Reading a BABA signal.

Anatomy of an alert

A BABA Premium signal that lands in your Telegram looks like this:

🤖 MD AI — TON short
Hyperliquid · $3 · 60s veto

TON -12.4% 7d, -5% 24h. Sustained downtrend,
not a fresh breakdown chase. Funding at -1.17% APR
means shorts are NOT overcrowded (longs paying shorts),
compatible with short. Reddit post confirms retail
caught long at highs — distribution signal.
ATR 1.44% manageable.

Entry $1.789 · SL $1.852 · TP $1.664
Voters: Sonnet PASS 0.72 · Grok PASS 0.68 · Gemini PASS 0.75

[✅ Approve & Fire] [⏭ Skip]

Every alert has seven fields. Internalizing them is the goal of this module.

1. Strategy + direction

The header tells you which pillar fired and which way: MD AI — TON short. MD AI is the directional perp pillar. "TON short" means betting TON's price will fall.

Possible strategies you'll see: MD AI (directional perps via the 3-vendor AI committee), ORACLE AI (Polymarket binary outcomes), PCE_v2 (high-conviction prediction-market mean-reversion), LP Agile (concentrated-liquidity rebalances). Each has a slightly different conviction profile.

2. Venue + size + execution mode

Hyperliquid · $3 · 60s veto. The venue tells you where to execute. The size is the recommended USD margin commitment. "60s veto" means the engine will auto-execute on its own wallet 60 seconds after this alert unless you tap Skip; if it's set to "manual" instead, you must tap Approve.

Default sizes are small on purpose — $3–$5 while track record is being built. You can size up once you've watched the engine fire 20+ signals on the same strategy and seen the win rate stabilize.

3. The thesis

The middle block is the AI's reasoning. It always starts with the price context (24h, 7d moves), names the technical pattern (downtrend, breakout, mean-reversion), checks funding (whether the trade is "chase-prone" or "fresh"), and references any news catalysts the context layer picked up.

The thesis is short on purpose. It's roughly what a sharp analyst would say in three sentences. If the thesis doesn't make sense to you — if you don't understand why funding being negative is bullish for a short — that's a signal you need Module 4 before you size up.

4. Entry, stop-loss, take-profit

Entry $1.789 · SL $1.852 · TP $1.664. These are the prices the engine calculated.

• Entry: the mark price at the moment the signal fired. By the time you tap Approve, the price will have moved slightly. Most signals tolerate 0.2–0.5% slippage; beyond that the engine flags it.
• SL (stop-loss): the price at which the engine will close the position to cap losses. On Hyperliquid this is placed as a venue-side conditional order — if you go to sleep, the SL still triggers.
• TP (take-profit): the price at which the engine closes the position to lock in gains. Also venue-side conditional.

The relationship between these three numbers is the risk-to-reward ratio. In the example: max loss = entry − SL = 3.5%, max gain = entry − TP = 7.0%. So R:R = 2:1. The engine targets R:R ≥ 1.5; setups that don't pass this filter never fire.

5. Voter status

Voters: Sonnet PASS 0.72 · Grok PASS 0.68 · Gemini PASS 0.75. The three AI judges' verdicts. For a signal to fire, at least 2 of the 3 must PASS at confidence ≥ 0.65.

When all three PASS (like the example), conviction is highest. When only two PASS and one BLOCKED, the trade still fires but you might want to size smaller. The voter line is your sanity check — if all three are at 0.65–0.7, this is a borderline trade; if all three are 0.8+, it's a high-conviction one.

When a signal arrives — a 10-second checklist

1. Read the thesis. Does it make sense given the market regime you've been watching? If yes → continue. If no → tap Skip.
2. Check the R:R. Is the max loss something you're comfortable with? At $3 size with a 3.5% SL, the worst case is −$0.11. Most people are.
3. Verify the venue. Is your account funded? If you've been at Hyperliquid all week, yes. If you haven't onboarded the venue, tap Skip.
4. Approve or veto. 60 seconds is plenty. The countdown is visible in the alert.
5. Move on. Once the trade is live, the engine manages SL/TP. You don't need to babysit it.

Common mistakes to avoid

• Overriding the size. The engine sized this at $3 because that's the per-pillar cap while you're learning. Bumping it to $50 because you "feel good about it" is how subscribers blow up.
• Removing the SL. Don't. The SL is the contract. If you removed it once and dodged a loss, that's survivor bias; the next one will be the liquidation.
• Trading the same signal twice. Once you've fired, that's the position. If you fire again on the same alert from a duplicate Telegram message, you've doubled your exposure unintentionally.
• Acting on stale alerts. Alerts older than 5 minutes are stale — the price may have moved past SL already. Tap Skip and wait for the next one.