What Is a Market Trend? Uptrend, Downtrend and Sideways, Read Swing by Swing

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-14 · Updated 2026-11-02 · Reviewed 2026-11-01

A market trend is the direction of a market’s swing points: higher highs and higher lows make an uptrend, lower highs and lower lows make a downtrend, and a mix of the two is a sideways market. On SPY’s daily chart since 1993 that rule read “up” on 42.53% of sessions, and the label always arrived after the move it described.

Most definitions stop at the first sentence. The useful part is what happens when you apply it to real prices one session at a time, because that shows how often a trend exists, how late you learn about it, and what the label does and does not tell you.

How to read a trend from swing points

Start with swing points, not candles. A swing high is a bar whose high is above the highs of the bars on either side of it; a swing low is the mirror. This page uses five bars each side, a common daily setting, and the higher high page covers how the choice of that number changes what gets picked.

Then compare the last two of each. There are only three outcomes:

• Uptrend: the latest swing high is above the one before, and the latest swing low is above the one before.
• Downtrend: both are lower.
• Mixed or sideways: one is higher and the other lower, the shape of a trading range or a market turning.

Two people applying this to the same chart will get the same answer. That is its main strength over reading a trend by eye, and it is the same test that market structure is built on.

The three types of trend also come in lengths. The same rule run on weekly bars finds the long swings behind a bull market or bear market; run on hourly bars it finds swings that last a day or two. A 10% to 20% drop inside a long rise is usually called a correction, and the long trend can survive it.

A worked example: SPY, July to October 2026

Here is the rule applied by hand to SPY’s daily bars over four months. Each swing point is the session’s high or low, and a swing is only usable five sessions after it prints, once the five bars to its right exist.

Late July: a downtrend that showed up at the top. The swing high on 15 July, $755.58, was below the 15 June high of $756.68, and the swing low on 29 July, $729.10, was below the 8 July low of $739.51. That 29 July low was only confirmed on 5 August, so the rule switched to “down” on 5 August, a day SPY closed at $769.79, which is $14.21 or 1.88% above the $755.58 high.

August: back to up in two steps. The 5 August high of $776.85 cleared $755.58, a higher high, confirmed on 12 August, so the reading went to mixed. Then the 20 August low of $762.04 sat above $729.10, a higher low, confirmed on 27 August, and the rule read “up” with SPY at $771.10.

September: down again without a crash. The 28 August high of $775.30 was below the 13 August high of $779.37, a lower high, and the 1 September low of $759.48 was under $762.04, a lower low. Confirmed on 9 September, the reading turned “down” with SPY at $762.40, only 2.18% under the 13 August swing high of $779.37.

October: up once more. The 1 October low of $758.79 was above the 16 September low of $749.60, and the 6 October high of $781.62 was above the 22 September high of $775.14. Both confirmed by 13 October, when the reading went back to “up” at $778.01.

Count the changes: eight in four months, while SPY’s whole range from the 29 July low to the 6 October high was $52.52, or 7.20%. The rule did exactly what it says. It is the market that kept producing swings close enough together to flip it.

Why every trend is identified late

A swing point cannot exist until the bars after it do. With five bars on each side, the earliest a swing low can be known is five sessions after it happened, and a trend needs two of them, so the label trails the turn by at least a week on a daily chart.

That lag is built into the definition, not a defect in it. Any rule that named a trend sooner would be naming something that had not yet met the test, which is a forecast, not a reading.

The worked example shows what the lag costs. On 30 October SPY closed at $755.72 with a low of $750.36, under the 1 October swing low of $758.79, yet the rule still read “up” because no new swing low can be confirmed until five more sessions have traded. Anyone reading the chart that afternoon saw a break the rule could not yet count.

A trailing stop or a moving average trades this lag for a different one. Each reacts to price sooner or later depending on its length, and none of them removes the gap between a turn and the evidence of it.

Why the timeframe decides the answer

The same price history holds several trends at once, and they can disagree. An hourly chart can be making higher highs inside a daily downtrend, and both readings are correct because they measure different swing points.

Fix the timeframe before looking. Decide which horizon you trade, read the structure there, and accept that other horizons will sometimes say the opposite. The failure is picking the timeframe after seeing which one agrees with the position already held, which multi-timeframe analysis is meant to prevent.

Shorter charts show more noise for the same reason. A one-minute chart flips between states as often as a daily one, but each swing is worth far less against the cost of a round trip.

The original data

How often SPY was in each state

We ran the five-bar swing rule over every SPY daily bar from 29 January 1993 to 30 October 2026. The first reading was possible on 7 April 1993, which leaves 8,450 sessions. Prices are Yahoo Finance daily highs, lows and closes, not adjusted for dividends; days with no range or no volume were dropped.

A fund that rose for three decades was in a clean uptrend on fewer than half its sessions. The typical up stretch lasted 18 sessions, under a month, and a third of all sessions had no trend at all by this test. Those mixed stretches are the condition that trend following systems pay to sit through. Every session’s reading is in the SPY trend-state data.

Did the reading say anything about the next month?

We then checked whether SPY closed higher 20 sessions later, split by the reading on the day. The base rate on all 8,430 sessions with 20 sessions after them was 5,391, or 63.95%.

After an up reading SPY was higher 20 sessions later 2,308 times out of 3,580, or 64.47%. After a down reading it was 1,283 of 1,921, 66.79%, and after a mixed one 1,800 of 2,929, 61.45%. None of these is far from the base rate, and the down reading came out slightly above the up reading, not below it.

These gaps are within what chance produces. Neighboring 20-session windows share 19 of their days, so the samples are far smaller than the session counts suggest. The plain reading is that on SPY’s daily chart the trend label summarized the last few weeks and carried no measurable information about the next four.

Do up days come in runs?

The same lesson shows up one bar at a time. We took every S&P 500 daily close from 3 January 1950 to 25 September 2026: 19,305 closes and 19,304 changes. The index closed up on 10,256 days, 53.1%; the 125 unchanged days count as not up. A run is an unbroken string of up days, and there were 4,565 of them.

Long runs turned up at almost exactly the rate coin flips would give. The longest was 14 straight up days, from 26 March to 15 April 1971. The full table is in the S&P 500 run-length data.

A streak did not predict the next day either. After three straight up days the next closed up 1,631 times out of 3,087, 52.8%, against 53.1% for all days; after five it was 462 of 868, 53.2% (two-sided binomial p = 0.75 and 0.97). After one to five up days the shares were 55.5%, 54.2%, 52.8%, 53.2% and 53.2%, on 10,255, 5,691, 3,087, 1,631 and 868 days.

The one exception has faded. After a single up day the next was up 55.5% of the time across the whole record (p < 0.001), but that came from the older decades: 58.5% against 52.6% for all days from 1950 to 1989, and 52.3% against 53.7% from 1990 to 2026.

When it fails

Calling a trend on two bars and sizing for a long one. A two- or three-day run is the kind chance produces every few weeks; on the S&P 500 three up days said nothing about the fourth.

Treating the label as a forecast. On SPY the down reading was followed by a higher close a month later as often as the up reading was. The structure describes where price has been.

Reading one broken low as the end. One lower swing point turns an up reading to mixed, not down, and in the 2026 example both down readings were back to up within five weeks.

Switching timeframe until a trend appears. Some timeframe always shows one, and choosing it after the fact fits the frame to the answer. Position sizing built on that choice inherits the error.

Confusing a trend with a reason. Higher highs say price rose; they contain nothing about why, or whether it continues.

Questions people ask

What are the three types of market trend?

Uptrend (higher highs and higher lows), downtrend (lower highs and lower lows) and sideways, where the two disagree or price holds inside a range. On SPY’s daily chart since 1993 the split was 42.53%, 22.77% and 34.70% of sessions.

How do you identify a market trend?

Mark swing highs and lows with a fixed rule, such as five bars each side, and compare the last two of each. Tools like ADX measure strength and the choppiness index flags trendless stretches, but the swing test is the definition they approximate.

How long does a market trend last?

It depends on the timeframe. On SPY’s daily chart the median up stretch was 18 sessions and the longest 143; the median down stretch was 15 sessions and the longest 76.

Was SPY in an uptrend at the end of October 2026?

By the five-bar daily rule, yes: the reading turned up on 13 October 2026 and still read up at the 30 October close of $755.72, though that close sat below the 1 October swing low of $758.79 and the next confirmed swing low could change it.

Related

Volatility covers how far the bars move while a trend does or does not exist, and breakout measures how often a move through a level keeps going. Support and resistance covers the levels the swing points are measured against, and a trend line is a drawn summary of the same swing lows or highs.

A break of structure is the event that confirms a new extreme, and a change of character is the first sign the sequence has stopped. Mean reversion is the opposite bet, made in the mixed stretches.

To act on one, how to trade a trend sets out the pullback method, and trend analysis compares three ways of deciding a trend exists.

This page is educational, not financial advice. Test every idea on your own charts before risking money.