The Mathematics of Money
A fishing trip, an adoptive grandpa, and the reason a bankroll grows: Kelly's 1956 paper read as Shannon's communication model, all the way to G = H(X) − H(X|Y).
Football · Probability · Information theory
You should not trust me.
Betting is an information game. You are trying to be on the right side of the asymmetry — and a tipster asking for your trust is the wrong kind of asymmetry entirely. So this is a blog that shows the working: the models, the mathematics, and a way to check that none of it was edited after the fact.
Under fair odds, the growth of your bankroll is the difference between the uncertainty the market has about an event and the uncertainty that remains given your model.
Gmax = H(X) − H(X|Y) — Kelly, 1956
The arrangement
I cannot prove I have an edge. I can prove I am not lying about one.
-
01
A manifest, before kickoff
Every morning at 10:30 CET, the games and markets that have been bet on are published to a public repository — the shape of the position, not the picks.
-
02
A signed fingerprint
Alongside it, one SHA-256 hash of the file holding the picks, the probabilities and the stakes, signed with a private key.
-
03
The picks, the next day
Recompute the hash, check the signature. Change a single comma after the fact and the verification fails, loudly.
-
04
A past that cannot be edited
Each day carries the hash of the day before, so one head hash covers the whole history. Delete a bad week and the head stops matching the one pinned in the channel bio.
Check it in one tap on the verification
page, hand the repository to an LLM
and ask whether the code does what it claims, or clone it and run verify.py yourself.
None of these asks you to trust me.
Latest