Three candidate indicators
New signals, published while they're still on the bench. Each one was backtested on the full 2020–2026 cycle and gets an honest passport — including the one that came back weak. These live boards read the current Binance-spot universe. Nothing here is promoted to the main Indicators hub yet.
These are experimental. Two passed the backtest and are usable as described; one is weak and shown as context only. Treat the live boards as research, not a buy list.
Leaders and Toxicity passed the backtest and graduated to the main hub: Relative Strength · Toxicity. Attention stayed weak, so it lives here as a context board.
loading live board…
3 · Attention / CT Heat WEAK · context only
what's popular in CT — context, not a signal
- On the archive the volume-attention proxy had an information coefficient of ≈0 and a 39% win-rate — most high-attention names bled.
- High attention meant higher variance (bigger tail both ways) and slightly deeper drawdowns, with no clean good-vs-toxic split.
- The live board adds CoinGecko trending for real search attention on top of the volume surge.
How to read it: popularity is a crowd-context layer, never a buy or sell trigger. Its real use is as a co-flag with Toxicity — a coin that's popular AND toxic is the classic distribution top.
Highest attention
loading…
How to read a row → a coin near the top with a CT badge and rVol 3.0: the crowd is piling in right now — treat it as context, cross-check Toxicity before ever calling it a buy.
Trending in CT right now
Honest limits
- Historical tendencies, not promises — the edge only shows over many trades.
- The exact numbers are noisier than they look (overlapping windows); trust the direction and the gaps, not the decimals.
- Dead/delisted coins aren't in the backtest, so real life is a bit worse, especially in risk-off.
- Attention is backtested on a volume proxy only — historical social data doesn't exist; the live CoinGecko layer is not itself backtested.
Full method and numbers are in the verification lab.