ALMA MATERMarket Intelligence
Cases — how the mechanisms work

Discipline, in the open.

These are records of how the system's risk and learning layers actually behaved — drawn from the production database. They illustrate mechanism, not prediction.

01 / 02
Case 01 · Timing layerKairos · AVGO · 2026-06-01 → 06-05 · paper

Four entry intents, four times unfilled.

−19.7% avoided · AVGO $485 high → $390 (06-05)

Over four consecutive sessions the decision layer produced an entry signal on AVGO — and none of them filled. Once a signal is generated, the execution layer puts it through a systematic review: both the overheat-protection layer and the risk-control layer must weigh in before an entry can proceed. Four times, the execution layer returned a decision to decline. Nothing was waived.

The mechanism, not the call

The point is not that the system “called” AVGO’s top. It is that a directional signal does not become a position on its own — it must first pass the execution layer’s review. On this name, the execution layer held its ground four sessions running, through both the run-up and the break that followed.

On the first three sessions, as AVGO pushed toward its $485 high, the overheat-protection layer flagged the entry — the name was running too hot to chase — and the execution layer declined it each time. A rising price was a reason to hold back, not to relent.

The fourth intent was different. It came after the break, on the way down — an attempt to catch the bottom rather than the top. This time the risk-control layer flagged it on different grounds: the risk-reward on offer no longer justified the entry. One name, two distinct reasons to stand aside — both caught by the execution layer’s review. AVGO then fell 19.7% from its high.

How it unfolded

06-01

AVGO near its high; the first entry intent formed and was downgraded by the overheat-protection layer. Too hot to chase.

06-02

Price higher still; a second intent, downgraded again. The climb was the reason to wait, not to relent.

06-03

The top at $485; a third intent, downgraded. The overheat-protection layer did not wear down as the price rose.

06-04

After the break, a fourth intent formed on the way down — an attempt to catch the bottom. Declined: the reward no longer covered the risk. A risk/reward call, not an overheat call.

Result check · 2026-06-05
−19.7% (AVGO $485 → $390) · four intents, zero fills · nothing was waived

What this does not prove

Honesty matters more than a clean story. This is a single, hindsight-favorable instance. The same risk-control layer could just as easily stop a strong name that keeps running — a false positive that costs a real gain. One success is not evidence a rule works. Effectiveness is measured over a large sample — how many losing entries were correctly stopped versus how many good ones were wrongly blocked — not by a single case. The layer lowers the odds of entering at the wrong moment; it does not eliminate loss. Its value is discipline, not foresight.

Case 02 · Timing layerKairos · META · 2026-07-06 → 07-13 · paper

Nine percent, while the market fell.

+9.07% realized · META entry 602.94 → exit 657.61

2026-07-06 → 07-13. The week the broad market drifted lower — SPY −0.28%, QQQ −1.53%. In the same tape, the system opened META and held it through its window.

The mechanism, not the call

This case is the mirror image of the first. Where AVGO was about a guard that stopped an entry, this is about an entry that held — through a week when the indices themselves lost ground.

The point is not that the system “picked” META. It’s that in a down week, with SPY and QQQ both negative, the system’s read on this one name diverged from the tape — and the divergence was expressed as a real, filled position, not a paper opinion.

How it unfolded

Entry

2026-07-06 — the system formed and filled an entry on META at 602.94. It cleared the same risk-control layer every entry clears; nothing was waived.

Hold

Over the following sessions the position was held with discipline — not shaken out by intraday noise, not overstayed. The exit system tracked it throughout.

Exit

2026-07-13 — the exit system took profit by rule at 657.61, locking in the gain. Realized: +$7,762.72, +9.07%.

Result check · 2026-07-13
+9.07% (META) vs −0.28% (SPY) · −1.53% (QQQ)

A single name, held through a down week, finished nine points ahead of the market it was drawn from. This is one outcome — a filled position that ran its course. It is not a claim that every entry does the same.