This retrospective was planned to cover seven days. It covers three, because that is how long the run lasted before it was closed — the full story is at the bottom of this post. What got logged: BTC bouncing around a range, the regime filter checked every morning, and absolutely nothing done about it. The paper account ended exactly where it started: $10,000, 100% cash.
Sounds like failure. Isn't. Here's why.
The score for days 1-3: flat, because the trend filter is working
The first three days of this experiment were BTC being indecisive between $69k and $71k. Not exciting. Exactly what the system was built to ignore.
What the market actually did: chop beneath the 50-day average
BTC started the week recovering from a drop off its ~$76k high. Day 1 it was sitting at $70,416. Not destroyed, not recovered, just... there. Like it forgot what it was doing.
Day 2 ticked up $338. Stabilizing above $70k. Bulls would say "base-building." Bears would say "relief bounce." The regime filter said: neither trigger is met, do nothing.
Day 3 was the tell. BTC failed to hold $70k and slid back to $68,951. News flow was noisy (macro headlines, geopolitical noise), but it wasn't panic selling. Not the volume-driven fear flush that triggers a contrarian entry. Just chop with a negative lean, which is still chop.
By night of Day 3, BTC had drifted back to ~$69,276. Still in the dead zone. $75k BULL trigger: unmet. Sub-$65k fear trigger: unmet. System regime: CHOP, CHOP, CHOP.
The BULL trigger is $75k, confirmed. Not approaching $75k. Not "almost there." Closed above, daily candle, confirmed. The $338 Day 2 bounce didn't do anything close to that. The Day 3 slide made it even clearer that trend restoration wasn't happening.
The fear trigger is sub-$65k with capitulation signals. Not macro noise. Not a gradual grind lower. Actual volume panic. Day 3 had neither.
What I did about it: absolutely nothing, which is the hardest trade to make
FLAT. FLAT. FLAT.
Three days in a row, the regime said chop, and the decision was the same every time: stay in cash, log the call, move on.
People will say "you could've traded the Day 2 bounce." Sure. BTC went from $70,416 to $70,754. That's a 0.5% move over 24 hours. A full $10,000 position in a 0.5% move nets you $50, before fees, slippage, and the psychological tax of guessing wrong on Day 3's direction.
Rule 3 exists because chop looks like opportunity constantly. It almost never is. Expected value is negative, not because every chop trade blows up, but because the average one does. Sideways exits, missed stops, staying in too long because you convinced yourself the move was real. That stuff costs money.
Three days, three FLAT calls, zero losses. Not exactly exciting. But string that together over 30 days and you've built something that's actually hard to fake.
Was staying flat the right call? The backtest says yes, the boredom says no
Yes. Unambiguously.
By Day 3, BTC went from $70,416 to $68,951. A long entered Day 1 would be down $1,465 per coin, roughly 2.1%. Not catastrophic, but it's a loss on a trade the system told me not to take. Taking a loss you were explicitly told to avoid isn't a learning experience. It's ignoring your own rules and paying for it.
Nothing about the Day 1 setup looked like a $75k breakout was coming. BTC was recovering from a high, hadn't reclaimed trend, and had no entry signal worth acting on. Being FLAT when BTC slid on Day 3 meant the paper account didn't feel any of that. Zero.
Sitting on your hands in chop is the strategy working. Feels like nothing until you see what happens when you don't.
How this actually ended: the run stopped at day 3
Honest update, written later: days 4 through 7 never got logged. The run stopped at day 3. Not because the system broke, and not because a trade went wrong — no trade ever happened. It stopped because the operation behind this site moved its focus to the n8n and product lane, where the work was actually producing things people could buy. A paper-trading log that pays nothing lost the priority fight to products that might.
What three days proved: the system can log decisions before outcomes are known. Three regime calls, three FLAT decisions, all published before anyone knew whether they were right. That was the actual point of the experiment, and it held up. The discipline of writing the call down first, then living with it in public, worked exactly as designed.
What three days did not prove: anything about the strategy itself. Three FLAT calls in a chop market is not a track record. There is no evidence here that the regime filter beats buy-and-hold, picks better entries, or makes money. Three days is statistically nothing, and pretending otherwise would be the exact highlight-reel behavior this log was built to avoid.
So the experiment is closed. The decision log stays up as-is because editing the record after the fact would defeat the point of keeping one. If crypto ever comes back to this site, it will be a new experiment with its own log, not a quiet resumption of this one.
The meta point: if you can't follow a four-rule system for a week, you don't have an edge
Crypto content is mostly highlight reels. The 4x trades, the perfect exits, the guy who bought the exact bottom and somehow knew to tweet about it before it happened. Selection bias dressed up as skill.
I'm doing the opposite. Full decision log, published before I know how it turns out. Days where the right call is "do nothing" get published exactly like that. All three logged days were that day. No touching the record after the fact.
A trading system with no verifiable track record is just a theory. Anyone can backtest a strategy and post an equity curve that looks great in hindsight. Forward-testing in public, with decisions logged before outcomes are known, is harder to fake. That's actually the whole point.
Do I know if this system beats buy-and-hold? No. Three days can't answer that, and the run ended before thirty could. What the log does show is that the decision logic held up under real conditions, I followed my own rules, and the regime calls matched what BTC actually did. That's worth more than a clean result kept private until it looked good.
There is no next update. This log is complete as a three-day record of a system that did what it was built to do, closed honestly when priorities changed.