Skip to main content

Why convergence beats single-source signals

One ping is noise. Many independent sources agreeing is signal.

bobby.engine · convergence · live000000
INPUTS · live chatter
echoAvalanche AM
echo from 3 unrelated chats
listRugfilter
safety pass, sellable
behOG Holders
cohort shift, not pump
behBase Insiders
$BRETT holders +18% in 4h
echoMemecoin AM
who else seeing this
convL2 Watch
6 sources agree
histSolana Pulse
historical match: 2024-Q3
GROUP CONVERSATIONSHOLDER BEHAVIORCROSS-COMMUNITY ECHOLISTING FOOTPRINTHISTORICAL CONTEXTCONVERGENCEOUTPUTS →
OUTPUT · alerts that earn the ping
$BRETTbehavior
holder cohort shift detected
$BONKlistings
listing footprint converging
Inputs: group conversations · holder behavior · cross-community echo · listing footprint · historical context.
Outputs: alerts that earn the ping.
[i] Five input streams. One convergence model. Behavior, never volume.
methodology

The single-source problem

A bot that pings on a single trade is a bot that pings on every manipulation. A bot that pings on a single mention is a bot that pings on every shill. A bot that pings on a single wallet move is a bot that pings on every wallet that wants to be seen.

The market figured this out years ago. Trust in single-ping bots collapsed because the pings stopped meaning anything. Members learned to ignore them. The bots kept pinging anyway.

Convergence is the answer

Bobby fires when multiple independent sources agree, weighted to be hard to game. The weighting is the part we do not explain. The convergence is the part we do.

What we deliberately exclude

Volume signals. The most manipulable input in crypto. Putting volume in the convergence model would be putting bait in a trust product. So we don't.

We watch behavior, not theatre.

Cross-community pattern matching

Coordinated accumulation gets detected before it surfaces publicly because the pattern shows up in multiple groups before it shows up on any chart. Years of data means historical context the rest of the market does not have. Even groups that no longer exist still contribute to pattern recognition.