← back to the platform

Practise the rule on a coin that does not exist

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 day down day 50-day high the rule watches you 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.