Day trading vs swing trading backtest: what changes and why

Contents

1. What changes in a backtest when a strategy closes every day instead of holding overnight?
2. How do you model the session-end exit in a day trading backtest?
3. Do swing backtests need to account for overnight gaps and the night session?
4. How do margin and holding costs differ between the two versions?
5. Why does a day trading strategy produce more trades, and does that make the result more reliable?
6. Day version or swing version of one breakout rule: which should you keep?
7. Frequently asked questions
8. References

What changes in a backtest when a strategy closes every day instead of holding overnight?

A day trading vs swing trading backtest differs in five settings: the exit rule, the bars you load, how stops fill across gaps, the margin behind each contract, and the number of trades. Treat the two as separate strategies, test each with its own settings, and compare them per trade after costs.

Day trading (當沖 in Chinese) is a style where every position is closed before the session ends, so no contract is held through the overnight break. Swing trading is a style where a position stays open for days or weeks and is closed by the rule itself, not by the clock. The same entry signal can feed both styles, but the moment you decide whether the trade may live past the close, almost everything downstream in the backtest changes. A trader who flips a "close at end of day" switch and calls the result "the swing version" has tested one strategy and misread it as another.

Setup choice	Day version	Swing version
Exit rule	Forced exit before the session close, plus the strategy's own exits	Rule-based exit only, which may come days later
Data loaded	Day-session intraday bars	Full-session intraday bars or daily bars
Gap handling	No gaps inside a session; a stop fills near its level	A stop can fill far past its level at the next open
Margin assumed	Broker intraday margin	Exchange initial margin for every night held
Trades per year	Hundreds	Tens

Most of the five types of futures algo trading strategies can be built either way, and a trend-following rule in particular behaves very differently once it is cut off at the close every day. If you are still choosing your first rule, the 30-day route to a first algo trading strategy suggests picking one holding style and keeping it fixed for the whole first project, because changing it halfway restarts the research.

How do you model the session-end exit in a day trading backtest?

A day trading strategy backtest needs an explicit exit placed on the last bar the strategy can trade, filled on the bar after it, a few minutes before the session close you will actually use live. The session matters as much as the exit. A US index futures contract such as ES trades almost around the clock on weekdays with a short daily break, but many day traders only keep the cash-equity hours, 8:30 a.m. to 3:00 p.m. Central Time. If the backtest uses one session and the trader keeps another, the forced exit lands at a different price and the whole trade list changes. Exchanges can change session times, so check the exchange's contract page or the futures market hours tool, which shows each session in your own time zone.

Next-bar execution is the practice of placing an order on the bar where the signal appears and filling it at the open of the following bar, because the close of the signal bar is only known once that bar has ended. For a session-end exit this detail is easy to get wrong:

1. Pick the exit time first. Write it as a clock time inside the session, for example 10 minutes before the close, not "the last bar".
2. Find the last bar that opens before that time. With 5-minute bars and a 3:00 p.m. close, an exit at 2:50 p.m. means the order is placed on the bar that closes at 2:50 and fills on the bar that opens at 2:50.
3. Confirm the fill bar is still inside the session. If the order is placed on the final bar of the day, the "next bar" is tomorrow's open, and the position was held overnight without you noticing.
4. Charge the exit like any other trade. A market order at the end of the day pays slippage; one tick per side is a reasonable starting assumption for liquid contracts.
5. Check that no entry fires after the exit bar. A rule that re-enters at 2:55 p.m. and exits at 3:00 is a different strategy again.

When the session, the exit time and the fill bar are fixed, the rest of the intraday work follows the steps in how to build and test a day trading algorithm. Because the forced exit sits inside the trading session you define, AlphaPrime (Valynox) backtesting charges commission and slippage on that exit and fills it on the next bar, the same as every other order in the candidate.

Do swing backtests need to account for overnight gaps and the night session?

Yes: a swing trading futures backtest must decide whether night-session bars are part of the data, and it must let a stop fill at the first available price after a gap, not at the stop level. An overnight gap is the difference between one session's last price and the next session's first price, created while the market was closed or thinly traded. A stop order placed below yesterday's close offers no protection against an open that lands below it; the fill is the open, and the extra loss belongs in the backtest.

The night-session decision comes first, because it changes what "overnight" means. TAIEX futures, for example, trade a day session from 8:45 a.m. to 1:45 p.m. and a night session from 3:00 p.m. to 5:00 a.m. the next morning, Taiwan time, according to the TAIFEX contract page (and the exchange can change these hours). A swing strategy that holds through the night faces one of two worlds:

• Night bars included. A stop can trigger during the night at a price close to its level, but night liquidity is thinner, so slippage should be set higher than in the day session. The gap is then the one between the night close and the day open, which is usually smaller.
• Night bars excluded, or daily bars only. The backtest sees one gap per day, from close to open, and every stop inside that gap fills at the open. This is the simpler model and the harsher one, and it is the right one if the trader does not watch or trade the night session.

Whichever world you choose, the live account must live in the same one. A backtest on day-only bars that holds overnight, traded live with night-session stops, will show a different drawdown than the trader experiences. The trade-offs between day-only and full-session bars in futures backtests cover the data side in detail. On the rule side, make sure the backtest engine fills gap-through stops at the open; if it fills them at the stop price, the swing version's drawdown is understated, and the day version, which never faces that gap, looks worse by comparison than it should.

How do margin and holding costs differ between the two versions?

Overnight positions need the exchange's full initial margin per contract and carry gap risk through every night, while intraday positions run on the lower intraday margin a broker allows, so the account size that survives the backtest drawdown is different for each version. Initial margin is the deposit the exchange requires per contract for a position held past the session close. Intraday margin is a smaller deposit a broker may allow for positions that are flat before the close; it is the broker's choice and can be withdrawn on short notice.

Item	Day version	Swing version
Margin per contract	Broker intraday margin, lower	Exchange initial margin, higher, re-checked each night
Who sets it	Broker	Exchange, with broker add-ons
Exposure per trade	Minutes to hours, inside one session	Days, including every night and weekend in between
Worst single loss	Bounded by intraday range and slippage	Can exceed the stop by the size of a gap
Account size to size the backtest	Intraday margin plus the day version's drawdown	Initial margin plus the swing version's drawdown plus gap allowance

The practical consequence is in position sizing. The same equity curve read against a small intraday margin suggests many contracts; read against initial margin plus a gap allowance it suggests far fewer. A swing backtest sized on intraday margin is a plan that breaks the first night it is held. How margin turns a backtest drawdown into a required account size is worked through in futures leverage and how margin changes backtest drawdown. Margin requirements are set by the exchange and the broker and change with volatility, so take the current figure from the exchange's contract page before sizing either version.

Holding costs are the second difference. A swing position in a futures contract pays no financing in the way a stock position on margin does, but it does face contract rollover: a trade held across the roll must exit the old contract and enter the new one, which is two extra executions with commission and slippage. A day version never rolls a position. Put the roll into the swing backtest as a cost, or the swing version gets a small advantage it would not have live.

Why does a day trading strategy produce more trades, and does that make the result more reliable?

A rule that must be flat by the close can only hold for hours, so it completes many more trades from the same history than a swing version that holds for days, and the larger trade count makes the average trade and the drawdown figures less uncertain. Reliability here means something narrow: how much the average trade would move if a handful of trades were different. It does not mean the strategy is better. A day version with a tiny edge per trade can be measured precisely and still lose to costs.

The swing side of this is easy to see in a real report. The run below is from AlphaPrime's engine on ES daily bars from 2019-01-02 to 2025-06-30, long and short, one contract, with commission and slippage charged on every trade. It tests a stand-in rule, built from a standard deviation condition that serves as the only entry and exit with no stop, target or filter, not the breakout rule discussed in the rest of this article. It is also the one candidate the search kept out of the 1,024 it tried, chosen for its return against its drawdown. Picking the kept one after the fact flatters it: the figures show what that single candidate did on that history, not what the rule earns or what a reader should expect.

151 trades across six and a half years is a few trades a month. An intraday version of a similar idea on 5-minute bars, forced flat each day, would typically complete several trades a week on the same history. The reason the count matters is how an average behaves. The one formula worth knowing is:

uncertainty in the average trade = spread of single trades ÷ √N

Here the spread of single trades is how far individual trade results usually sit from the average (the standard deviation), N is the number of completed trades, and √N is its square root. The square root grows slowly, so four times as many trades are needed to halve the uncertainty.

Two cautions follow. First, the day version's many trades carry a cost per trade that is a much larger share of a much smaller gross move, so its precision is often precision about a number near zero. Second, the swing version's small sample is made worse when it is the best of many candidates: Bailey, Borwein, López de Prado and Zhu (2014) show that the more variants are tried, the longer the history needed before the best one can be told from luck. A swing version therefore needs either a longer history, a walk-forward design that yields more out-of-sample trades, or a stricter rule for accepting it.

Day version or swing version of one breakout rule: which should you keep?

Run the breakout rule twice, once as a day version with a session-end exit and once as a swing version that holds, then keep the version that passes a rule you wrote before the test, not the one with the larger net profit. A breakout rule is a rule that enters when price moves above the highest high (or below the lowest low) of the last N bars, on the idea that a new extreme starts a move. Both versions use the same entry; the day version adds the forced exit and the swing version adds a rule-based exit such as a close back inside the channel.

The pre-written rule should name the figures and the thresholds. A workable set for comparing the two versions:

1. Gross average trade at least three times the round-trip cost. This protects the result against a slippage estimate that turns out optimistic.
2. Maximum drawdown that the intended account can carry with the margin that version actually needs, intraday for the day version, initial plus gap allowance for the swing version.
3. Enough completed trades that the uncertainty band from the previous section does not reach zero.
4. Overnight exposure the trader is willing to hold. A trader who cannot watch the night session should treat every overnight position as a cost of sleep, not just a number.

Note that the two versions are not rivals that cancel each other. A trader with the account size and the tolerance for gaps may run the swing version; a trader who wants to be flat every night may keep searching for a day rule whose gross edge clears the cost test. What neither trader should do is pick by net profit and then adjust the pass rule to fit the winner.

Your next step: take one breakout rule you already have, write the four thresholds above on paper, then run the day version on day-session bars with a forced exit ten minutes before the close and the swing version on full-session or daily bars with gap-through stops filling at the open. Put both results into the table and see which one passes the rule you wrote before you looked.

Frequently asked questions

What is the difference between a day trading backtest and a swing trading backtest?

A day trading backtest forces every position flat before the session close, uses day-session bars, never faces an overnight gap and can be sized on broker intraday margin. A swing trading backtest lets the rule close the trade days later, must choose whether night bars are in the data, must fill gap-through stops at the next open, and is sized on exchange initial margin. The same entry signal produces a different trade list, drawdown and cost profile in each.

Does a swing strategy backtest need to account for overnight gaps?

Yes. When the market opens beyond a stop level, the realistic fill is the opening price, and the difference between the stop and the open is a real loss. A backtest that fills such stops at the stop price understates the swing version's drawdown. Decide whether night-session bars are in the data as well, because that changes where the gap sits and how large it tends to be.

Why does a day trading strategy produce more trades?

Because it must be flat by the close, each trade can last only part of one session, so the same history holds many more completed trades than a rule that holds for days. The larger count makes the average trade and drawdown figures less uncertain, since the uncertainty shrinks with the square root of the trade count. It does not make the edge larger; the day version's gross move per trade is smaller, so costs take a bigger share of it.

Which is better, the day version or the swing version of the same strategy?

Neither wins across the board. Compare them with a rule written before the test: gross average trade against round-trip cost, maximum drawdown against the account size each version needs, enough trades to trust the average, and whether the trader will hold positions overnight. Keep the version that passes that rule, and do not choose by net profit and then adjust the rule.

References

• Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
• Taiwan Futures Exchange. Mini-TAIEX Futures. https://www.taifex.com.tw/enl/eng2/mTX