Market

HOLD — the trend is up

On the 2026-10-03 23:59 UTC close, price was at $2,687 and the 150-day average at $2,082 — so price sat +29.1% above the average, and the state has held for 46 days. The rule switches to HOLD when price climbs more than +2% above the average, and to CASH when it drops below −2%. Between those two thresholds it does nothing — so it doesn't flip you in and out on every small swing.

Read from the daily decision record, computed on KuCoin candles. This page does not recompute it.

◈

Where the money actually comes from — worth reading once: the rule gets the direction right about 50.5% of the time, which is exactly what you'd get flipping a coin. That is not where the gain comes from. The gain comes from the size of the moves: while in HOLD, ETH gains an average of +3.55% over 30 days, while in CASH only +1.04%. In other words: it is wrong as often as anyone else, but it catches the big moves and misses the small ones. That is why there is no point "improving" it by trying to be right more often.

7d -0.3% · 30d +7.2% · 90d +50.5%

100%

of normal exposure · far below its peak

ETH is 44% below its all-time high. Historically the best place to be buying, so size stays full.

◈

What this is, exactly: a prudence cap, not an edge. It predicts nothing — it simply holds back exposure near the highs, and holding back less capital shrinks the tail of losses.

What it is not: the favourable numbers you may have seen for it (20.4% a year against 19.2%, drawdown 27% against 38%) come from the one 5.5-year window the thresholds were chosen on. Tested properly — thresholds picked only on past data, applied to the next unseen block — it came out below the plain rule, randomly chosen thresholds did just as well, and simply multiplying exposure by a constant achieved a better tail without any of this logic. Over 9 years it is worse than the plain rule.

So: unconfirmed, not disproven. Treat the percentage above as a ceiling you have chosen for prudence, and not as evidence that this app knows when to step back.

Volatility, 30d

40%

relatively calm

Usual move in 14 days

±7.8%

neither good nor bad — the normal

From the yearly high

-42.6%

high $4,684

Trend duration

46 days

on the same side of the SMA150

The "usual move" is the reference you need in order not to panic for nothing: a swing of ±7.8% over two weeks is normal for ETH at the current volatility. If a tool gets agitated by moves smaller than that, it is reacting to noise, not to something real.

Median year

+11.5%

median across 12-month windows

Reaches +20%

45.6%

of 12-month windows

Loses money

33.5%

one year in three

Worst year

−35.9%

buy & hold: −76.7%

Maximum drawdown

39.4%

buy & hold: 79.3%

Time to +20%

113 days

median, from any starting day

Run php scripts/portfolio_whatif.php 10000 20 to recompute with your own amount and target. These figures come from the frozen rule, not from a forecast. One year in three you lose money — that is a property of the asset, not of the code, and no version of this app changes it.

Greed

Exploratory observation, not a validated result. Inside the same trend state, ETH has behaved differently depending on the mood: with the trend up and fear below 45 it averaged −9.8% over 30 days; with the trend up and greed above 55, +6.3%. But the 45/55 bands were picked after looking at the data, there are only about 5 independent checks behind the figure, and Fear & Greed is itself computed partly from price and volatility — so some of the effect may just be the trend under another name. The test that would settle it has not been run yet. Read it as a reason for caution, never as a signal.

Historical figures come from real prices, after costs. The past does not repeat exactly. Updated 2026-10-04 07:00 UTC

01What the rule says

02Where the price is

03How much to expose

04Context

05What to expect from the rule

06Market mood