$ Paper money loading the latest snapshot…

%

$10,000 · since launch

Just buying Bitcoin instead: 

How it works

A very patient robot

The strategy has historical backtests. They do not establish future returns. This paper track records decisions as they happen, including losses and gaps in service.

01
4h

It checks the market every 4 hours

No staring at screens, no panic. Six looks a day at the Bitcoin price — that’s the whole schedule.

02

It follows trends, with strict rules

The rules set a paper target: up to 100% long or 50% short. This track does not apply the old wallet or daily-loss gates.

03
···

When unsure, it sits out

The strategy can choose cash. If market data is missing, the record retains the last position until valid data returns.

04
=

This page keeps the score

Results follow saved decisions, with modeled fees and funding. Missed decisions during outages are not filled in afterward.

Lab notes

We publish the failures too

Most of what we test doesn’t work. Writing that down is the point — it’s how you know the scoreboard above is honest.

Did not pass8 September 2026

Does holding Bitcoin and adding exposure in uptrends help?

Strong older returns did not pass the final registered screen. Severe drawdowns and weaker recent comparisons close this strategy search.

Read the write-up →
Mixed result 8 September 2026

We sized Bitcoin positions by volatility. Did it help?

A controlled test links part of the apparent advantage to missing data. The strategy still faces deep drawdowns and unresolved data quality.

Read the write-up →
Mixed result 8 September 2026

The bot lagged Bitcoin. Did buying-pressure filters help?

One promising filter avoided a single losing trade. Nearby timing choices and multi-day pressure tests still failed to establish a dependable advantage over holding Bitcoin.

Read the write-up →
Negative result 5 July 2026

We ran the bot at 10× leverage. It liquidated 34 times.

Can you leverage the bot to get rich faster? We ran 11 years at 2x–10x, with real fees and funding, and tested stop-losses too. Leverage isn’t an edge — it’s a risk dial with a fuse on the end.

Read the write-up →
Mixed result 23 June 2026

Does Bitcoin really run on a 4-year clock? We counted the bars.

The “23-bar bottom” is wrong — the bear is ~12 months. The 47/35/12 cycle is real, but only because it’s the halving clock, only fits two cycles, and only looks clean if you delete the 2020 crash.

Read the write-up →
Negative result 21 June 2026

We tested the “follow the money” signals. One was backwards.

Negative funding, rising open interest, spot vs. futures — the flow-trading playbook, measured against years of real funding and open-interest data. The most-repeated signal pointed the wrong way.

Read the write-up →
Negative result 11 June 2026

We tested 20 candle patterns. The famous ones were all noise.

1,005 bullish engulfing candles, hammers, dojis, flags — measured against 11 years of data. The only real signal was the quietest thing on the chart.

Read the write-up →
Negative result 11 June 2026

We spent a day trying to beat our own bot. It won.

Four serious attempts to improve the strategy — cut costs, buy bottoms, fade the crowd. All four failed, including the one that looked like it doubled the money.

Read the write-up →
Method 11 June 2026

What 11 years of backtests can and can’t tell you

Why we trust the process but not any single month — and why this site shows drawdowns first.

Read the write-up →