Smart money vs retail crypto divergence log in Notion (BTC, ETH, SOL)

Build me an agent workflow that maintains a research log of smart money vs retail divergences on crypto perpetuals. Trigger is cron every 8 hours, aligned roughly with the standard perp funding intervals (00:00, 08:00, 16:00 UTC).

For each of BTC, ETH, and SOL on Binance, call CoinGlass "Get Global Long/Short Account Ratio History" (this is the retail-crowd LSR) and "Get Top Trader Long/Short Position Ratio History" (this is the smart-money LSR) for the last few snapshots at a reasonable interval like 4h. For each coin, compute the divergence as (top trader LSR minus global LSR) on each snapshot.

A signal fires for a coin when the most recent snapshot shows the absolute divergence is at least 0.25, AND the divergence has the same sign across at least two consecutive snapshots, so we are not logging noise from a single print. Direction is "smart-money-long-vs-retail-short" when top traders are net long while retail is net short, or the reverse.

When a signal fires, pull current price, funding rate, and open interest for the coin from CoinGlass "Get Futures Coins Markets" for context. Then write a one-sentence agent-authored read on what the setup implies (e.g. "Smart money is positioned long against a heavily short retail book on BTC, a classic short squeeze setup if price holds").

Before writing, dedupe against the existing log: for the coin, use Notion "Query a Data Source" on the user-configured database, sorted by timestamp descending, page size 1, to fetch the most recent row for that coin. If that row already has the same direction and the divergence has not changed materially (say, within 0.05), skip writing for this cycle. Otherwise, append a new row via Notion "Create a Page" with these properties: timestamp (now, ISO 8601), coin (BTC / ETH / SOL), direction (smart-money-long-vs-retail-short or smart-money-short-vs-retail-long), top trader LSR (number), global retail LSR (number), divergence gap (number, signed), price (number, USD), funding rate (number, percent), open interest (number, USD), and a short notes field with the one-sentence read.

If no coin clears the threshold this cycle, the agent should stay quiet and write nothing. Do not post to Slack, do not send email, do not create empty rows.

At setup time, ask the user for: the Notion database ID for the "Smart Money Signals" log (must already exist with the columns above and be shared with the Notion connection), the coin list (default BTC, ETH, SOL), the divergence threshold (default 0.25), the consecutive-snapshot requirement (default 2), and the exchange (default Binance).

Integrations: CoinGlass (long/short ratio history and futures coins markets) and Notion (query data source, create page).

Additional information

What does this prompt do?

Checks BTC, ETH, and SOL three times a day, aligned with funding intervals, for moments when top traders are leaning one way and the broader retail crowd is leaning the other way.
Only logs a setup when the gap is large enough to matter and shows up across at least two snapshots in a row, so noise stays out of your journal.
Pulls live price, funding rate, and open interest for context, then writes a one-sentence read on what the setup implies.
Appends each signal as a fresh row in your Notion database with timestamp, coin, direction, the two ratios, the gap, and the market context.
Skips writing anything if the same divergence is already the most recent row for that coin, so you do not get duplicates.

What do I need to use this?

A CoinGlass account with an active API plan, since the long short ratio history feeds are paid.
A Notion workspace with a database set up for the log, with columns for timestamp, coin, direction, top trader ratio, global retail ratio, gap, price, funding rate, open interest, and a notes field.
The Notion database shared with your General Input connection so the agent can read and write to it.

How can I customize it?

Change the coin list to whatever you actually trade, for example add BNB, AVAX, or a memecoin you watch.
Tighten or loosen the divergence threshold and the consecutive snapshot requirement to match how aggressive you want the journal to be.
Swap the exchange the ratios are pulled from, or compare two exchanges, if you trust one venue's crowd more than another.
Adjust the schedule, for example run every 4 hours instead of every 8 if you want a denser log.

Frequently asked questions

What does smart money vs retail divergence actually mean here?

Top traders on a futures exchange are usually the better informed side of the book, and the broader retail crowd often sits on the other side at the extremes. When those two groups disagree by a wide margin, traders watch it as a classic contrarian signal. This workflow surfaces those moments and logs them so you can study how they played out.

Will it spam my Notion database every 8 hours?

No. The agent only writes a row when a coin clears your divergence threshold across at least two consecutive snapshots, and it checks the most recent existing row for each coin first. If the same standing setup is already logged, it stays quiet.

Do I need to pay for CoinGlass?

Yes. The long short ratio history feeds require a paid CoinGlass API plan. Plans start around 29 dollars per month. A free CoinGlass account is not enough.

Can I run this on coins other than BTC, ETH, and SOL?

Yes. After clicking Use this prompt, just tell the agent which coins you want to track. The same logic works for any coin CoinGlass supports futures data for.

How is this different from the existing CoinGlass funding rate journal?

The funding rate journal snapshots every cycle no matter what, into Google Sheets. This one is event driven, only writes when a real divergence shows up, lives in Notion where you can annotate setups with your own commentary, and tracks position ratios instead of funding.