← Back to all studies

Study · AMZN · against QQQ

Every $AMZN Drop of 10% or More Since 2014

Fifteen drawdowns, from 10.8% to 56%. Only four of them were the whole market falling, and the one it is in right now is not.

Data through the close of 23 September 2026

Published Sep 23, 2026, 8:55 PM ET

Data through Sep 23, 2026

TL;DR

  • Since 2014 $AMZN has fallen at least 10% from a record high fifteen times, counting the drop it is in now.
  • The completed ones bottomed anywhere from 10.8% to 56.1%. There is no usual depth.
  • Four were market-wide: Q4 2018, COVID, the 2022 bear and the April 2025 tariff shock. Strip those out and the finished ones still run from 10.8% to 30.5%.
  • Of the ten completed episodes that ever reached 16%, seven kept going past 21%.
  • In the drop it is in now, $AMZN is down while $QQQ is up over the same window. This one is not the market.

A claim goes around that $AMZN tends to find a bottom once it is down 16 to 20 percent. It is a tidy idea and it is easy to check, because every drawdown it has had is a matter of record. So we listed all of them.

There is no usual depth

Since the start of 2014, $AMZN has closed at least 10% below a record high on fifteen separate occasions. Fourteen of those have finished. They bottomed at 10.8%, at 56.1%, and at ten different places in between.

Only three of the fourteen finished anywhere in the high teens. The band that gets named as the bottoming zone is where the smallest group of them landed.

Four of them were not about Amazon

Four of the fifteen ran into broad market crises: the fourth quarter of 2018, the COVID crash, the 2022 bear market, and the tariff shock of April 2025. Those four took 34.1%, 22.7%, 56.1% and 30.9%.

It is fair to ask what the rest look like with those removed, so we did that too. The ten finished episodes that are left still run from 10.8% to 30.5%. Taking out the crises narrows the range at the top and changes nothing about the spread underneath it.

Two of the deepest drops survive that filter. In 2014 $AMZN fell 29.5% while the Nasdaq 100 rose, and in early 2016 it fell 30.5% while the index fell half as far. Neither had a bear market to point at.

What actually repeats

Most of the finished episodes stopped in the teens or close to them. The rest kept falling until they had lost closer to a third. What repeats is that split, not a depth.

The sharpest version of the point: of the ten completed episodes that ever reached 16%, seven kept going past 21%. Once $AMZN is down 16%, carrying on through the zone has been the ordinary outcome rather than the exception.

The one it is in now

The current drawdown began at the record close of 3 August 2026 and reached 13.4% below it on 16 September. It has not finished, so it has no final depth and it is listed here without one.

One thing does separate it from most of the list. Over the same window $QQQ is slightly higher. Whatever is happening to Amazon here, the market is not doing it too.

Every episode (15)

From record highTo low$AMZN$QQQ same window
Jan 2014Jan 2015-29.5%+14.3%
Aug 2015Aug 2015-13.7%-12.3%
Dec 2015Feb 2016-30.5%-15.7%
Oct 2016Nov 2016-14.8%-3.5%
Jul 2017Sep 2017-10.8%-1.0%
Mar 2018Apr 2018-14.2%-10.5%
Sep 2018Dec 2018-34.1%-22.4%
Feb 2020Mar 2020-22.7%-25.2%
Sep 2020Mar 2021-16.4%-0.6%
Jul 2021Dec 2022-56.1%-26.8%
Jul 2024Aug 2024-19.5%-10.6%
Feb 2025Apr 2025-30.9%-17.3%
Nov 2025Feb 2026-21.7%-4.7%
May 2026Jul 2026-17.6%-4.8%
Aug 2026Sep 2026still openstill open

An episode marked still open has not finished. It is shown where it sits so far, never as a final figure.

How this was measured

Definition. An episode opens at a record-high closing price measured from 2 January 2014, its low is the lowest close before the first close strictly above that peak, and it counts only if that low is at least 10.0% below the peak. An episode with no subsequent record high is still open and is shown without a final depth. Basis. All figures are CLOSING prices, dividend-adjusted. Intraday lows were deeper in every episode, because a low day's low is at or below its close. Do not compare these numbers with high-to-low figures read off a chart with wicks: those will be larger, and on an intraday basis the episode boundaries themselves would fall on different days. The $QQQ column. It shows what $QQQ did between the same two dates. It is NOT $QQQ's own maximum drawdown over that window, because it is measured from $QQQ's close on Amazon's peak date rather than from $QQQ's own high. $QQQ's own drawdowns were deeper. The column is there to show whether the market was falling at the same time, not to size the market's decline. Sensitivity. The count of fifteen is not fragile: any threshold from 9.6% to 10.8% produces the same fifteen episodes. Source: TradingView daily closes. Every episode was independently recomputed from the raw series before publication.

Sources

Educational commentary, not investment advice. See the full disclaimer.