> 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/cost-share-across-the-band.md).

# Cost share across the band

The line chart, and the table at the bottom of the report headed **"The sweep — every step tested across the band"**.

{% hint style="warning" %}
**The common misreading.** That the lowest point on the chart is the best step. On most assets the line simply keeps falling to the right, and the lowest point is just the widest step tested. That's arithmetic, not a finding.
{% endhint %}

### What the chart shows

The Profiler doesn't just calculate the two ends of your range. It walks across it in seven steps and measures each one: how far price has to retrace, how long that historically took, and what share of the profit target costs would eat.

The chart plots that last figure. Step size runs along the bottom; the share of the target consumed by costs runs up the side. A red dashed line marks your tolerance, and the red tint above it is the region you've said is unacceptable.

### Why the line usually falls

Your costs at entry, spread and commission, are fixed. They don't care how wide your step is. But the profit target grows with the step. So a wider step means the same fixed cost is spread over a bigger target, and the share falls.

This is why "widest is cheapest" is a mathematical certainty on most assets rather than a recommendation. The chart is telling you the shape of a trade-off, not pointing at an answer.

{% hint style="info" %}
**Then why look at it at all?** Two reasons. First, the *level* matters. A curve sitting at 3% is a very different asset from one sitting at 30%, even though both fall. Second, on assets with heavy overnight costs the line stops falling and turns back up. Where that happens, there is a real answer, and the report says so. See [Recommended Step](/products/asset-profiler/recommended-step.md).
{% endhint %}

### When the axis zooms in

On cheap assets the whole curve can sit far below the tolerance line. Drawing the axis all the way up to 35% would flatten the curve into a straight line and hide its shape.

So when the tolerance is far above the data, the chart zooms to the curve instead, and says so in green at the top right: *"axis zoomed to the curve — every step is far below the 35% tolerance"*.

{% hint style="success" %}
That message is good news. It means costs aren't close to being a constraint on this asset.
{% endhint %}

### The sweep table

The same seven measurements, in numbers:

| Column          | What it is                                                                                 |
| --------------- | ------------------------------------------------------------------------------------------ |
| **Step**        | The step size tested, in your quote unit and in ticks.                                     |
| **Retrace**     | How far price has to come back, in ticks, to close a pair at this step.                    |
| **Median wait** | How long that retrace historically took, half the time.                                    |
| **p80 wait**    | How long it took in the slower four cases out of five. This is the number to plan against. |
| **Cost share**  | The share of the profit target costs would consume.                                        |
| **Samples**     | How many times this retrace occurred in the measured history.                              |

**Use the p80 column, not the median.** The median is the comfortable case. The p80 is what you should assume when you're deciding whether an asset's waits fit inside your carry runway.

**Watch the samples column.** It falls as the step widens, because bigger retraces happen less often. A row with few samples is a less reliable measurement. If any row drops below the reliability threshold it's dimmed, and the panel header counts how many.

### What to do with this

* Read the level of the curve, not its direction. Direction is almost always downwards.
* Plan against the p80 wait column, and compare it to your carry runway.
* If rows at the wide end are dimmed, treat the wide end of your range as less well evidenced.