LONA is an imaginary coin. Its price is generated in your
browser from a random seed. There is no exchange, no account, no money and no
way to deposit any — and nothing you do here is recorded anywhere.
What this is for. Not to make you good at trading. To show you, with
your own hands, why a boring mechanical rule beats an attentive human on a
chart — and then to show you that one run of anything proves nothing at
all, which is the single hardest lesson in this whole project.
How it works
The coin is fake. The physics are not. LONA's price is
generated from statistics measured on 46,779 real market-days of the
markets this rule actually trades: 7.72% daily volatility, extreme fat tails,
and — surprisingly — negative daily autocorrelation. Returns
mean-revert. What the rule eats is not trend persistence but a conditional
effect at the breakout itself: the day after a 50-day high returns
+0.902% against +0.060% on every other day. Nothing here was tuned
until the rule looked good.
You are shown one day at a time. On each day you may buy
(if flat) or sell (if long). You always trade the full stake — no
sizing decisions, so the only thing being tested is timing.
Beside you, invisibly, finding 008's rule plays the same
days: buy when the close is strictly above every one of the previous 50 closes,
hold exactly 5 days, then out. It never changes its mind. Both of you pay the
same 10bp round-trip fee.
Market:
LONA / USD · 1d · simulated
up daydown day50-day high the rule watchesyou were long
Your position
Order ticket
Finding 008’s rule
Your fills
The same rule, running for real
Everything above is a fake coin. This is not. It is the
live paper book from this project’s forward record — the identical
rule, on real Hyperliquid markets, written down once a day and
never revised afterwards. It reads whatever the platform last published,
so it is as current as the machine that keeps it.
Forward record · live
reading the published record…
The honest part
Whatever happened above, it was one season. So when a
season ends, this page silently replays the rule across 500 more randomly
generated seasons and shows you where a single run sits in that spread. That
picture is the reason this project pre-registers everything and refuses to
believe a good backtest.