Copy Desk Floors Into Trading Agent Rules: A Desk Checklist for a Normal UTC Day

Copying desk floors into Trading agent rules means one thing: every bar you already apply by hand on ranks 51-200 gets written down as a setting before the agent is allowed to act on it. On a normal UTC day that is a short checklist. Sort by Breakout Score™, require climb, require confidence, and encode each refusal as a rule so the agent skips what you would skip. This post is the day-of version of score floors for Trading agent entries: less about why floors matter, more about what you type and when. Breakflare is an attention tracker; core research does not place trades or publish price targets, and floors only govern the optional agent.

Key Takeaways

• Write your hand-starring bars as agent rules after the UTC snapshot, never from a loud feed.
• Use four fields as the minimum: Breakout Score™ floor, climb requirement, confidence floor, band membership.
• If you cannot say a bar out loud on a quiet day, you cannot encode it on a loud one.
• Keep paper mode on until paper fills match what you would have starred by hand.
• Board first, agent second: the agent automates a habit, it does not create one.

What does it mean to copy desk floors into agent rules?

It means translating judgment into configuration without changing the judgment. Your desk already has an informal standard: some rows get starred, most get parked, a few get dropped. The agent has no standard until you give it one, and an agent with no floors will treat every row as eligible.

The copy step has three parts. First, state the bar in plain language (“I do not star a row unless climb is present and confidence is clear”). Second, map each clause to a field the agent exposes. Third, test the mapping against rows you already decided by hand. If the agent would have entered a row you parked, the rule is looser than your habit and needs tightening before anything else.

Breakflare is a daily hype-momentum attention tracker for CoinGecko ranks 51-200. It scores hype, 3d/7d momentum, Climb, Breakout Score™, and confidence. The scoring stack hub explains how those layers fit together; this post assumes you can read them and focuses on turning your reading into rules.

Why run the checklist on a normal UTC day?

A normal UTC day is the baseline your rules must survive. Nothing unusual happened overnight, no single sector is dominating the board, and the snapshot landed on schedule. Rules written on a calm day describe your real standard. Rules written during a mania day describe your mood.

The practical reason is calibration. On a calm day you can compare what the board offers against what you would star, and the gap between them is the information you need. If the board hands you twelve finalists and you would star two, your floors are doing less work than your eyes. Writing rules from that gap is how a desk standard becomes numeric.

Anchor the routine to the UTC snapshot so every comparison uses the same cut. The daily pipeline explains when the lock happens, and the live board versus UTC lock note covers what changes intraday. Floors should be written against the locked view, then watched against the live one.

What is the checklist, step by step?

Run these after the snapshot, in this order, and write down the answer to each before moving on.

1. Restrict the universe. Only CoinGecko ranks 51-200. Anything outside the band is a different research problem and should never reach the agent.
2. Sort by Breakout Score™. Use the Breakout Score™ sort and keep a short finalist list only. The shortlist size limit applies to rules too: if your floors clear dozens of rows, they are too loose.
3. Read momentum next to hype. Level without acceleration is a warning, so check 3d versus 7d momentum before you trust a loud row.
4. Require climb. Rank progress is the confirmation column. See the Climb score note for how it reads.
5. Require confidence. Coverage honesty comes last because it can veto everything above it. The confidence explainer covers what the column measures.
6. Decide star, watch, or drop. One word per finalist, with a one-line reason.
7. Encode the reasons. Each reason you typed in step 6 becomes a rule, so refusals stop being memory and start being configuration.

Step 7 is the whole job. Without it you have a good morning routine and an agent that ignores it.

How does each hand step map to an agent field?

Think in a mapping table rather than a settings page. The left column is what you do by hand; the right is the rule that makes the agent do the same.

What does a worked day look like?

Take an illustrative normal UTC day with three finalists. These are made-up rows to show the thinking, not real tickers or readings.

Row A has a strong Breakout Score™, clear climb, and confidence above your bar. You would star it by hand. The agent should be allowed to enter it, so the rules need to pass it. If your rules would block Row A, they are too tight for your own habit.

Row B has the loudest hype on the board but a flat rank. By hand you park it, which matches the high hype, no climb desk checklist. The agent must skip it, so the climb requirement has to be a gate, not a suggestion.

Row C shows climb with unremarkable hype. By hand you may keep it research-only, as described in the climb without hype desk standard. Decide once which way you lean and write it down. If climb alone is research-only on the desk, the agent must not get a loophole that promotes it.

The test of the day is simple: does the agent’s eligibility list equal your star list? Agreement means the copy worked. Disagreement means either your hand habit loosened or the rule is wrong, and you resolve it before the next snapshot.

Which fail modes should you name before you write the rules?

Name them out loud, because a named fail mode is easier to encode than a vague unease.

• Zero floors for activity. Lowering a bar to see the agent do something trains it on marginal rows.
• Hype-only eligibility. A floor on one layer lets a loud feed substitute for the composite.
• Paper looser than live. Paper then teaches a fill distribution you will not accept later.
• Silent edits. A floor that changes without a dated note cannot be reviewed after a bad week.
• Override creep. A one-day exception that renews daily is a lower floor in costume.

Every one of these is a process failure, not a settings failure. The fix is a written habit, not a cleverer number. For the specific case of thin coverage, the low-confidence skip note shows what a hard skip looks like, and confidence zero covers the edge where a row has no honest coverage at all.

The most useful property of a floor is that it makes disagreement visible. When your hand list and the agent list differ, you learn something about your own drift. A desk that never compares the two lists is not running a standard; it is running two opinions that happen to share a screen.

How do you test the rules before trusting them?

Use paper mode as the test bench. Trading agent starts in paper by default, which means the rules can be wrong without costing you anything, and that is exactly when to find out. Read how paper mode works and why paper then live is the intended path.

A clean test runs for a defined window, for example several UTC days, with one rule changed at a time. Compare three things against your hand journal: which rows the agent entered, which rows you would have starred, and which rows each side skipped. Count the mismatches and read each one. A mismatch tells you which side is looser.

What should the rules never replace?

Floors decide eligibility. They do not decide size, exit, or shutdown, and a perfect floor with no stop is still incomplete. Pair them with the controls in the guardrails set, including max hold and stop rules. Live fills use your Binance trade-only keys and PAYG credits at 0.5% of fill notional, so a loose floor has a direct cost beyond the position itself.

Floors also do not replace reading the board. If you only watch agent fills, your sense of what a careful promote looks like will fade, and the floors will drift with it. Keep the morning read alive, and keep pruning stale names using the prune rules.

How do you turn this into a written habit?

Board first, agent second. The board routine is the checklist above, run on every UTC day you research. The agent routine is a single page in your journal that holds the mapping table, the current floors with dates, and the override log.

A practical weekly rhythm:

1. After the snapshot on a normal day, run the seven steps and record star, watch, drop.
2. Compare your list to the agent’s eligibility list and log any mismatch.
3. Once a week, review the mismatch log. Change at most one floor, with a date and a reason.
4. Keep paper on until mismatches are rare and explainable. Test one floor change at a time so you know what helped.

What is the one-line desk standard?

Write the bar in words before the number, map it to a field, test it in paper, and change one floor at a time with a date. If your hand list and the agent list do not match, the stricter reading wins until they do.

If you want to run this against the live rank 51-200 board, the 14-day Plus trial includes live ranks, alerts, whales, trackers, and Trading agent in paper mode. You can start a Plus trial when you are ready to configure floors on the live product. Plus is $399/mo or $3,990/yr. There is no free delayed board.

Frequently asked questions

Is copying desk floors into Trading agent rules a trade signal?

No. Floors are eligibility filters for an optional agent, and the research board stays an attention tracker. Copying a floor does not say a row will climb; it says the row passed the same gates you would apply by hand.

What changes on a normal UTC day?

Very little, and that is the point. A calm day is the right time to write rules because the board is not pulling your judgment around. On louder days the checklist and the floors stay the same, and you resist editing them in the moment.

Should Trading agent use this as a hard skip?

Yes for the gates you already treat as refusals. If you would not star a row because climb is missing or confidence is below your bar, the agent should skip it, not shrink the position and enter anyway.

Does Plus change the columns involved?

No. Plus unlocks live ranks, alerts, whales, trackers, member chat, 90-day pick history, and Trading agent in paper mode. The columns and the floor logic are the same; Plus changes how fresh the board is when you read it.

How many floors are enough?

Start with four: band membership, Breakout Score™, climb, and confidence. Add a momentum rule only if you already apply one by hand. More fields are not more safety if you cannot explain each one in a sentence.

What if my hand list and the agent list keep disagreeing?

Treat it as a process bug. Either your starring loosened or the configuration drifted. Use the stricter reading until you can explain every difference, then change one floor with a dated note.

Related reading

• Score floors for Trading agent entries: the why behind each floor.
• Breakflare scoring stack explained: the layers your floors read.
• Daily rank 51-200 watchlist workflow: the board routine that comes first.
• What is a crypto signal confidence score?: the second lock in plain terms.