First time on the top traders desk?

The desk runs its own numbers and publishes its calls next to the account that trades them. The short route:

INDICATOR — CONTEXT

Attention / CT Heat CONTEXT ONLY

This board shows where the crowd is right now, which coins are pulling unusual volume and search interest. It is the one board on this hub that makes no forecast, and we publish it saying so.

The idea, and what it actually reads. The score is a crowd proxy: a coin's volume against its own 30-day normal, plus a bonus when it sits on CoinGecko's trending-search list and when an up-move comes with that volume. Posts on X are read too, through the desk's attention feed, but they are counted beside the score, not inside it: the X columns show mentions in the last 3-hour window, and a z-score once that ticker has 4 windows of its own history to be unusual against — 12 hours at this cadence. The columns disappear when the feed goes stale rather than showing an old count as current. Share of voice ranks the board: it is this coin's share of all engagement counted in the window — likes, reposts, quotes, replies and bookmarks, views excluded — and a dash means the coin was not in the window at all, which is not the same as zero. The CT tag marks a name the X window named. Published as context: a crowd can arrive before a move, after one, or instead of one.

loading live board…

Regime is risk-ON. Attention still forecasts nothing; read it for what you are walking into.

Attention / CT Heat — where the crowd is (live)

loading…

Attention volume + search interestX posts mentions in the last 3h window, shown while the feed is freshrVol volume vs its own 30d averageCT named in this X window

How to read a row. A high attention score says the crowd is here — more volume, more search, wider dispersion and slightly deeper drawdowns. It does not say the coin will rise: the measured information coefficient is about zero. Use it to know what you are walking into, not to decide.

What the backtest showed

The information coefficient is about zero and the board was right 39% of the time. As a predictor it does not work, and no amount of framing changes that.
High attention did come with wider dispersion and slightly deeper drawdowns — the crowd adds volatility, not direction.
There is no clean split between coins that went on to do well and coins that turned out toxic. Attention does not separate them.
The archive test used a volume proxy; the live board adds CoinGecko search interest on top, and that addition has not been tested.

Where it breaks: everywhere, if you read it as a forecast. It is published because a desk that hides the boards which did not work is showing you a filtered record.

How to use it: as the last check before acting on some other trigger, to know whether you are stepping into a crowded trade. Never as the reason for the trade.

All indicators Also in the lab: Attention (CT heat) →