Crypto signals your agent can read — with the losses printed next to them

The honest hook: it does not beat holding

Most crypto signal pages lead with the best year and hope you don't ask what it's compared against. We'll do the opposite. Over 2016–2026, Coil's crypto trend rule compounded at 59.6% CAGR versus 74.9% for the obvious benchmark — a 50/50 BTC/ETH portfolio, monthly rebalanced, just held. If your only goal is maximum long-run CAGR and you can genuinely sit through an −87% drawdown, buy and hold beats this rule. That sentence stays on this page for as long as it's true.

What the rule buys with the CAGR it gives up is the crash years. In 2022, holding BTC/ETH lost −66.0%; the rule finished 0.0% — flat, in cash, out of the way. In the October 2025 → March 2026 crash, holding drew down −55.7%; the rule drew down −19.7%. Roughly 80% of hold's compounding with the worst years cut down: a risk posture, not a moonshot. If someone shows you crypto signals without the number for just holding printed next to them, ask why.

What the Crypto book actually is

Since July 15, 2026, the Coil Scanner board pairs the Equities book (S&P 500, Nasdaq-100, and Macro) with a Crypto book covering BTC and ETH. The rule is deliberately simple and fully stated here — no black box:

• Long-only, two sleeves. The book is split into two 50% sleeves — one Bitcoin, one Ethereum. Each sleeve is either in its asset or in cash. Never short, never levered.
• The trend gate. A sleeve is IN only while its own price is above its 50-day moving average and BTC is above its 200-day moving average. BTC's long trend is the regime gate for both sleeves: when the market's anchor asset breaks its long trend, everything goes to cash.
• Daily cadence, stated plainly. Signals evaluate on the UTC daily close, and fills are modeled at the next daily close — no intraday heroics, no look-ahead. The board itself refreshes on weekdays during US market hours. Crypto trades 24/7; our refresh does not, and a weekend move won't show until the next market morning. This is a daily-cadence trend read by design — if you need tick-by-tick coverage, this isn't that.

The record — ugly parts first

The numbers below come from a cold-validated, pre-registered study: 2016–2026 Coinbase daily data, with worst-case costs of 130 bps per side modeled on every trade (the bottom-tier taker fee — most real accounts will pay less). The benchmark is the honest one: 50/50 BTC/ETH, monthly rebalanced, held throughout.

Read the losing rows twice, because they are the real cost of this rule:

• 2024 was brutal. The rule lost −22.6% in a year holding made +74.7% — a choppy tape whipsawed the 50-day gate in and out, paying costs each time, while holders just rode it. Trend rules have whipsaw years. This one had a bad one, and there will be more.
• It gives up big chunks of the manias. In 2017 and 2021 — the monster bull runs — the rule captured far less than holding did, because every shakeout knocked sleeves to cash and re-entry came at the next close, higher. Trend-following sells some of the top and buys some of it back.
• −64% is still a crypto-sized drawdown. The rule cuts the worst of it, but "cut down" in crypto means you can still lose nearly two-thirds from a peak. If a −64% drawdown would break you, no gate fixes that — the position size has to.

And the standard caveat, which is not boilerplate here: this is a research backtest under modeled conditions. It is not live results, and past or backtested performance does not predict future returns.

How an agent reads it — and where it executes

Coil publishes the read; your agent executes at your broker. Coil executes nothing, holds no funds, and never touches your keys. Two ways to get the board:

• Subscription: Coil Scanner at $12/mo (or $99/yr) — the full board in a browser or, for your agent, the same license key against the board API. The whole board — the Equities book and the Crypto book — regime verdicts, buy-list lanes.
• Per read, no account: autonomous agents can buy a single board read over the open x402 protocol — GET https://coil.trade/api/board/agent, $0.25 in USDC on Base, payment and receipt on-chain, no card, no email. The agent endpoint docs.

On the execution side, the landscape in July 2026 is uneven and worth stating exactly. Coinbase supports agent trading today — its Coinbase for Agents tooling (a CLI and MCP interface over its retail exchange) lets an AI agent hold an isolated portfolio and place spot trades with scoped keys. Robinhood has announced agentic crypto (July 10, 2026) with no launch date — when it ships, the same board read applies there. Wherever your agent trades, the sequencing is yours: the board says which sleeve the rule has in or out; your agent, your broker, your rules, your risk.

FAQ

Do Coil's crypto signals beat buy-and-hold?

No — 59.6% CAGR versus 74.9% for a 50/50 BTC/ETH hold in the 2016–2026 cold backtest, and we lead with that. The rule's case is drawdown: 2022 flat versus −66.0% for holding, −19.7% versus −55.7% in the Oct 2025 → Mar 2026 crash, max drawdown −64.3% versus −87.4%. It is a risk posture, not a moonshot, and backtests do not predict future results.

How does an AI agent trade crypto with Coil's signals?

Your agent reads the board (subscription, or $0.25/read over x402) and executes at your own broker. Coinbase supports agent trading today via Coinbase for Agents; Robinhood has announced agentic crypto with no US date yet. Coil publishes the read and executes nothing.

Are the signals live or backtested?

Tracked live since July 15, 2026. Everything before that is hypothetical backtest, and the Crypto book is not traded with real money by Coil. The board refreshes on weekdays during US market hours; crypto trades 24/7, so weekend moves appear at the next market morning.