The problem
A rising market makes everyone look like a genius
Screenshots of winners are free to produce and free to select. The comparison is what costs something.
What you see
The calls that worked, posted after they worked. Nobody screenshots the other ones.
What's missing
Whether the chain was up anyway. We score returns in excess of peer tokens over the same window, so a market-wide melt-up never reads as skill.
The uncomfortable finding
Past performance does not predict future performance here. First half against second half of the same account's history: rho −0.024. One account went +22.8% to −10.7%.
What does predict
Entry timing — rho −0.188, p<0.01. The later you are, the worse you do. Small effect, real effect, and the only one we found.
What you get
Three things, all of them measurements
No predictions, no scores you can buy, and no rankings built on a correlation we couldn't find.
Call history
Every post we have scored for that account, with performance measured against peers rather than against zero. The losers are in there too.
Entry timing
How early each call was relative to the move — the one component with proven predictive value, presented as the small effect it is.
Audience check
Whether the replies under a call are real people or wallet addresses. Useful the moment you are deciding how much a chorus of agreement is worth.
The numbers
Measured, not asserted
We would rather tell you that performance doesn't persist than sell you a leaderboard implying it does.
−0.024
skill persistence
first half vs second half, same account
−0.188
entry timing vs return
p<0.01, the one real predictor
31%
of token posts aren't calls
we classify before we score
$0
for a cached check
unlimited, no account
Questions