How AQMath Works — What the Signals Mean and What You Execute

Published: 2026-08-07

Version: v2.1
Audience: every BLACK user, shown as a must-read before first use
Status: 📖 USER GUIDE — the single source of truth for the in-app explainer

1. What AQMath is (and is not)

AQMath is an analytical tool that produces trading signals. It is not a trading bot, it never connects to any exchange, and it never executes a trade. You place every order yourself, on your own exchange account, at your own discretion. You can ignore any signal at any time — nothing happens without your manual action.

There is no guarantee of profit, and losses are possible in every market regime. The numbers AQMath shows come from mathematical risk models validated in backtests; backtests are simulated history, not a promise about the future.

2. The two loops

The MACRO loop — KKT risk parity (slow, frozen)

When you save your portfolio for the first time, the engine computes KKT risk-parity weights for your token set: each token receives a share of capital proportional to how much risk (not price movement) it contributes, subject to structural limits (per-token caps, at most 60% of capital in risky assets, the rest treated as stable reserve).

These weights are then frozen. They are not recomputed when you press a button and not recomputed when prices move. The only thing that changes them is the macro loop: every 180 days the engine re-optimises once and freezes the new weights. You are notified when this happens.

The daily loop — Proteus (v18) Deleverage Shield (fast, frozen configuration)

Once per day, after market close, the engine runs the Proteus (v18) Deleverage Shield over your frozen weights. The shield is a regime modulator: it reads the drawdown and downside volatility of your basket and continuously scales your target exposure between fully invested and defensive (more reserve, less risk). Its drawdown trigger is scaled by your basket's own trailing volatility, so one configuration adapts to calm and wild baskets alike. The configuration is frozen at Proteus (v18) — validated out-of-sample on the exact production path and no longer tuned. Portfolios with too little history to calibrate the adaptive threshold fall back to its fixed-threshold predecessor, Aegis (v14), until they have warmed up.

3. What a signal means

After each daily run the engine compares the shield-adjusted target weights with the holdings you declared. If the difference for a token is material, you receive a signal of the form:

SELL 0.5 BTC ≈ 31,200 USD — your position is above target; the model suggests reducing it by roughly this amount.
BUY 120 SOL ≈ 18,400 USD — your position is below target; the model suggests adding roughly this amount.

The amounts are computed from the prices and the portfolio value at run time. They are guidance about size and direction, not limit orders: you decide how and when to execute, and slippage or fees are yours to manage.

Notifications are deliberately rare. You only receive a message when:

1. the Deleverage Shield turns ACTIVE (defensive mode engaged), 2. the Deleverage Shield turns back OFF, 3. a macro re-optimisation freezes new weights (every 180 days), 4. a material BUY/SELL rebalance is indicated, 5. the daily run fails — you are told the calculation did not run.

No message on ordinary days means the model sees nothing worth changing.

4. Your data

To personalise the calculation you declare which tokens you hold and in what amounts. These holdings are stored on our servers tied to the hash of your activation key, and are used only to compute your personal signals. You can update them any time, and deleting your key deletes your holdings, your consent records and your notification subscription. Full details are in the Privacy Policy.

5. Notifications

Notifications are delivered through our own first-party push server — no third-party provider sees your signals. You install the free ntfy app, add our server address and subscribe to your personal topic — no login or password is required, because the random 32-character topic name itself is your private address and is read-only. Turning notifications off closes the topic immediately; signals are still computed and logged in the app.

6. The honest summary

AQMath tells you what a disciplined risk model would trade today.
You trade it — manually, on your own exchange, if you choose to.
Weights are frozen at entry and re-optimised every 180 days; the shield modulates exposure daily; both run without your intervention.
There is no guarantee. The shield cuts drawdowns in backtests; markets can still lose value, including rapidly.