Coil vs Trayd (trayd-mcp)
An open-source Robinhood-via-Claude execution bridge vs. a pre-built, validated, self-tuning engine — closest architectural cousins, very different products.
Of everything we get compared to, Trayd (trayd-mcp) is the closest architectural relative. Both projects let an AI agent place real trades in a Robinhood account using the Model Context Protocol. If you've used Trayd and you're wondering how Coil is different, the honest answer is: they sit at opposite ends of the same pipe. Trayd is the bridge. Coil is the engine that runs over a bridge like it. This page is a fair comparison of what each one is — and isn't. None of it is advice, and nothing here is a recommendation to buy, sell, or hold any security.
To be clear up front: Coil is an independent project and is not affiliated with, endorsed by, or connected to Trayd or its authors. Trayd is open-source software with its own community; we describe it from its public project, and any capability we attribute to it is theirs, not a claim we make on their behalf.
What Trayd actually is
Trayd (the trayd-mcp project) is a free, open-source MCP server that exposes Robinhood to a Claude agent so you can trade in natural language — "buy 10 shares of NVDA," "show me my positions," "close everything" — often driven through a recurring runner like /loop. It's a flexible execution bridge: it gives an agent hands on your account. What it deliberately does not ship is a strategy. There's no validated edge in the box, no backtest harness, no risk-ladder, no opinion about what to trade or when. You bring — or author — the rules yourself.
Trayd's genuine strengths
These are real, and we're not going to pretend otherwise:
- Free and open-source. No price, full source. You can read every line, fork it, and change anything.
- Hackable and general-purpose. Any ticker, any side, any logic you can express to an agent. It isn't boxed into one instrument.
- Multi-account / multi-asset reach. Because it's a thin bridge, it doesn't constrain you to a single pair or a single account the way a fixed engine does.
- Developer-friendly. If you're comfortable writing and testing your own strategy, a bridge plus an agent is a powerful, cheap foundation.
A capable developer can build a great deal on top of Trayd for nothing but their own time. That's a legitimate path, and for some people it's the right one. If you want to understand the underlying pattern — an agent with broker hands — our explainer on whether Claude can trade stocks and the broader Claude + Robinhood agentic trading guide cover it without selling you anything.
What Coil is
Coil is not a bridge. It's the layer that's missing from a bridge: a pre-built, validated, self-tuning strategy engine. It runs over the same kind of plumbing — built for Claude as a scheduled agent, built for Robinhood (or any equivalent broker MCP) — but what you're buying is the part that's hard to get right: the rules, the safety rails, and the reproducible evidence.
- A specific, narrow strategy. Coil trades only SOXL and SOXS, the 3x long and inverse semiconductor ETFs. Entries are a "compression-to-ignition" leg-rider — a sustained 2%+ intraday move treated as a short-timeframe reversal inside the prevailing 1-hour trend. Exits are deterministic: a 0.8% counter-move soft trail (roughly 2.4% on the 3x ETF) plus a 5% hard stop, and the inverse ETF is never held overnight.
- A risk ladder, not a vibe. An equity-high-water-mark circuit-breaker ladder governs everything: a −6% day halts new entries, −10%/−15% cut size, −25% is a hard stop, and no single symbol exceeds a 65% cap. This is the part you'd otherwise have to design, test, and trust yourself.
- A cold-backtest harness + 20 integrity guards. The same harness the developer uses ships with the engine: a fresh process per market regime so cross-run cache warmth can't flatter the numbers. That harness once caught its own three backtest bugs — a look-ahead, a next-day leak, and a sign-inverted short book.
- Local self-tuning, on a leash. Coil re-fits within hard whitelisted bounds and rolls back automatically when a change doesn't hold up. Why that matters is covered in self-tuning vs static trading bots.
- One-time $9.99. You download the full engine and own it. Optional Pro ($25/mo or $249/yr) delivers each new cold-validated version via Coil Sync — you're paying for fresh research, not for permission to run software you already have.
The caveat that applies to both, equally: SOXL and SOXS are 3x leveraged and can lose value rapidly, including total loss of the capital you put in. A bridge like Trayd gives an agent the ability to act; an engine like Coil gives it rules to act on — but neither removes market risk. With Trayd, an under-tested rule you wrote can trade exactly as confidently as a good one. With Coil, the rails reduce single-day damage but stops can still gap straight through a price. In both cases you own the keys, the capital, and the risk.
Bridge vs engine — the comparison
| Question | Trayd (trayd-mcp) | Coil |
|---|---|---|
| What it fundamentally is | Execution bridge — agent hands on Robinhood | Strategy engine that runs over a bridge like it |
| Trading rules | You bring or author them — none in the box | Pre-built, specific SOXL/SOXS strategy ships with it |
| Validated edge / evidence | None out of the box — depends entirely on your rules | Backtested across regimes; treat as hypothesis, not proof |
| Backtest harness | Not included — you build your own testing | Cold-backtest harness, fresh process per regime, included |
| Risk controls | Whatever you implement | Equity-drawdown circuit-breaker ladder + 65% symbol cap, built in |
| Integrity checks | Up to you | 20 integrity guards (caught its own 3 backtest bugs) |
| Self-improvement | You code it if you want it | Local self-tuning within whitelisted bounds, auto-rollback |
| Scope | Any ticker, any side, multi-account | SOXL/SOXS pair only, deliberately narrow |
| Open source / hackable | Yes — free, fully readable, forkable | No source license; engine is readable and yours to keep, not a fork base |
| Price | Free | $9.99 once; optional $25/mo or $249/yr for cold-validated updates |
| Who owns the risk | You | You |
The last row is identical on purpose. In both worlds you are the account holder and the leveraged-ETF total-loss risk is yours. The real difference is everything above that line: with Trayd you supply the strategy and the safety; with Coil they're the product.
The honest numbers — and why they're a hypothesis
Trayd doesn't advertise a track record, because it's a bridge — there's nothing to track until you write a rule. Coil does publish figures, and we'd rather over-disclaim them. Everything below is backtested or forward-tested under modeled execution — not client or live returns, best-case rather than typical, and past results do not predict future ones:
- Best trailing 250-session window (to 2026-06-13): +78.3%, profit factor 3.87, max drawdown 6.4%. This is the single strongest window in the test — do not anchor on it.
- 2024 chop: +11.4% (PF 1.51). 2023 quiet bull: +3.1% (PF 1.19), with idle cash on no-setup days earning the broker's variable sweep (for example, Robinhood Gold quoted ~3.35% APY in early 2026 — that's the broker's yield, variable, not paid by Coil, and not risk-free).
- 2022 bear: −1.4%. The honest weak spot — it was −3.6% before a stand-down-to-cash gate on confirmed bear days.
Sample-size caveat, in plain terms: this is roughly 115 trades a year on one ETF pair, under ~500 trades of total validation. Treat every figure above as a hypothesis, not proof. The point of comparing to Trayd isn't that Coil's numbers beat Trayd's — Trayd has none to publish — it's that a validated, evidence-backed strategy is exactly the thing a bare bridge leaves you to build and prove yourself.
How to choose
If you're a developer who enjoys authoring and testing your own logic, wants any ticker and any side, and values free, forkable, open-source code, Trayd is a genuinely good foundation — and you should use it without guilt. If instead you want the validated strategy, the deterministic exits, the circuit-breaker ladder, and a reproducible backtest harness handed to you for the price of a sandwich — so you're not betting real capital on rules you've never stress-tested — that's the gap Coil is built to fill. The two even compose: Coil runs over the same agent-plus-broker pattern a bridge like Trayd demonstrates.
Neither tool is "guaranteed better," and neither can promise a profit. For the wider field, see Coil vs trading bots and signal services; when you've decided which kind of builder you are, the pricing page has the $9.99 download and the Pro tier.
The strategy a bridge leaves out
Trayd gives an agent hands. Coil gives it validated rules, a circuit-breaker ladder, and a reproducible backtest harness — one-time $9.99, yours to keep. You hold the keys, the capital, and the risk.
See pricing — from $9.99Coil is software you install and run yourself, with your own brokerage credentials and capital. It is not investment advice, not a managed account, and not a signal service. Leveraged ETFs such as SOXL and SOXS can lose value rapidly, including total loss. All performance figures are backtested or forward-tested under modeled conditions — not client returns; past performance does not predict future results.