Noetfield Runway · Stock & Crypto
A daily crypto screen for traders who already have rules. Twelve pairs scored before dawn, falsifiable plans for the few that survive, and published rejections — with reasons — for the rest.
No card · Research only, it never places an order · Next brief Sun 07:00 PDT
Every number on this page comes from the agent's own receipts.
Symbols screened
108
across every cycle, in plain arithmetic
Plans refused by the gate
8 of 8
the model proposed; deterministic code declined
Days with no setup worth taking
—
the model declined to propose anything at all
Orders placed
0
structurally — there is no order path
There are no testimonials here and no reader counter. These are counts of what the agent has actually done, checkable against the receipts.
The most recent brief, exactly as it ran
Not a screenshot and not an example — the live receipt from the last cycle, including the mornings when nothing qualified.
Score
78
Confidence
70%
Max size
312.50
Data age
38m
| Symbol | Score | Read | Why |
|---|---|---|---|
| ATOMUSDT | LONG | Base and higher timeframe both up. | |
| POLUSDT | LONG | Base and higher timeframe both up. | |
| LTCUSDT | LONG | Base and higher timeframe both up. | |
| SOLUSDT | LONG | Base and higher timeframe both up. | |
| XRPUSDT | SHORT | Base and higher timeframe both down. | |
| BNBUSDT | NO_DIRECTIONAL_EDGE | Base and higher timeframe both up. | |
| AVAXUSDT | NO_DIRECTIONAL_EDGE | MACD histogram agrees with the trend direction. | |
| DOTUSDT | NO_DIRECTIONAL_EDGE | -0.97% over the momentum lookback. | |
| ETHUSDT | THIN_VOLUME | Base and higher timeframe both up. | |
| ADAUSDT | THIN_VOLUME | Base and higher timeframe both up. | |
| LINKUSDT | THIN_VOLUME | Base and higher timeframe both up. | |
| BTCUSDT | THIN_VOLUME | Base and higher timeframe both up. |
Next brief Sun 07:00 PDT. Full evidence view →
Six frontier models were given $10,000 each to trade crypto unsupervised
It is running live right now at Alpha Arena, on real money, on a real exchange. Same models, same market, same week. One is up 46%. GPT-5 is down 75%.
Best of the six
+46%
Worst of the six
−75%
Human oversight
None
That spread is the whole argument. A model good enough to write a convincing thesis is not therefore good enough to be trusted with the decision.
So here it doesn't get one. The model writes the plan. Deterministic code decides whether it passes, using rules it cannot see or influence. When it refuses, the refusal is published with its reason — and you are reading those on the sales page.
Figures from nof1.ai Alpha Arena, a live public leaderboard. They move; check them.
Cheap arithmetic first, a model only where it earns its keep
Step one · free
Trend, momentum, volatility, volume, MACD, band position and higher-timeframe alignment, computed across the whole watchlist before anything is paid for. Most candidates die here, and dying here costs nothing.
Model spend
$0.0069
Last cycle, across every call.
Step two · paid
Survivors reach a model, which returns a thesis with entry, stop, target and explicit invalidation.
Step three · free
A deterministic gate returns approved, watchlist or rejected — checking staleness, direction coherence, reward-to-risk, exposure, daily loss and drawdown. The model gets no vote here.
Orders placed
0
Structurally, not so far. There is no order path.
Worth being blunt about this
What else you would buy instead
Public pricing, checked July 2026. It moves — the point is the shape, not the cent.
| What it is | Typical price | Tells you what to buy | Shows its rejections | Places the order | Promises returns |
|---|---|---|---|---|---|
| This daily research brief | Free · $29 | Yes, with levels | Every one, with reasons | Never | No |
| Signal groups Telegram, VIP channels | $47 – $99 · to $750 | Yes | Rarely | No | Usually |
| Screeners TradingView, altFINS | $15 – $70 | You screen | No verdict to reject | No | No |
| AI scanners Trade Ideas | $118 – $228 | Yes | No | No | Implied by track record |
| Bot platforms Composer, 3Commas | $20 – $50 | You author the rules | No | Yes, automatically | No |
| On-chain data Glassnode, CryptoQuant | $29 – $799 | Raw metrics | No verdict to reject | No | No |
The row that matters is the third from the right. Everything here either tells you what to buy and hides what it discarded, or hands you data and leaves the judgement entirely with you. This publishes the discards, which is worth less than a promise and considerably more than a hunch.
A free tier that is actually useful, and one paid tier
Public brief
Free
The same brief that runs every morning, on a fixed twelve-pair watchlist.
Operator
$29 / month
Your instruments, your hour, delivered — with the receipts to review it later.
Desk
Custom
Multiple seats, custom gate thresholds, webhook delivery.
| Feature | Free | Operator | Desk |
|---|---|---|---|
| Daily ranked screen | ✓ | ✓ | ✓ |
| Full plans and gate verdicts | ✓ | ✓ | ✓ |
| Watchlist | Fixed 12 | Yours | Yours |
| Delivery time | Fixed | Your hour | Your hour |
| Telegram delivery | — | ✓ | ✓ |
| Receipt archive | Latest only | Full | Full |
| API access | — | ✓ | ✓ |
| Custom gate thresholds | — | — | ✓ |
| Seats | 1 | 1 | Multiple |
The paid tiers are not open yet. Joining the list costs nothing and takes no card — you'll hear before there is anything to pay for.
The questions worth asking first
No. Every plan is a candidate for a person to review, and nothing it produces is actionable on its own. Read the full risk note →
Because the gate checks staleness, direction coherence, reward-to-risk, exposure, daily loss and drawdown — and the model gets no vote in it. See today's rejections →
No. There is no order path, no wallet and no exchange credential anywhere in the service. Market data comes from a public read-only feed. See how a cycle runs →
You shouldn't, yet. Run the free tier on paper for a few weeks and judge it on profit factor and gate-pass rate, not win rate. Start free →
The free brief runs a fixed twelve-pair USDT watchlist. Operator lets you set your own. Compare tiers →
It is stored and nothing else — no name, no tracking, no third parties, no card. Join the list →
It isn't open yet, and the page says so where you'd click. Joining the list costs nothing. See what it includes →
Yes — the current brief is served as JSON, and Operator adds API access to the archive. See the raw state →
Tell us where to send it
The free brief goes out every morning at 07:00 Pacific. No card, and you can stop at any time.
We store the address and nothing else. No tracking, no third parties, no card.
The part most of this category leaves out
Not financial advice. No track record. Judge it on paper for weeks.
Nothing this service produces is actionable on its own. Every plan is a candidate for a person to review.
It has run 9 cycles. That is not a track record. Run it on paper for at least several weeks and thirty-plus setups before it influences a real position, and account for slippage, fees, data latency and rejected orders when judging results. Judge it on profit factor and how often it passes its own gates — not on win rate, which flatters a system that wins small and loses big.
It holds no exchange credential. There is no order placement, no wallet and no execution path in the service. Market data comes from a public read-only feed.
Approved does not mean take this trade. It means every machine check passed and a person should now look.
Every prompt change and every risk-rule change makes a new version that needs evaluating from scratch.