its selectivity is the obvious one: in a stock that more than twelvefolded, every stretch it judged too disorderly to trade was upside it never touched.

Three things are worth watching if this ever trades at size. The first is drawdown: the worst stretch in the backtest was about eighteen percent, on a single long held too long. The second is the long-only exposure, because the model has no way to profit from or hedge a sustained decline. The third is the distance between live and backtested behavior, the earliest sign the reading is no longer marking the same regimes.

Risk and honest limits

Treat everything here as backtest evidence about how the rules behaved on saved AAOI history, not a verdict on where the stock goes next.

Lifecycle