Trading vs Investing: What SPY's Sessions and Nights Show
Trading is buying and selling over short holding periods to profit from price moves; investing is buying and holding for years to own the growth of the assets. On SPY from 1993 to 2026, most of the price gain arrived between one close and the next open, outside the hours a day trader holds.
Trading and investing can use the same market, the same broker and the same ticker. What separates them is how long a position is held, and that one choice decides which part of the market’s movement you are exposed to. SPY’s own record from 1993 to 2026 shows how much that matters.
What each one is
Trading is buying with a plan to sell over a short horizon, minutes to weeks, and the profit has to come from the price move within that window. Day trading is its fastest form: every position closed before the session ends, so nothing is held overnight.
Investing is buying to hold for years, with the return coming from the growth of the business or the market and from dividends. Buy and hold is its plainest form, and investing basics covers the accounts and decisions around it.
Both are bets on price, and neither is free. The trader pays costs every round trip and needs an edge that survives them. The investor pays far fewer costs and accepts sitting through every fall.
Where they differ
Holding period. A trader is exposed to a slice of time; an investor is exposed to all of it, including the nights, weekends and the worst weeks.
Number of decisions. A day trader makes a buy and a sell decision every session. A buy-and-hold investor in SPY from 1993 made one purchase.
Costs. Every round trip crosses the bid-ask spread and may pay a commission, so costs scale with the number of trades rather than with the size of the gain.
What has to go right. An investor needs the market to rise over years, which it has done in most long windows of its history. A trader needs a repeatable edge over a short window, which is a much narrower and harder claim to prove; the is day trading worth it page shows how many sessions it takes to tell a small daily edge from noise.
Sessions versus nights
SPY’s price gain did not arrive evenly through the day. Split every day from 1 February 1993 to 25 September 2026 into two parts: the session, from the open to the close, and the night, from the previous close to the open. Multiplying the session parts together gives what a position held only in market hours earned; multiplying the night parts gives what a position held only while the market was shut earned. The two together give buy and hold exactly.
| Held | $10,000 became | Gain | A year, compounded |
|---|---|---|---|
| Every session only (open to close) | $12,772.75 | +27.73% | 0.73% |
| Every night only (close to next open) | $137,445.88 | +1,274.46% | 8.10% |
| Both: buy and hold, price only | $175,556.18 | +1,655.56% | 8.89% |
| Buy and hold with dividends reinvested | $320,680.71 | +3,106.81% | 10.85% |
Sessions were still up more often than down: 52.31% of the 8,471 sessions closed above their open, against 54.90% of nights. The difference in where the money came from was in the size of the moves, not only in how often they went up.
The split held in every stretch, at different sizes. From 29 January 1993 to 29 December 2000 the sessions lost 31.38% while the nights gained 335.10%. From 2000 to 2010 the sessions lost 11.54% and the nights gained 8.36%. From 2010 to 2020 both rose, the sessions 49.57% and the nights 98.78%, and from the end of 2020 to 25 September 2026 the sessions rose 40.67% and the nights 46.66%. In each stretch the night part was larger.
A worked example
Take a hypothetical trader who bought SPY at every open and sold at every close from February 1993, starting with $10,000 and paying nothing to trade. That is the sessions-only row: $12,772.75 by 25 September 2026, and as low as $4,057.47 on 6 March 2009.
Now add a cost. Suppose each round trip cost 0.02% of the position, two basis points, for spread and commission together. Over 8,471 round trips that keeps (1 − 0.0002)^8,471 = 0.1837 of the money, and the $12,772.75 becomes $2,346.55. At one basis point a round trip it becomes $5,474.89.
The investor who bought once on 29 January 1993 at $43.94 a share and held to the $771.35 close of 25 September 2026 paid one spread and ended with $175,556.18 on price alone. Neither path is a forecast; the gap between them is the cost of being flat every night plus the cost of trading every day.
Missing the best days
An investor who is out of the market on a few days can lose a large share of the result. Using SPY’s total return, $10,000 from 29 January 1993 grew to $320,680.71. Remove the 5 best days and it becomes $191,661.29; the 10 best, $139,148.85; the 20 best, $80,405.97, a compound rate of 6.39% a year instead of 10.85%.
The best days sat next to the worst ones. All ten best days came in 2008, 2009, 2020 and 2025: the largest was 13 October 2008, up 14.52%, and the most recent was 9 April 2025, up 10.50%. Seven of the ten came within ten sessions of one of SPY’s ten worst days. A trader who stepped aside after a bad day would have been out of the market for some of them, because that is exactly when these days arrived.
This cuts both ways and should be read that way. Missing the 10 worst days instead would have turned the same $10,000 into $768,916.23. Both sets are known only in hindsight, which is the point: the investor’s result depends on not trying.
Which one to use: trading or investing?
Use investing for money you can leave alone for years, when you want the market’s return rather than a bet against other traders. On the S&P 500 since 1950, 74.6% of one-year holds and 93.1% of ten-year holds ended higher, against 53.1% of single days, per the holding-period figures behind the market timing page.
Use trading only when you have a written rule, a record of it and an edge measured after costs, and only with money whose loss you can absorb. The sessions-only line above is the honest starting point: on SPY, the hours a day trader holds have given almost nothing back on their own.
Some people do both, with separate accounts and separate rules, so that a trading loss never forces a long-term holding to be sold. That separation matters more than the label.
The original data
The sample. SPY daily open, high, low, close and adjusted close from Yahoo Finance, 29 January 1993 to 25 September 2026: 8,472 sessions, 33.65 years. Session part = close ÷ open of the same day; night part = open ÷ the previous close. The product of all 8,471 pairs equals the 25 September 2026 close divided by the 29 January 1993 close exactly, so nothing is lost or double-counted. Over those 33.65 years the night part compounded at 8.10% a year and the session part at 0.73%.
Price only for the split. The ex-dividend drop in SPY’s price falls in the night part, because the share opens lower on that morning, so the night figure is if anything understated. The total-return row and the best-days figures use the adjusted close, which includes dividends.
The figures are in two files: the stretch-by-stretch split and the month-end values behind the chart.
One caution on the early years. This page cannot check the 1990s opening prices in Yahoo’s file against a second source, and the 1993-2000 stretch shows the widest gap. The three later stretches, built from the same file, point the same way at a smaller size, so the result does not rest on the early data alone.
When it fails
The investing side fails when the money is needed early. Buy and hold pays across years, and most of the best and worst days above arrived together in 2008 and 2020. A holder who must sell during a stretch like that turns a paper loss into a real one.
It also fails when “investing” means one stock. These figures are an index fund. A single company can fall and not recover, which the index absorbs by replacing it.
The trading side fails on costs first. The worked example shows two basis points a session turning $12,772.75 into $2,346.55. A method that trades every day has to clear that before any skill shows up.
It fails a second time on the night. A day trader who is flat every night gives up the part of the day that, on SPY, carried most of the price gain, so any edge has to come from inside the session.
And this page fails if it is read as a strategy. It measures when SPY’s gains arrived. It does not say the next 33 years will split the same way, and holding only overnight would itself be a trading method with a round trip, and its costs, every day.
Related
Trading covers the decisions inside a single trade, and buy and hold covers the other end of the scale. The is day trading worth it page measures how long it takes to prove a daily edge on SPY, and why traders lose money collects the research on how most short-term traders actually did.
Before I decide how often to trade something, I look at when its gains actually arrived. If most of them came while the market was closed, a method that is flat every night has to find its returns somewhere else.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.