Docs  /  AI Studio Targets  /  Mean Reversion

AI Studio Answer Guide: Mean Reversion

Status: Per-family deep reference (companion to the consolidated Fusion AI Studio & ONNX User Guide) Audience: first-time AI Studio user Answer family: Mean Reversion Prediction style: Regression

1. The Plain-English Question

Mean Reversion asks:

“Over the next N bars, how much is price expected to move back toward today’s mean?”

This family is different from normal return prediction.

It is not mainly asking:

It is asking:

“Will price end up closer to its current mean, or farther away from it?”

2. What Fusion Is Actually Building

Example:

Meaning:

Fusion compares:

Current trainer formula:

Target_Reversion10[t] =
(|Close[t] - SMA20[t]| - |Close[t+10] - SMA20[t]|) / ATR14[t]

Plain-English translation:

3. The One Important Mental Model

This family predicts reversion strength, not direction by itself.

That is the most important thing to understand.

A positive value means:

“Price is expected to end closer to the mean.”

But that does not automatically mean:

“Go long.”

Why?

Because you still need to know which side price is stretched on right now.

If price is:

So this family works best when you pair it with current extension context such as:

4. Which Step 1 Settings Control This Family

This family uses its own Step 1 section:

Important

For this family:

This is another place where the UI needs clearer family scoping.

5. What The Horizon Means Here

If the mean reversion horizon is:

10

the question becomes:

“Will price end up closer to today’s 20-bar mean 10 bars from now?”

If you enter:

5,10,20

Fusion can create:

Then Step 2 chooses one of those columns for one model.

6. What The Model Output Means

This is a regression family.

The correct meaning is:

positive = price is expected to end closer to today's mean
negative = price is expected to end farther from today's mean
near 0   = little expected change in distance from the mean

Important

This output is:

But the sign is about:

not automatically:

7. What 0 Means Here

For Mean Reversion, 0 means:

That is different from:

Here, 0 means:

“No strong expected reversion or further stretching.”

8. Why Traders May Like This Family

This family matches a very common trader question:

“If price is stretched away from the mean right now, how likely is it to snap back?”

That makes it useful for:

It is especially useful when your trading style already cares about:

9. Why This Family Can Be Misused

This is where users often go wrong.

They see:

positive = bullish
negative = bearish

But that is not always true.

The correct reading is:

positive = stronger expected move back toward the mean
negative = stronger expected move away from the mean

So you still need current context to know which side is stretched.

This is why the product copy already recommends pairing it with:

10. What “Today’s 20-Bar Mean” Means

This family currently uses:

as part of the built-in target design.

Important current product truth

These are not currently separate Step 1 knobs for this family.

So the user should think of this as:

“Fusion’s built-in 20-bar mean reversion label measured in ATR14 units.”

not:

“A fully custom mean-reversion formula builder.”

11. Step-By-Step Mean Reversion Walkthrough

Step A: Prepare the raw dataset

Open Export and make sure:

Because this family needs current extension context, good feature choices often matter more here than with simpler families.

Step B: Turn on mean-reversion answer generation

In AI Studio Step 1:

Example first pass:

Step C: Create the prepared dataset

Click Create New Capture.

Now Fusion creates answer columns such as:

Step D: Choose the actual training question

In Step 2:

Example:

Now the model is learning:

Step E: Choose the features

Pick the inputs that may help answer:

“Is price likely to move back toward the mean from here?”

Useful first feature types often include:

Good first rule:

Step F: Train

Choose:

Good first algorithms:

Step G: Read the result

Think of the output like this:

positive = stronger expected move back toward the mean
negative = stronger expected move away from the mean

Then combine that with current price location:

below mean + positive reversion = can support a long fade idea
above mean + positive reversion = can support a short fade idea

That is the right way to use this family.

12. Good First Session

If you want a clean first experiment with this family:

Step 1

Step 2

Step 3

Feature tip

Make sure the feature set includes at least one current extension measure such as:

Without that, this family is much harder to interpret correctly.

13. What Not To Do

Do not treat positive output as automatically bullish

Positive reversion only means “closer to the mean,” not “up.”

Do not use this family without current extension context

If the model does not know which side is stretched, the output is harder to turn into a trade idea.

Do not assume this uses the normal look-ahead field

This family uses Mean Reversion Horizons, not Look-ahead Windows.

Do not assume this is a custom formula builder

This is currently a built-in reversion target built around today’s SMA20 and ATR14.

14. Current Product Notes From The Mean Reversion Guide Pass

These are the product improvements this guide exposes.

A. The UI should say explicitly that positive does not automatically mean long

Suggested note:

“Positive Mean Reversion means price is expected to end closer to today’s mean. Pair it with current extension context such as ZScore to know whether that supports a long or a short setup.”

B. The family picker should show the built-in anchor clearly

Suggested note:

“Current built-in formula uses today’s SMA20 and measures reversion in ATR14 units.”

C. Step 1 should clearly scope the mean-reversion controls

Suggested note:

“Mean Reversion Horizons create mean-reversion answer columns only. They do not control Future Return or ATR-Normalized Return.”

15. One-Sentence Rule To Remember

Mean Reversion means:

“Predict how much price is expected to end closer to today’s 20-bar mean over the next N bars, measured in ATR units, and combine that with current extension context to decide trade direction.”

← All target families
NinjaTrader® is a registered trademark of NinjaTrader Group, LLC. Fusion Indicators LLC is an independent software vendor and is not affiliated with, endorsed by, or sponsored by NinjaTrader.