12 AI agents sound like more coverage. In practice, they're eleven things nobody's coordinating and one you forgot you built. That's the problem this issue's AI x Trading section actually solves, not "more automation," but who reads the output before you do.
The same instinct shows up in the Skills & Prompts section right under it. The best of the 5 prompts we're sharing isn't the one that alerts you fastest. It's the one that knows when to say nothing.
Markets this week has a version of the same lesson, just more expensive if you get it wrong. The Crypto Clarity Act looked like a clean binary a week ago. It isn't anymore, 18 state attorneys general came out against it days after the market had already decided the outcome. And in the Discord, Miles is watching LIT break out into that same vote and still hasn't bought it. Chart looks good. He's waiting for the level to confirm first.
None of that is a coincidence we're forcing. It's just what happens when you build a system, or a trade, or a newsletter, around checking the signal instead of reacting to it.
Start with the 12 agents, and the 1 that actually matters.
🤖 AI x Trading
12 agents, and the 12th one is the only one that matters
Everyone building agent stacks right now is solving the wrong problem. They're asking what each bot should do. The question that decides whether the thing works is who reads the output.
Run 11 agents and you have 11 feeds. Each one competent, each one reporting independently, none of them aware the others exist. Two overlap. One drifts for a week before you notice. By Thursday you're skimming all of them at 6am, which is precisely the job you built them to take off you. That's not automation. That's twelve more tabs.
So the 12th slot doesn't get a job. It gets the other eleven. One coordinator, tracking what each agent is doing, catching overlap and drift, handing you a single combined status view instead of eleven check-ins. That's the difference between a swarm and a pile of assistants, and it's the line most people cut when they build this themselves because it feels like a wasted slot.
Two passes, never one
The second thing that breaks these setups is building before designing. You describe what you want, the model produces twelve roles, and nine of them are things you'd never open.
Split it. Pass one is an interview: what actually needs ongoing coverage, which tools you're already logged into, what keeps falling through. Only then does it propose twelve roles, with a hard rule that no two overlap. Pass two builds them, one at a time, each verified before the next goes live. Miles is blunt on that last part and he's right — activating twelve agents simultaneously with nothing verified is how a swarm goes sideways in an afternoon.
What a trader's roster looks like
Most published rosters are built for operators. Comms, Sales, Finance, Eng. Yours shouldn't be. Funding and OI drift on your open book. An unlock calendar that reports size as a percentage of float, not dollars. Governance forums for the three protocols you're actually exposed to. A reconciliation agent that tells you what you're holding versus what you think you're holding.
None of those find alpha. That's not the claim. The claim is that eleven things you keep meaning to check get checked before you wake up, and one agent hands you the summary.
The limitation, stated honestly
Nothing in this stack should be able to act. Anything that sends, publishes, buys, or takes an irreversible step holds for your approval, no exceptions, no low-risk carve-outs. Twelve agents also means twelve sets of tool logins — twelve blast radii. Audit access before it goes live.
And even at its best, a swarm reports. It doesn't weigh. It can't tell you what's already priced, who's positioned against it, or whether the account that broke the news has been wrong four times this year. That judgment still runs on people in a room talking to each other, which is what MHC members get every day.
Do this before Tuesday
Run pass one only. Build nothing. Answer the interview properly, look at the twelve roles it gives you, and cut to four — the four you'd read every single morning. If you can't name four, you don't have an agent problem, you have a process problem, and twelve bots will make it worse.
Twenty minutes. No keys handed over.
🛠️ AI Skills & Prompts
Take your Grok Bot to the next level using these prompts
5 prompts, adapted for a trader instead of an operator. Copy them as-is, swap the bracketed parts, run one per morning until it earns its slot.
1. Personal Chief of Staff → book brief
Every morning at 7am, check my open positions across [exchange/wallet]
and report in this order:
1. What needs a decision today — anything at or near a level I flagged
2. What moved against me overnight and by how much
3. What I said I'd close or trim and haven't
One message. No commentary on the market. If nothing needs a
decision, say "nothing today" and stop.2. Competitor watch → narrative watch
Watch [ticker/protocol] across X, news, governance forums and their
own docs. Only alert me when something material changed — funding,
unlocks, a governance vote, a team departure, a contract change.
Price moving is not material. Threads restating old information
are not material. If nothing changed, stay silent.3. Note-taker/analyser → pattern audit
Go through my last 30 days of [trade log/journal/notes]. Find the
mistake I keep repeating — not the biggest loss, the most frequent
error. Tell me what it's costing me and what it's blocking.
Be specific. If I keep sizing up after a win, say that.4. "Sparring" partner → pre-trade objection
Before I enter this trade, argue the strongest case against it.
Include the boring bear case, not just the tail risk.
Then tell me which single objection actually lands, and what price
or data would prove it right.5. Brainstorming partner → backtest stress test
Take this rough strategy and turn it into a working [script/sheet].
Then show me where it breaks — which market condition, which input,
which edge case — before you tell me it works.The pattern underneath all five
Give it a schedule, a clear threshold for staying quiet, and permission to disagree with you. Miss any one of those and the bot turns into noise you stop reading within a week.
Schedule — an agent you have to prompt is a chatbot. The value is that it fires at 7am whether you remember or not.
Threshold for silence — "if nothing changed, say nothing" is the highest-leverage line on this list. Most agent stacks die from being too loud, not too quiet.
Permission to disagree — prompts 4 and 5 are the ones people skip because the output is uncomfortable. They're also the only two that have ever stopped a bad entry or a broken model before it shipped.
One warning on prompt 3
It will tell you something you don't want to hear. Run it weekly and log the output. If the same error surfaces three weeks running, that's not an AI insight, that's a process problem the bot happened to find first.
Do this
Pick one, not five. Run the book brief for five days and check whether you actually open it. If you don't, the problem was never the tooling.
Where this stops: a sparring partner argues against your trade using what's public. It doesn't know three people in the room already ran this setup last month and got chopped up. That part isn't promptable — it's in the Discord.
📅 Upcoming Events
Date | Event | Why you care |
|---|---|---|
Tue | Crypto Clarity Act — Senate vote | Senate Republicans' reported "final" draft, said to include a Trump-backed ethics provision. Needs 60 votes to advance. The whip count isn't independently confirmed as of writing, treat this as on-the-calendar, not in-the-bag. Clears and it's a real multi-quarter catalyst for the sector. Fails and nothing changes, since it was never priced as a certainty. |
Wed | FOMC rate decision | Lands the day after Tuesday's vote fallout is still being digested. Recent Fed prints have moved crypto more than most macro data this cycle, worth attention even outside the usual macro crowd. No strong positioning read on this one specifically, flagged because skipping it has been more costly lately than usual. |
Wed | Strategic Bitcoin Reserve (H.R. 8957) — House Financial Services Committee consideration | Bill to formalize a federal Bitcoin reserve using existing Fed resources rather than new appropriation. Committee consideration is a procedural step, not a floor vote, so this doesn't pass or fail Wednesday, it either advances toward one or stalls in committee. A 3rd Wednesday catalyst on top of the Fed and Arc, worth tracking regardless of immediate price impact given what it would signal about federal accumulation. |
Wed | Arc — public mainnet launch | Same day as the Fed decision and the Bitcoin Reserve committee session. On the radar because Bonk's team has been signal-boosting it. The launchpad space is crowded and this isn't a high-conviction catalyst, worth a small opportunistic look if onchain activity holds into launch, not a thesis to size around. |
What this means
Four catalysts now, three of them landing on the same day. Wednesday was already a hard day to read cleanly, a Fed decision, Tuesday's regulatory fallout, and a mainnet launch all competing for the same attention. Add a congressional committee taking up a federal Bitcoin reserve bill the same day and Wednesday stops being 1 signal and becomes 4 overlapping ones, none of which will be cleanly separable in the price action. A rally on Wednesday could be the Fed, the Clarity vote's outcome, the Reserve bill advancing, or none of them and just Arc noise, and the market won't sort that out cleanly until well after the fact.
Size Wednesday like a day with no clean read on causation, because there genuinely won't be one. If anything, this addition raises the case for staying small into Wednesday rather than sizing up on any single headline that crosses that day.
📊 Markets
All eyes on the Clarity Act
A week ago this was a straightforward setup: Senate GOP drops a final draft, adds a Trump-backed ethics provision, cloture vote Tuesday, needs 60. Clean binary, clean date. It isn't that anymore.
What happened
The ethics provision landed well initially, read by some as Trump working with Democrats to get the bill over the line, which is the kind of signal that usually firms up a vote count. That read didn't hold. Sentiment on whether Clarity actually has 60 votes has since gone soft, and a second, separate story has attached itself to the same week: a coalition of 18 state attorneys general, led by New York's Letitia James, has come out publicly urging Congress to reject the bill as written. James's line was blunt: "As written, the Clarity Act would embolden scammers." The coalition's specific objection is that the bill as drafted could let federal rules override state registration and enforcement powers, and they want that language fixed before anything advances.
That's a real objection from people with actual power to slow this down, not noise. It's also the reason the vote count has gotten softer rather than firmer since the ethics news, and it's worth being clear about that link rather than treating the two stories as unrelated.
Why this is different from how the market's been trading it
Most of the sector has been pricing Clarity as the catalyst. It might not be. SEC and CFTC both look set on pushing forward with clearer crypto rules regardless of whether this specific bill clears the Senate, and SEC innovation exemptions are reportedly on a separate track that could land Friday or early next week whether or not Clarity gets its 60 votes. If that's right, the bill passing is a confidence signal on top of a direction of travel that's already set, not the thing that sets it. The market treating Tuesday as make-or-break is arguably mispricing what Tuesday actually decides.
How to position
Not a directly tradeable event on its own, and nothing here is a reason to build a position around Tuesday specifically.
High conviction: none. This isn't a setup with a clean trigger, it's a data point that shifts how much confidence you put behind the broader regulatory-clarity thesis.
Medium conviction: if you're already long the "US regulatory overhang lifts" thesis through majors or infrastructure names, this is a week to hold that thesis rather than add to it. The AG intervention is a reason for patience, not a reason to exit.
Low conviction / watch only: anything sized specifically around a Tuesday pass or fail. The vote count is soft enough right now that a position built purely on the binary is a coin flip dressed up as a thesis.
The risk
The specific way this breaks: a failed cloture vote gets read by the market as "regulatory clarity is dead," when the more accurate read, if the SEC and CFTC keep moving independently, is that the direction hasn't changed at all. That's a reflexivity problem. A no vote could trigger a real, ugly, short-term drawdown in anything trading on the regulatory-clarity narrative, entirely disconnected from whether the underlying regulatory picture actually got worse. Thin, headline-driven liquidity around a Tuesday miss is exactly the environment where that kind of overreaction happens and doesn't correct for a few days.
What to watch
Whether the AG coalition's objection gets addressed in the bill's language before Tuesday, or gets ignored and forces a delay.
The actual cloture result Tuesday: pass, fail, or delayed.
Whether SEC innovation exemptions land Friday or early next week regardless of the vote outcome. This is the one that tells you whether the "direction of travel is set independently" read is right.
Whether a failed vote produces a sharp, short-lived drawdown that recovers within days, which would confirm the reflexivity risk above rather than a genuine repricing.
💼 MHC Trades: A few setups we're watching
Miles, $LIT, 17:30
He has been watching LIT show strength into the Clarity vote. Chart's got a bull flag forming, trying to confirm a breakout. Could be early positioning on the Lighter regulatory narrative, could be nothing.
He's not in it. Wants a retest of the parallel channel first before adding exposure ahead of the event. That's the discipline: a decent-looking chart into a binary catalyst still isn't a reason to size in without the level confirming first.
Fabian D, $MET
Holding his bag through the dip below $0.20, and looking to add lower. His case isn't vibes: MET and RAY are pulling near-identical 30-day revenue ($2.3-2.6m), but MET trades at roughly a third of RAY's market cap. He argues that gap is buyback policy, not fundamentals, and thinks the market's underpricing MET until Meteora turns discretionary buybacks into a programmatic one.
The discipline is in the number, not the conviction. He's sized around a specific revenue-to-marketcap read (23% of current cap), not around "this looks cheap."
Paradise, $PONS
Calling PONS attractive on 3 things lining up at once: a 50% pullback from highs, the 0.5 psychological level, and the first retest of the daily "noodle." Sell-off came as mindshare rotated to STONK, but he doesn't read that as the end of the PONS thesis.
The discipline is waiting for 3 separate signals to stack in the same zone rather than calling a level on 1 of them alone.
Morin, $PUMP (Spot + Perp), 16:03
A fully conditional plan, not a live position. Wants capitulation into the $0.0032 daily demand zone before longing the perp, with a stop at $0.0029 if it's wrong. Spot side is a DCA around the range low, no attempt at the exact bottom.
The discipline is the structure itself: entry, invalidation, and size all set before price gets there, so there's no decision left to make emotionally in the moment.
These are positions and updates our analysts posted in the room, not recommendations. Levels and reasoning are theirs.
The room's been right lately, too. For example, CrossfitPiano flagged MET and ZEC as working the same week Fabian and Paradise called them, and members have posted real fills off these setups, including a +183% long on PUMPFUN that scooped Fabian's call within a day of it going up in the channel.
Every one of these was visible in Discord before it showed up in this email. Join MHC and you're reading the setup, not the recap.
Closing Out
Free Skool Community
For more behind-the-scenes, including AI prompts, workflows, and research, join Miles’ free Skool community here:
