A stock breaks out. A new three-month high, sometimes a new all-time high. The chart is obvious. Everybody can see it.
Do you buy it here, or wait for it to come back to its 50-day average?
Most people wait. Buying at a new high feels like paying up, and the pullback offers a better price and a line to lean on. It is the patient choice, and it sounds like the safe one.
But is it the right one?
Apple, 2018. One stock, one move, two ways in. The question is which dot you want to be.
I ran both entries on every breakout in my coverage universe since 2010.
Drumroll…
Waiting loses money. It lost on calm stocks and wild ones, on index funds and single names, and on every definition of a pullback I tried. The reason is a piece of arithmetic sitting inside the 50-day average (or whatever average you use) itself.
What I mean by breakout and pullback
The same Apple move, labelled. The blue zone is the pullback zone, 3% either side of the gold line.
A moving average is the average closing price of the last fifty days, redrawn daily. It is a smoothed version of where the stock has been. It also moves, which will matter more than anything else here.
A breakout is a close higher than every close of the previous three months. That is all. No chart patterns, no ranges, no all-time highs required. Whatever shape the stock made getting there, a new 60-day closing high counts.
A pullback is price coming back into a zone around the 50-day average. It has to be a zone, because price almost never lands exactly on an average. I used 3% either side, so any close between 3% above the line and 3% below it counts as a touch. That answers the obvious question: a close 2% under the average is a pullback here, same as one 2% over. In practice, the first touch usually lands just above the line, a median of 2.2% above, because price is coming down from higher up and the rule fires on the first close inside the zone.
Why 3%? It is the middle of the five widths I tested. I ran nine definitions in total, from 0.5% up to 5% and four more scaled to how jumpy the stock is. Every one gives the same answer, and I show all nine later.
Last term. An episode is the life of one complete move. It opens on the breakout and ends once the stock has spent five straight days below its 50-day average. The sale happens on the next close. Five days is a guess, so I tested three and ten as well, but their outcomes are more or less the same.
The test
Two buyers, one stock.
Buyer One buys at the close of the breakout day. Buyer Two waits and buys at the close of the day price first touches the zone. Both sell at the same close, once the episode has ended. Same stock, same trend, same exit. The only difference between them is the day they enter the trade.
If the pullback never comes, Buyer Two never enters the trade, but that missed episode counts against him.
Apple again. Buyer one paid 43.11 at the close of 4 May and was in for 125 trading days. Buyer two waited 31 days, paid 43.71, and was in for 94. Both sold at the close of 31 October at 51.70.
The example above is just one instance, but it is there to make the rules clear.
Now the study. Every breakout on all 193 tickers I cover, 5,224 of them, from 2010 to August 2026. Across that set, Buyer One is in the market for a median of 48 trading days per trade. Buyer Two waits a median of 21 days and then holds 23, so they own the stock for about 60% of the days Buyer One does.
The answer
Each bar counts trades. The number along the bottom is buyer two’s return minus buyer one’s, on the same trade. Anything right of zero is a trade where waiting won.
The median trade came out 0.9 points ahead of waiting. Waiting beat chasing on 57.3% of trades. So far the “wait-for-pullback” advice is holding up.
Now the mean. Add every outcome together and divide, and waiting comes out 4.4 points behind. Weighting each of the 193 names equally, so no single busy ticker can dominate, the cost is 5.35 points per trade, with a range of 3.57 to 7.90 and no path to zero.
Look at the shape of the chart, and you can see why the median and the mean disagree.
The right side is squat and wide. The left side stretches out a long way.
When waiting wins, it wins an average of 5.0 points. When it loses, it loses an average of 17.0 points. The best 5% of trades gained 11.6 points or more. The worst 5% gave up 25.4 or more.
That asymmetry is the trade you are actually making. On a good day, you save a little on the entry price, and there is a limit to how much you can save, because the stock only falls so far before the trend is over. On a bad day, the stock leaves without you and keeps going, and there is no limit to that at all.
Waiting wins most of the arguments but loses the war.
Why waiting loses: the 50-day average rises while you wait
Here is what actually happens while you wait.
The 50-day average rises while you wait. It rose during 98.7% of the waits in this study. So two things are moving towards each other, and where they meet is the price you pay.
Two numbers decide it.
The Price-MA gap is how far above the 50-day average the stock sat on the breakout day. Median 7.8%.
The MA drift is how far that average climbs before the two meet. Median 4.2%.
Two real trades. Watch what the gold line does between the two dots.
Nvidia broke out on 21 May 2024, sitting 7.7% above its 50-day average. The pullback arrived 38 days later. In those 38 days the average itself climbed 29.9%. Price had 7.7% of room to give back and the average came up nearly four times that far, so the two met well above where buyer one got in. Chasing returned 22.7%. Waiting returned minus 0.8%.
Apple in August 2022 is the reverse. The price-MA gap was 14.1%, the average only climbed 6.9% in the 19 days it took, and buyer two got a genuine discount. Waiting won by 4.3 points.
Every trade where a pullback arrived. The diagonal is where the two are equal. Below it the price-MA gap was bigger. Above it the MA drift was bigger.
Almost every blue dot sits below the diagonal, and almost every gold dot sits above it.
That is the finding in one picture:
If you knew only which of the two numbers (Price-MA Gap vs. MA Drift) was bigger, and nothing else about the stock, you would call the outcome correctly nearly every time.
On 45.2% of the trades where a pullback did arrive, the pullback price was higher than the breakout price. Nearly half the time you wait a month for a discount and pay more.
Averaged out, the price-MA gap is 9.6% and the MA drift is 11.8%. The average climbs further than the price falls, and that one comparison is the entire result.
A dip is a race. Price has to fall further than the average climbs, and on the things worth owning, it usually fails to.
Can you screen out the bad trades?
That is the obvious next move. If a fast-rising average is what costs you, screen those breakouts out and wait only on the rest.
I tried four screens. Each one uses information you have on the breakout day. How steep the 50-day average was over the previous 20 days. Whether the 50-day sat above the 200-day. Whether price sat above the 200-day. And how big the price-MA gap was to start with.
Fourteen screens, each one checkable on the breakout day. Every bar below zero is a screen that still lost.
That makes fourteen groups of trades, and waiting loses in all fourteen. In thirteen of them, the range never reaches zero. The best group of the lot is the one where price started closest to its average, and it still costs 1.07 points a trade.
One group runs against instinct. You would expect the worst case to be a steeply rising average, because a rising average is what eats the discount. The worst is a falling one, at -9.25 points.
Picture that setup. The stock has crashed, gone quiet, then jumped to a three-month high while its 50-day average still points down. Then the moving average turns. It climbs 21.2% off a low base, and the “pullback” becomes a higher price.
Nothing I tested makes waiting the better trade. I expected the 200-day screen to sort the good from the bad, but it doesn’t.
Nor does the way I defined things. Two of my choices were arbitrary, so I re-ran everything without them.
The first is how close price has to get to the 50-day average before it counts as a pullback. I used 3%. I also ran 0.5%, 1%, 2% and 5%, plus four more that widen the zone on a jumpy stock and narrow it on a calm one.
The second is what counts as a breakout. I used a new three-month high. I also ran a cross back above the 50-day average and a new three-month high that is also above the 200-day.
Nine pullback definitions times three breakout definitions is 27 versions of the test. Waiting loses in all 27.
Nine ways to define a pullback, three ways to define a breakout, measured identically.
Where it hurts most
Volatility measured over the 60 days before each breakout.
The mechanism predicts this before you look. Drift is the speed of the average, and averages move fastest on things that move fastest. So the cost of waiting should rise with how much a stock swings about.
It does. Sort the breakouts by how much each stock tends to move. Waiting cost 1.63 points on the calmest third. On the wildest third, which swings about three times as far, it cost 5.99.
The same pattern shows up in what you are buying. A broad ETF cost 0.78 points. A single stock cost 4.30. Wait for pullbacks only on index funds and the habit is far cheaper, but it is still a cost.
Bitcoin is the mechanism in caricature. Chasing its breakouts returned 65.4% per trade, and waiting gave up 37.0 points of it. Tesla gave up 15.7.
Google is the only one of the twelve where waiting won, by 0.3 points over 30 trades, which is close enough to nothing.
What to do with this
Three lessons, in the order I trust them.
Judge the price-MA gap against how fast the thing moves. A quiet index fund has a slow-moving average, and the discount can arrive. A semiconductor in a mania has an average that climbs fast, and waiting will likely not provide a better entry.
If you wait, put a clock on it. Twenty days, then buy or drop it. That recovers 40% of the loss for no cleverness at all.
Ask whether you wanted to own it. If the answer was yes on the breakout day, the entry-timing question is smaller than it feels, and the record says getting cute costs 5.35 points a go.
What the next piece covers
This one asked when to buy. The next one asks whether the 50-day average is worth watching at all.
Three things come out of it. A touch of the 50-day is followed by a real excess return over the following week, measured against a control that strips out momentum. The first touch of a move is worth more than double any later touch, and that survives a check for whether it is simply measuring the age of the trend. And when I sweep every average length from 20 days to 200, the 50 sits at the top of the range. After two pieces in which a famous number turned out to be folklore, this one earns its reputation.
I run daily Systematic Portfolios and Multi-model Signals, including deep dives into everything they hold. We aim to beat the S&P 500 in a systematic manner, using liquid, easy-to-trade names without any narratives.
You can see how they work now.
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.