How to Trade the 50-day Moving Average Pullback

Last time I asked when to buy a breakout. Waiting for the pullback to the 50-day average costs 5.35 points per trade compared with buying straight away.

This time I take the pullback on its own terms. Suppose you want to trade the touch of the 50-day. How should you do it?

The idea is simple. A stock in a trend dips back to its 50-day average. The average holds. Buyers step in. So the touch looks like a good day to buy.

Here is SPY (the S&P 500 ETF) over the last two years. The gold dots are the touches I count.

SPY comes back to its 50-day far more often than seven times. Its close crossed the line 30 times in those two years. So why only seven gold dots?

Because a touch is a dip into a zone, not a crossing of the line. Three rules decide what counts.

The zone. Price rarely lands exactly on the line. So I draw a zone around it, 3% either side. Rule 1 below tests other widths.

From above. A touch is the first close back inside the zone after a close above it. Price is coming down into the line from higher up. Once inside, it can cross the line again and again without a new touch. From November 2025 to March 2026, SPY spent 87 trading days inside its zone without a single close above it. Price climbing back into the zone from below, as in April 2025, does not count either.

One every 20 days. A pullback often bobs in and out of the zone for a few weeks. I want to count it once. So I count at most one touch every 20 trading days, about a month. SPY dipped into its zone from above 20 times in the window. Thirteen of those came within 20 days of a touch. They are the hollow circles. Why 20 days and not 10 or 15? I come back to that once the main result is in.

For each touch, I take the stock's return over the next five days. Then I subtract what the same stock usually returns over five days, on days with similar momentum. A stock that has just fallen behaves differently from one that has just risen, so this keeps the comparison fair. What is left is the excess return: what the touch itself adds.

In the chart below, the gold band is the five days after one SPY touch.

Over its full history, SPY touched its 50-day moving average 66 times. Its average excess return over the next five days was just 0.06%. One name tells you very little, though. So I ran the same test on all 193 names I cover.

Each result below comes with a range in brackets. It is the 95% range across names. A range that stays above zero is a result I trust.

A little. Across 193 names and 11,583 touches, the average excess return in the five days after a touch is 0.28% (0.14% to 0.44%).

Is 0.28% a lot? On its own, no. It is the extra return from one touch, over one week. A typical stock gives you about four touches a year.

I set the 20-day gap before running any test. A gap of 10 or 15 days would count more dips. So I ran the whole test again, with gaps from none to 30 days.

The smaller gaps give the same answer, only weaker. The excess return is 0.17% with a 10-day gap, 0.22% with 15 and 0.28% with 20. With no gap at all, counting every dip, it falls to 0.04%. Its range then runs through zero.

The reason is the repeat dips. At the 3% zone, the dips a 20-day gap leaves out had an excess return of minus 0.27% (minus 0.46% to minus 0.10%).

Now the rules…

How wide should the zone be? I tested nine widths. Five are a fixed share of the average: 0.5%, 1%, 2%, 3% and 5%. The other four scale with how much the stock usually moves in a day. That daily move is its Average True Range (ATR). The four widths are 0.5, 1, 1.5 and 2 times it.

A wider zone does not mean a deeper dip. It changes when the touch fires. With a 1% zone, the stock has to close more than 1% above the line, then close within 1% of it. So the touch only fires when price is right on the line. With a 3% zone, it fires earlier, on the first close back within 3% of the line. Most of the time, that close is still above the line.

The tight zones are not useful. At 0.5% and 1%, the excess return is slightly negative. At 2% it is positive, but its range still includes zero. From 3% upward, and across all four ATR widths, the excess return is positive, with a range above zero. The 3% zone earns 0.28%.

The 3% touches show why. Four in five fired while the stock was still 1% to 3% above the line. Those earned 0.38% (0.14% to 0.66%). The touches that closed within 1% of the line earned just 0.04%. Their range runs through zero. So the touch pays on the first approach to the line. A close sitting right on it tells you little. That is why 3% is the main zone in this piece.

My intuitive read is this. To count as a touch from above at 3%, a stock first has to close more than 3% above its 50-day. That only happens in a clear trend. A 1% zone scenario also lets in a stock that never got far from its average.

The wide zone also fires earlier, and the positive returns suggest that if a name is in demand, dip-buyers act near the moving average, not wait for it to be right on it.

By the time price reaches the average, the early buying hasn’t held it up, and the stock may look less attractive. The trend is being tested, and the next move is closer to a coin flip.

So the 3% zone buys the first dip in a clear trend. A tight zone buys that dip only after it has fallen all the way to the line.

How long should you hold after a touch? I measured the excess return 1, 5, 10, 20 and 60 trading days later. The gold line is the 3% zone.

The edge peaks at five days, at 0.28%. It is 0.08% after one day and 0.20% after ten. By 20 days, it is down to 0.12%. Its range now runs through zero. By 60 days, it is minus 0.44%. All nine widths are below zero by then.

What if the dip keeps falling? It happens.

At the 3% zone, one touch in five closed below the bottom of the zone within the next five days. Those touches had an excess return of minus 5.44%. The rest earned 2.26%. On the touch day, you cannot tell which is which, so I count them all. The 0.28% average already includes the ones that fell through.

So should you use a stop? I tested the obvious one: sell at the close if the stock closes below the bottom of the zone. It made things worse on all nine widths. At the 3% zone, it cost 0.10% per touch (0.03% to 0.16%). It was hit on about one touch in six. After it was hit, the price on average won back some ground by day five. The stop mostly locked in losses that would have partly recovered.

So the touch tells you about the next week. Hold through it.

Is there anything special about 50 days? I ran the same test on thirteen averages, from 20 days to 200.

At the 3% zone, the 50-day scores highest, at 0.28%. Its neighbours are close behind: 0.24% for the 40-day, 0.22% for the 60-day and 0.20% for the 80-day. The 200-day scores minus 0.04%.

Change the zone width and the top spot moves. On eight of the nine widths, the best average is somewhere from 40 to 80 days. The 50-day comes top on only two. So the 50-day works, but its neighbours work about as well.

The touch does not work every year.

The gold bars are the four years when SPY rose less than 2% or fell: 2011, 2015, 2018 and 2022. In each of them, the touch had a negative excess return on all nine widths. 2022 was the worst, at minus 1.22% (minus 1.82% to minus 0.68%). In the other twelve years, the 3% zone was positive in eleven.

But you only know it was a bad year once the year is over. A trading rule needs something you can check on the day. So I tested three trend filters, each checked at the close of the touch day:

Is SPY above its 200-day average? Is the stock above its own 200-day? Is the stock's 50-day average higher than a month ago?

In gold are the touches each filter keeps. In blue are the ones it drops.

The market filter barely matters. SPY was above its 200-day for 86% of touches, so the filter drops few of them. Within each stock, the touches it keeps beat the ones it drops by just 0.01%.

The two stock filters point the wrong way. Touches with the stock above its own 200-day had an excess return of 0.23%. Touches with it below had 0.61%. The rising 50-day filter is similar: 0.22% with it and 0.53% without.

Those gaps are too noisy to make into a rule of their own. What is clear is that none of the three filters helps. The bad years are real. These filters do not see them coming.

A 0.28% edge sounds tradeable. So I built the simplest version. Buy every touch at the 3% zone, hold five days, then sell. Then compare that with simply owning the stock.

The Sharpe ratio measures return per unit of risk, so higher is better. Buying every touch scores 0.25. Owning the stock scores 0.60. The touch rule beats owning on only 11.4% of names.

The problem is idle money. One stock touches its 50-day about four times a year. So the five-day rule is invested on just 8.9% of days.

To fix that, I spread the money across names. Each day, I put equal money into every name that had touched in the last five days. That covers 188 names, leaving out crypto. The book held about 14 names on a typical day and was almost always invested.

Before costs, this rotation made 27.5% a year against 15.2% for SPY.

Most of that gap comes from the names, not the touches. Simply owning all 188 names equally made 22.1% a year. These are the names I cover today, so the list leans towards stocks that did well.

Against owning the same names, the rotation was still ahead before costs. The touches added 4.9% a year, with a slightly higher Sharpe ratio: 1.23 against 1.13. But the range runs from minus 1.4% to plus 11.3%, so that lead could be luck. Buying the same names on random days, the same number of times, made 23.3% a year on average.

Then come the costs and the work. The rotation trades every day. It holds about 14 names at a time. It turns over its whole book about 69 times a year. At 0.10% a round trip, it made 18.4% a year, with a Sharpe ratio of 0.88. That is below simply owning the names, which takes far less trading.

So there are two ways to use the touch. As a daily system, it roughly matches holding before costs. After costs, it falls behind, and it takes a lot of work. As a timing tool, it costs nothing extra. If you already want to own a stock, a touch is a slightly better day to buy.

1. Use a zone of 3% or wider. The touch pays on the first approach to the line. A close sitting on the line tells you little.

2. Hold about a week, without a stop. The edge peaks at five days. A stop at the bottom of the zone made it worse.

3. Use any average from 40 to 80 days. The 50-day works. Its neighbours work about as well.

4. Skip the trend filter. The touch fails in flat and falling years. The simple filters cannot see those years coming.

5. Use it to time a buy you already want. As a daily system, it roughly matches holding before costs. After costs, it falls behind.

The 50-day touch is a timing tool, good for about a week.

The daily Systematic Portfolio and Multi-model Signals, including the deep dives on every holding, sit alongside this letter.

You can see how they work at yxinsights.com/systematic.

DISCLAIMER: This newsletter is strictly educational. Any information or analysis in this note is not an offer to sell or the solicitation of an offer to buy any securities. Nothing in this note is intended to be investment advice and nor should it be relied upon to make investment decisions. Any opinions, analyses, or probabilities expressed in this note are those of the author as of the note's date of publication and are subject to change without notice.