Backtesting Jeff Sun’s 50 SMA ATR extension heuristic

... and taking it one notch higher

Last Friday, someone I know got in touch about AAPL 0.00%↑ and asked for my view on whether it was extended.

I pulled up the chart and looked at the distance from the 50 SMA in ATR terms. It was sitting at about 6.7x ATR, so I replied, “It looks like it has some more room to run, and once prices start reaching 7.5x to 8x ATR, that’s when it usually starts pulling back.”

He challenged me straight away and fairly so.

“How can you be so certain?” he asked. “How can you be certain when there’s no certainty about anything in trading?”

He had a point. Trading rewards humility. The market has a way of making fools of anyone who speaks in absolutes. But while there may be no certainty, there is probability. After years of staring at these charts and using Jeff Sun’s ATR extension indicator, I had experiential evidence — a strong intuition built from repetition.

The question was whether it was anything more than that.

SwingTheStrat has five years of historical data across roughly 2,700 tickers. That’s millions of candles. And for the first time, I realised we could actually test it properly — find out whether this was a well-worn heuristic or something genuinely empirical.

---

Why extension matters

At its core, the market behaves a bit like a rubber band.

When price stretches away from a moving average, tension builds. Sometimes that tension keeps expanding for longer than your gut thinks it should. Sometimes it snaps back fast. The key is learning the difference between a normal stretch and an extreme one.

That’s what ATR extension multiple is trying to capture.

Instead of asking, “How far is price from the moving average in dollars?” we ask a more useful question: “How far is price from the moving average relative to this stock’s own daily volatility?”

That matters because a $500 stock and a $10 stock can both be stretched, but not in the same way. ATR lets you compare them on equal terms. It standardises the move.

So if a stock is 7.5x ATR above its 50 SMA, that doesn’t mean the same thing as 7.5% or $7.50. It means price has stretched 7.5 times the stock’s normal daily range away from the anchor.

The 50 SMA bell curve

The 50 SMA is where this whole thing becomes especially useful, because it sits right in the middle of trend behaviour.

It’s fast enough to matter for swing trading, but slow enough to smooth out the noise. When you plot the extension of thousands of stocks against the 50 SMA, you don’t get randomness.

You get a bell curve.

That bell curve tells you what normal looks like.

We took over five years of daily candles across roughly 2,700 tickers and measured the 50 SMA extension in ATR multiples for every single candle. Then we plotted the whole lot. The result was exactly what you’d hope for if markets were behaving like a statistically messy but broadly understandable system: a clustered middle, thinner tails and a few extreme outliers.

Most candles live near the middle. The further you go out, the rarer the event becomes.

Translated into the numbers from the dataset, that means:

• 1 sigma is roughly 2.5x ATR.

• 2 sigma is roughly 5x ATR.

• 3 sigma is roughly 7.5x to 8x ATR.

What the sigma levels mean

It’s worth being precise here, because people often throw sigma around without really explaining it.

If we say a move is within 1 sigma of the mean, we mean it sits inside the range where about 68% of observations live. That’s ordinary behaviour.

At 2 sigma, you’re covering about 95% of the distribution. Anything beyond that is unusual enough to merit attention.

At 3 sigma, you’re out in the tail. That’s the territory where only about 0.3% of observations should live if the distribution is behaving normally. In plain English, that’s extremely rare.

That gives you a practical framework:

• 1 sigma means normal.

• 2 sigma means stretched.

• 3 sigma means you are in the extreme tail.

This is why the 7.5x to 8x ATR zone matters so much on the 50 SMA. It’s not just “a bit extended”. It’s the point where the market is telling you the move has already travelled a long way from its mean and is now vulnerable to mean reversion.

And mean reversion is often where traders get caught.

They don’t buy when the stock is still building. They buy when the stock has already done the heavy lifting.

Which brought us back to the original question: was Jeff Sun’s heuristic — that a stock reaching 7.5x to 8x ATR from the 50 SMA is likely to revert — actually supported by the data? The answer is an overwhelming YES!

So when a stock is sitting near 7.5x to 8x ATR above the 50 SMA, you are deep into statistically rare territory. It doesn’t guarantee an immediate reversal, because trading doesn’t work like that, but it does tell you the odds are no longer in your favour if you’re still chasing late.

That was the answer I was really giving about AAPL 0.00%↑ .

It wasn’t, “This stock will definitely pull back here.” It was, “This is where the odds start getting ugly.”

That’s a much better way to think.

The other moving averages

The 50 SMA is only the beginning.

SwingTheStrat also captures extensions for the 10 EMA, 21 EMA and 200 SMA. Once I plotted those too, a more complete picture started to emerge. Each moving average has its own personality and the numbers make that personality concrete.

The 10 EMA is the tight leash. Because it reacts fast, the extension bands are narrow. One standard deviation sits at just 0.93 ATR to the upside. By 3 sigma, you’re only at 2.69 ATR. That might not sound like much, but it means the 10 EMA reaches its extreme zones quickly. A stock doesn’t need to travel far before it’s statistically overstretched on this timeframe.

The 21 EMA gives a stock a little more room to breathe. The 1 sigma boundary sits at 1.45 ATR, and 3 sigma stretches out to 4.17 ATR. It’s still a responsive measure, but the bands are wider, which means you’re capturing a slightly more developed move before the signal becomes extreme.

The 200 SMA is a different animal entirely. Because it moves so slowly, price can travel a very long way from it before the extension becomes statistically significant. One sigma doesn’t arrive until 5.42 ATR. Two sigma is 10.72 ATR. Three sigma is 16.02 ATR. These are not small numbers.

But the more interesting thing about the 200 SMA is the shape of the curve itself. Look at it closely and you’ll notice it’s slightly skewed — the right tail is a little heavier than the left. That asymmetry is telling you something real about long-term trending behaviour. Stocks can spend a surprisingly long time extended to the upside.

Standard deviation extension bands for key moving averages

Using the bell curves, we can now set proper thresholds for positive and negative extension across the different anchors.

That table is more useful than a single number because it shows both sides of the distribution.

A lot of traders only think in terms of the upside extension. But the downside matters too.

A stock that is deeply extended below the 200 SMA is not just “cheap”. It may be broken, capitulated or in the middle of a violent unwind. Likewise, a stock stretched far above the 50 SMA may still keep climbing, but the odds of a sharp pause or mean reversion are rising fast.

A better way to think about extension

The best way to use extension is not as a yes-or-no switch.

It’s a context tool.

If a stock is only just starting to stretch, that doesn’t mean it has to reverse. It just means you should start paying attention.

If it becomes stretched on one moving average, that’s useful.

If it becomes stretched on two, that’s more interesting.

If it stretches on three or four at the same time, you start getting real conviction that mean reversion is getting closer.

A stock can be extended from the 10 EMA and still continue higher. It can even be extended from the 50 SMA and keep grinding. But when you start seeing the 10, 21 and 50 all moving into stretch or trap territory together, that’s when the move starts to look exhausted.

Jeff’s indicator that73K traders around the world have viewed is super useful, but it’s still one-dimensional. It tells you the extension in a column, not the shape of the move. The extension cloud fixes that by plotting the stretch directly on the chart, so you can see how price behaves around the 10, 21, 50 and 200 moving averages in real time.

This is why the extension cloud is so helpful visually.

You don’t just want a number. You want to see how the candles behave inside the cloud, how quickly they move through it and whether they’re stretching on one anchor or all of them at once.

---

AMD: a stock that kept stretching

AMD is a good example of why extension shouldn’t be treated as a hard stop.

A stock can look extended in one dimension and still have fuel in another.

That’s why people get into trouble when they look at one moving average in isolation. The 50 SMA might be saying, “This is stretched”, while the 10 EMA and 21 EMA are still telling you the trend is alive.

In that kind of situation, the right response isn’t to panic. It’s to pay attention.

AMD kept running even as the extension cloud started to thicken. That told me the stock was strong enough to absorb the stretch. The real question wasn’t, “Is it extended?” It was, “How many of the anchors are now showing strain?”

That’s a much more useful question.

---

CAR: when the rubber band really snaps

CAR gave a very different picture.

That was one of those charts where multiple moving averages were all flashing at once. The 10 EMA, 21 EMA and 50 SMA were all stretched. That’s where the rubber band analogy really earns its keep.

When the shorter anchors are all showing the same kind of stress, the stock is no longer just extended. It’s becoming mechanically vulnerable.

That doesn’t mean the reversal must happen immediately. But it does mean the stock has reached a point where the probability of continuation is lower and the probability of exhaustion is much higher.

That’s the sort of move where you stop thinking in terms of chasing and start thinking in terms of protecting capital.

And if you’re looking for the moment when a stock is most likely to capitulate, that simultaneous stretch across multiple timeframes is exactly what you want to watch for.

Apple, revisited

So let’s come back to Apple.

The important point here is that I didn’t need the extension cloud to know Apple still had room to run. Jeff’s heuristic already told me that. Once a stock gets up into that 7.5x to 8x ATR zone from the 50 SMA, that’s where the odds start shifting against fresh chase entries.

That was the original insight. The cloud doesn’t replace it, it just makes it easier to see.

What the extension cloud gives you is a visual way to watch the stretch build in real time. Instead of relying on a number in a column, you can see how price moves through the bands over time, how the candles behave inside the cloud and when the move starts to look crowded.

On Apple, that matters because the stock is not yet fully compressed across every anchor. The 10 EMA and 200 SMA are still relatively calm, whilst the 21 EMA and 50 SMA are starting to show more stretch. That tells you the move is still alive, but the pressure is building.

So the message is simple: the heuristic told me there was still space. The cloud lets me see that space.

That’s why the cloud is useful. It turns a rule of thumb into something you can watch unfold on the chart.

The heuristic gives you the threshold. The cloud shows you the stretch.

Stock-specific thresholds

The cloud had already been useful, because it let me visualise extension instead of just reading a number. But I wanted to take it one step further. I didn’t want to rely only on thresholds built from the full universe of stocks. I wanted a model that could adapt to the stock in front of me.

What if you could treat each stock on its own merit?

What if, instead of using one fixed set of thresholds for every ticker in the market, you could look at a stock’s own previous behaviour and ask, what is normal for this ticker?

That was the exam question.

That’s the point of ‘adaptive thresholds’. It doesn’t compare Apple to the whole market. It compares Apple to Apple’s thresholds. It uses that stock’s own price history to work out what stretched, extreme or exhausted actually looks like for that name.

So when you look at the chart, the comparison is the point.

On the left, you’ve got the out-of-the-box model, where Apple is measured against the broader population. On the right, you’ve got the adaptive model, where Apple is measured against its own historical behaviour. Same stock, same chart, two very different ways of reading the stretch.

That matters because stocks do not all behave the same. Some are volatile. Some are smooth. Some can keep stretching for ages before they mean-revert. Others turn much sooner. A stock-specific threshold lets you see that difference clearly.

So the real question is not just whether the stock is extended.

It’s whether it is extended for that stock.

And that’s what makes the adaptive model such a big step forward.

A stock like Apple will have one personality. A stock like SNDK will have another. MU will behave differently again. HOOD will have its own rhythm.

A stock-specific model lets you study the stock on its own terms.

And once you do that, the extension cloud becomes much more powerful.

SNDK, MU and HOOD: studying the winners

If you really want to learn how to use this properly, study the winners.

That’s where the real lessons are.

Big winners do not move in a neat straight line. They stretch, pause, re-accelerate and stretch again. If you go back through historic multi-baggers, you can see how price behaves as it pushes deeper into the cloud. That tells you a lot about how much fuel a move still has and when the pressure is starting to build.

The point is not whether extension exists. It’s how far a stock can stretch before the move finally gives way.

That’s why SNDK, MU and HOOD are so useful. They show you three different versions of the same basic story. SNDK shows a strong trend that kept finding fuel even as it stretched. MU shows a steadier move that ground higher, paused and then expanded again without falling apart. HOOD shows what happens when momentum and narrative come together and the stock starts pushing through the cloud faster than most traders expect.

That’s the kind of chart you want to study.

The important thing these charts show is this: big movers often keep stretching longer than you expect before they finally snap back. That doesn’t mean you chase them blindly. It means extension is not just a warning light, it’s a way of reading pressure.

How to use this in practice

Here’s the practical part.

First, identify the anchor you care about. For swing trading, that’s often the 21 EMA or 50 SMA. For longer context, the 200 SMA matters more.

Then look at the ATR multiple.

If the move is still inside the normal range, fine. If it is stretching, pay attention. If multiple moving averages are stretched at the same time, take that seriously.

A simple framework would look like this:

• 10 EMA stretched: short-term momentum is hot, but not necessarily finished.

• 21 EMA stretched: swing momentum is becoming extended.

• 50 SMA stretched around 7.5x to 8x ATR: the move is entering statistically rare territory.

• 200 SMA stretched by 15x or more: you are looking at a major structural move or a late-stage exhaustion event.

Then combine that with price action.

If the stock is stretched and you see a reclaim candle, a hammer or a bullish engulfing bar after a washout, that’s when the setup becomes interesting.

If it’s stretched and all the moving averages are flashing at once, that’s when you start thinking about exits, not entries.

The real lesson

The biggest mistake traders make is thinking they need to be first.

They don’t.

They need to be right enough often enough.

ATR extension analysis helps with that because it keeps you from forcing trades at the wrong point in the move. It also stops you from dismissing great trends too early just because one moving average looks stretched.

You want to know when a stock still has fuel. You want to know when it’s starting to overheat. You want to know when the rubber band is still elastic and when it’s about to snap.

So the next time you look at a stock like Apple, or AMD, or one of the real monsters like SNDK, MU or HOOD, don’t just ask whether it’s up a lot.

Ask how stretched it is.

Ask which moving averages are stretched.

Ask whether you’re looking at a normal trend, an extended trend or a stock that has entered the sort of zone where prices have run too far, too fast.

And most of all, don’t confuse extension with certainty.

It isn’t certainty.

It’s a map of probability.

One more thing: scanning for extension

There’s one more capability worth knowing about.

Once you understand extension thresholds, the natural next step is to scan for them. Instead of manually checking stocks one by one, you can run a screener that finds every name currently sitting beyond a specific ATR multiple across any combination of moving averages.

That’s exactly what SwingTheStrat lets you do.

The ATR extension scan inside the platform lets you build fully customised filters around ATR multiples. You can screen for stocks that are beyond 2.5x on the 10 EMA, beyond 4x on the 21 EMA, beyond 7.5x on the 50 SMA or beyond 16x on the 200 SMA. You can combine them with ANY logic (at least one condition true) or ALL logic (every condition must be true). You can mix and match however the setup demands.

The screener shown here, for example, is scanning for stocks where any of the four anchors has breached its approximate +3 sigma threshold simultaneously. That’s the multi-anchor trap scan, and it’s exactly the kind of filter that surfaces the CAR-style setups before the reversal happens, not after.

The individual stock view goes even further. You can see the ATR multiple for each anchor at a glance, alongside the slope, the status and the zone. ARM, for example, shows a 10 EMA at 2.7x, a 21 EMA at 3.8x, a 50 SMA at 6.0x and a 200 SMA at 7.7x. All four anchors in HOLD territory, all climbing. That’s the kind of picture that tells you a move is real and building, not just a one day spike.

You don't need to build it yourself. It's already there, running across more than 2,700 tickers every day, ready to surface the setups when they appear.

And if you enjoyed this article, please consider supporting the work by picking up a copy of TheStrat Protocol. TheStrat Protocol is currently ranked #36 in Best Sellers in Stock Market Investing on Amazon, which is beyond anything I could have imagined. If you already own a copy, an honest review and a share in your trading communities would mean the world.

---

If you found this article valuable, here’s the follow up article on how to detect ‘Super Momentum’ Stocks based on ATR extension:

---