> For the complete documentation index, see [llms.txt](https://help.hydravision.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.hydravision.ai/products/asset-profiler/retrace-how-far-and-how-long.md).

# Retrace: How far and how long

{% hint style="warning" %}
**The common misreading.** That the median wait is how long a grid cycle takes. It isn't. It's how long **one two-position close** has historically taken. A real grid is deeper, and needs several of these.
{% endhint %}

### What "retrace needed" means

For PCL to close a pair of positions profitably, price has to come back far enough to cover three things at once:

* the gap between the two positions' average price and the one at the frontier;
* the profit target you want out of the close;
* everything the pair cost you: spread, commission, and any swap accrued while waiting.

Add those together and you get the retrace required. It's shown in your quote unit, with a second figure telling you what share of the asset's four-hour range that represents.

That second figure is the useful one for comparison. Anything under about 50% of the four-hour range is an easy ask. Well over 100% means you're depending on multi-bar moves.

### How long it takes

The panel underneath measures it. The Profiler walks the history and records, every time the retrace target was set, how long it took to arrive.

| Row                                    | What it tells you                                                            |
| -------------------------------------- | ---------------------------------------------------------------------------- |
| **Retrace required at the band floor** | The distance being measured.                                                 |
| **Median time to deliver it**          | Half the time it arrived faster than this.                                   |
| **80th percentile**                    | Four times out of five it arrived within this. **Plan against this one.**    |
| **p80 at the wide end of the band**    | The same figure at your widest step, so you can see how much the wait grows. |
| **Against a runway of**                | How long your carry budget allows. Compare it directly to the p80.           |
| **Target vs one M15 bar**              | How big the retrace is relative to a fifteen-minute bar.                     |

The comparison that matters is **p80 wait against runway**. If the retrace usually arrives well inside the runway, the wait costs you nothing meaningful. If they're close, you're relying on the fast half of the distribution.

### Two notes the panel may show

**"Measured on M1."** Every step in the range retraced inside a single fifteen-minute bar, so the measurement was re-run at one-minute resolution. Without that, every step would report the same wait and the comparison would be meaningless.

**"The retrace is small relative to an M15 bar."** The distance being measured is smaller than the bars used to measure it, so the reported median is a floor rather than a precise figure. Treat it as "very fast" rather than as an exact duration.

### If the panel says "not measured"

The required retrace never arrived, not once, in the whole sampled history.

That's a finding, not an error. This asset does not give back what a two-position close needs at this step. The verdict will read *Not viable*, and the honest response is to look at a different symbol, or at a window that better represents how the asset normally behaves.

### Remember the depth multiple

Everything above measures **one** two-position close. Real grids run deeper.

As a rough scale: a shallow grid typically needs one and a half to two steps of retrace to work through; a deeper one, two and a half to four. So if your strategy runs deep grids on an asset whose p80 wait is already eight hours, plan for considerably more than eight hours.

### What to do with this

* Compare the p80 wait to the runway. That single comparison is most of what this panel is for.
* Check the retrace as a share of the four-hour range. It's the figure that compares across assets.
* Multiply the wait by your grid depth before deciding an asset is fast enough.