> ## Documentation Index
> Fetch the complete documentation index at: https://docs-d0.getdonut.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Acting on Thorp Signals

> The signal engine watches your list and pushes setups that clear its filter, with the trigger, the hit rate, the payoff, and the evidence chain attached.

## The Scenario

Most signal services send you a ticker and a direction. That's not enough to act on: you can't tell whether it's a high-probability setup or a lottery ticket, and you can't check the reasoning.

Thorp signals arrive with the trigger that fired them, the historical hit rate, the risk-to-reward, and, for event-driven signals, the chain of confirmations that had to pass before it was sent. You can disagree with a signal, but you can always see what it was built on.

## Try It

* *Add BNB to my Thorp watchlist and push me technical signals*
* *Why did you push that signal? Show me the evidence chain*

## What You Get Back

### A technical signal, honestly labelled

This one comes from your watchlist on its own, without a prompt:

> **D0:** BNB on your watchlist just printed a 4H MACD death-cross short signal 👀\
>
> 🍩 **BNB · Short · 4H**\
>
> **Technical trigger**
>
> * 4H MACD death cross: DIF −3.524 crosses below DEA −3.3681
> * Histogram −0.1559, turning down, short-term bearish
>
> 🎯 Entry \$586.75 → Target \$546.07 (≈ −6.9%)\
>
> 🛑 Stop \$607.09 · R/R ≈ 1:2\
>
> 📊 30d hit 22.7% · avg −1.3% · 22 trades\
>
> → A low-win-rate, high-payoff "odds-type" setup\
>
> \[View details] \[Short in one tap] \[Ignore]

A 22.7% hit rate looks bad on its own, but at 1:2 R/R, losing three trades out of four and winning the fourth still works out. The engine pushes these deliberately and labels them as odds-type. A feed that only ever showed high win rates would be hiding the trades that pay.

### An event-driven signal, with its confirmations

> **D0:** 🗞️ **Thorp signal: ZEC news-driven short**\
>
> **Tier 1: news capture**
>
> * Bearish ZEC news detected: security research disclosed a potential protocol vulnerability; social concern building
> * News sentiment: 81% negative, spreading fast
>
> **Tier 2: on-chain / order-book validation**
>
> * ✅ ZEC perp OI rising + funding turned negative, shorts entering
> * ✅ Exchange net inflows rising (potential sell pressure)
>
> **Tier 3: execution & result**
>
> * Shorted at \$432 → now \$415.15 (−3.9%)
> * 3x leverage · short unrealized ≈ +11.7%
> * TP −8% / SL +5%
>
> 🎯 News + on-chain double confirmation passed\
>
> ⚠️ Bearish news must be cross-verified; size carefully, not financial advice\
>
> \[View details] \[View evidence chain]

Nothing gets pushed on the news alone. The story has to be corroborated by positioning and flow data before it becomes a signal, which is why the tiers are numbered in the message.

> [!WARNING]
>
> Hit rates, average returns, sample sizes, and PnL figures in these examples are sample output from a specific moment, shown to illustrate the format. They are not performance claims or predictions, and a signal clearing the engine's filter is not a recommendation to take it.

## When to Use

* Coins you hold but can't watch: put them on the watchlist and let the technical scan run.
* When you want entry, target, stop, and R/R in one message.
* Event-driven trading: news breaks constantly, and only some of it shows up in positioning. The double-confirmation filter checks for that.
* Auditing your own read: ask for the evidence chain and compare it with your reasoning.

## Tips

* Judge hit rate and R/R together; neither number means anything alone. A 70% hit rate at 1:0.3 is a losing system.
* Ask before you act. `"Why did you push that signal?"` returns the full chain, including what nearly disqualified it.
* Keep the watchlist short. Signal quality drops when the list is long, because the engine has more chances to find something marginal.
* Set your own size. One-tap execution still takes your size and leverage. See [Risk Controls](https://docs-d0.getdonut.ai/features/risk-controls.md) for caps.
* Skip freely. Ignoring signals is the normal case.

## Related

* [Thorp Strategy](https://docs-d0.getdonut.ai/features/thorp-strategy.md): how signals are generated and scored
* [How the Agent Decides](https://docs-d0.getdonut.ai/introduction/how-the-agent-decides.md): the reasoning D0 exposes for any action
* [Risk Controls](https://docs-d0.getdonut.ai/features/risk-controls.md): leverage caps and position limits
* [Distilling Videos and Threads](https://docs-d0.getdonut.ai/use-cases/distill-research.md): sentiment that feeds the engine


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.