Strategies we backtested CPI day breakout

Tested by BacktestingMax

CPI day breakout on NQ, ES and gold: 94 CPI days backtested

Does the CPI day breakout strategy work on NQ, ES and gold futures?

No, not in our backtest. CPI, the main US inflation report, comes out once a month at 08:30 New York time. The breakout marks the range: the highest and lowest price from 08:30 to 08:45. It buys when a candle (one bar of the chart) closes above the range, and sells short (sells first, to profit if price falls) when one closes below it. The stop-loss, which closes a losing trade, is at the other side of the range. The take-profit, which closes a winning trade, is twice as far from the breakout candle's close as the stop-loss (2R). Trades still open at 12:00 close then. We tested these basic rules on 94 CPI days from 2018 to 2025, on MNQ, MES and MGC (micro NQ, ES and gold futures), each on 5-minute and 15-minute charts. After trading costs, 1 of the 6 tests made money: MGC on the 5-minute chart, +0.9% of the account in 8 years. The other 5 lost 5.4% to 9.7%. Trades opened before the 09:30 US stock market open lost money in all 6 tests. With about 12 CPI days a year, no test reached 100 trades. So the 'Where it worked' box, which counts only tests with 100 trades or more, shows 0 under 'Made money'.

6 backtests 2018–2025 Trading costs included Tested 9 November 2026

Where it worked 6 backtests
Made money
0
of 6 tests, with the basic rules
Made money in both periods
0
2018–2021 and again 2022–2025
Best backtest
+0.9%
MGC (Micro Gold), 5-minute chart: account growth 2018–2025
Wait for the 09:30 stock market open
0
tests made money in both periods, compared with 0 with the basic rules

The results

Results on every market

Each box shows one backtest of the CPI day breakout with the basic rules, from 2018 to 2025, after trading costs: the % result, and a small chart of the account balance. The solid line is the account balance and the dashed line is the starting balance. Green results made money; red results lost money. Under each result: whether it made money in both 2018–2021 and 2022–2025, or passed all four of our checks (see "How a test passes" further down).

Market 5-minute 15-minute
MNQ (Micro Nasdaq 100) −10% Too few trades −5% Too few trades
MES (Micro S&P 500) −9% Too few trades −7% Too few trades
MGC (Micro Gold) +0.9% Too few trades −6% Too few trades
All the numbers, test by test
Market Chart Trades Win rate Profit factor Result Max drawdown 2018–2021 2022–2025 Monte Carlo drawdown Trading costs
MNQ (Micro Nasdaq 100) 5-minute 77 35.1% 0.73 −9.7% −17.5% −8.4% −1.4% −17.5% $780
MNQ (Micro Nasdaq 100) 15-minute 75 38.7% 0.83 −5.4% −13.4% −3.8% −1.6% −14.7% $720
MES (Micro S&P 500) 5-minute 86 43% 0.75 −8.8% −18.6% −11.1% +2.6% −16.7% $1,650
MES (Micro S&P 500) 15-minute 84 45.2% 0.78 −7.3% −16.9% −9.7% +2.7% −15.3% $1,517
MGC (Micro Gold) 5-minute 89 55.1% 1.03 +0.9% −4.2% −0.5% +1.3% −10.1% $1,103
MGC (Micro Gold) 15-minute 87 52.9% 0.72 −6.4% −7.0% −5.2% −1.2% −11.9% $940

Profit factor is the money won divided by the money lost. Above 1 means the test made money. Max drawdown is the biggest fall from the highest balance. 2018–2021 and 2022–2025 show the result in each period on its own. Monte Carlo drawdown: we shuffled the trades into a random order 1,000 times; in 95% of those runs, the worst drawdown was this number or less. Result is the change in the account, with the risk of each trade set to 1% of the account at that time.

The account balance

The best test and the middle test

Best test MGC (Micro Gold), 5-minute chart +0.9% from 2018 to 2025
Middle test (median): half of the 6 did better, half did worse MGC (Micro Gold), 15-minute chart −6.4% from 2018 to 2025

What we found

What the 6 backtests show

  1. After trading costs, only 1 of the 6 tests made money from 2018 to 2025 with the basic rules: MGC on the 5-minute chart, +0.9%. The other 5 lost between 5.4% (MNQ, 15-minute chart) and 9.7% (MNQ, 5-minute chart). All 5 lost money even before trading costs. All 6 tests lost money in 2018–2021. In 2022–2025, 3 of them made money, at most +2.7% (MES, 15-minute chart). Each period's % is the change in the account during those years. With only 75 to 89 trades per test, a result of a few percent either way can come from luck.

  2. We measure results in R. 1R is 1% of the account at the time of the trade, the risk each trade aims for. Only 43 of the 498 trades (8.6%) reached the 2R take-profit, and they made +1.84R each on average. 175 trades hit the stop-loss and lost 0.94R each on average. The other 280 were still open at 12:00 and closed then: 181 of them made money, and the 280 made +0.18R each on average. Breakouts went up more often than down: the 301 buys were close to break-even (+4.0R together), and the 197 sells lost 40.3R.

  3. CPI did not always move the market much. From 2018 to 2020, the typical (median) range from 08:30 to 08:45 on CPI days was 20 points on MNQ. That was about twice the range on other weekdays, and about half the range on the days of the US jobs report, the other big report at 08:30. From 2021 to 2025, as US inflation rose, the typical CPI range was 144 points. That was 5 times the range on other weekdays, and wider than on jobs report days. In 2022, it was 273 points. MES and MGC changed the same way. The breakout's losses came mostly from 2019 and 2020. From 2018 to 2020, the 6 tests together lost 38.3R in 207 trades. From 2021 to 2025, they were close to break-even, about +2R in 291 trades. Range size alone does not explain the results. 2018 had the smallest range and made +2.3R, and 2023 lost 7.0R with a wide one (see the table 'CPI days, year by year' below).

  4. Trades opened before 09:30, when the US stock market opens, lost money in all 6 tests. With the basic rules, 254 of the 498 trades opened before 09:30. Together, they lost 46.7R. The other 244 trades opened at 09:30 or later, and together they made +10.3R. On MNQ and MES, 83 of the 159 early trades hit the stop-loss, compared with 48 of the 163 later trades. On MNQ and MES, the early trades did worst from 2018 to 2020, when the CPI range was small and the stop-loss was close. Then, 56 of the 74 early trades hit the stop-loss. We did not plan this split: we found it after looking at the results. We also tested waiting for 09:30 as a rule, one of four planned changes. It lost less than the basic rules, but it still lost money in all 6 tests (see 'Can the CPI day breakout be improved?').

  5. On MNQ, the biggest CPI days were too big for 1% risk. On 13 of the 94 CPI days on the 5-minute chart, even one contract would have risked more than 1% of the account, so the test skipped the day. On the 15-minute chart, it was 12 days. 7 of the 13 were in 2022. On 13 September 2022, for example, one MNQ contract would have risked $934, about twice the 1% at the time. To see what the skipped days did, we ran the same rules on a $1,000,000 account, where no day was skipped. There, the 13 days made +2.4R together. That account lost 10.2% on the MNQ 5-minute chart, a little more than the 9.7% on the $50,000 account. Without rounding down, every trade there risked the full 1%. So the skipped days did not hide a profit.

CPI days, year by year

Big losses in 2019 and 2020, then about break-even as CPI ranges grew

Each row is one year. The three 'MNQ range' columns show the typical (median) range on MNQ from 08:30 to 08:45 New York time. They cover CPI days, jobs report days and all other weekdays. On half of the days, the range was smaller. The last two columns add up the trades of all 6 tests with the basic rules, after trading costs.

Year MNQ range, CPI days MNQ range, jobs report days MNQ range, other weekdays Trades, 6 tests Result, 6 tests
2018 16 points 35 points 8 points 72 +2.3R
2019 17 points 32 points 9 points 69 −10.5R
2020 27 points 47 points 19 points 66 −30.1R
2021 87 points 80 points 19 points 69 −0.7R
2022 273 points 141 points 34 points 46 −6.2R
2023 128 points 106 points 26 points 66 −7.0R
2024 145 points 132 points 31 points 60 +11.4R
2025 149 points 119 points 33 points 50 +4.4R

MES and MGC (micro gold) changed the same way. The typical CPI range on MES was 5.75 points in 2018–2020 and 30.25 points in 2021–2025. On MGC, it was $4.00 and $17.70 of the gold price. On all 3 markets, the CPI range was about half the jobs report range in 2018–2020, and about 1.2 times the jobs report range in 2021–2025. 2022 had the fewest trades. On MNQ, one contract would have risked more than 1% of the account on 7 of its 12 CPI days on the 5-minute chart. On the 15-minute chart, it was 5 days. Those days were skipped. Each R figure is a trade's result after trading costs, divided by 1% of the account at the time of the trade.

Before or after the 09:30 stock market open

Trades opened before 09:30 lost money in all 6 tests

We did not plan this split: we found it after looking at the results. CPI comes out at 08:30 New York time, an hour before the US stock market opens at 09:30. The cards split the trades of the basic rules from 2018 to 2025, after trading costs, by the time each trade opened. The MNQ and MES cards add up the 4 stock index tests, and the MGC cards the 2 gold tests.

MNQ and MES: trades that opened before 09:30
−0.258R
Average per trade, 159 trades. 83 of them hit the stop-loss.
MNQ and MES: trades that opened at 09:30 or later
+0.061R
Average per trade, 163 trades. 48 of them hit the stop-loss.
MGC: trades that opened before 09:30
−0.060R
Average per trade, 95 trades. 29 of them hit the stop-loss.
MGC: trades that opened at 09:30 or later
+0.005R
Average per trade, 81 trades. 15 of them hit the stop-loss.

In 2018–2020, when the CPI range was small, the early MNQ and MES trades lost 0.424R each (74 trades), and 56 of them hit the stop-loss. In 2021–2025, they lost 0.114R each (85 trades). The later MNQ and MES trades lost 0.049R each in 2018–2020 (64 trades), and made 0.132R each in 2021–2025 (99 trades). Test by test, the early trades lost money in all 6 tests, though only 0.13R in total on MGC's 5-minute chart. The later trades made money in 5 of the 6, all but MGC on the 15-minute chart. The later trades came on days when no candle closed outside the range before 09:30. Because we found this split in the results, it needs its own test on new data before anyone trades it. Each R figure is a trade's result after trading costs, divided by 1% of the account at the time of the trade.

Every version, test by test

No version made money in more than 1 of the 6 tests

Each row is one market and chart, from 2018 to 2025, after trading costs, on an account that started at $50,000. Each column is one version of the rules: the basic rules and the four changes under 'Can the CPI day breakout be improved?'. Each cell shows the change in the account, with the number of trades in brackets.

Market Chart Basic rules Wait for 09:30 Stop-loss at the middle of the range Hold until 16:00 Only against the first 15 minutes
MNQ 5-minute −9.7% (77) −5.3% (77) −14.9% (88) −10.4% (76) −2.7% (44)
MNQ 15-minute −5.4% (75) −1.6% (74) −14.8% (85) −5.2% (75) −0.5% (41)
MES 5-minute −8.8% (86) −2.5% (84) −4.2% (88) −7.2% (86) −6.1% (43)
MES 15-minute −7.3% (84) −3.7% (82) +2.5% (86) −6.2% (84) −3.8% (41)
MGC (micro gold) 5-minute +0.9% (89) −0.8% (86) −6.2% (89) 0.0% (89) −9.3% (46)
MGC (micro gold) 15-minute −6.4% (87) −0.6% (83) −11.9% (88) −7.3% (87) −7.4% (40)

With the stop-loss at the middle of the range, a trade risks about half as far. So 1 contract fits the 1% risk on more days: MNQ traded on 88 and 85 days, compared with 77 and 75 with the basic rules. But that version lost the most: 50.1R across the 6 tests, compared with 36.4R with the basic rules. 82 of its 524 trades reached the take-profit, and 258 hit the stop-loss. Holding until 16:00 changed each result by 1.6% or less. Trading only against the first 15 minutes traded on 40 to 46 days, about half as many as the basic rules, and lost money in all 6 tests. We tested it because our NFP day breakout backtest (the same rules on US jobs report days) found that on MNQ and MES, breakouts against the first move did better. They did on CPI days too: with the basic rules, they lost 0.018R per trade (100 trades), compared with 0.132R for breakouts with the first move (222 trades). But the rule also takes a later breakout against the first move, on days when the first breakout went with it. Those 69 later trades lost 10.7R, so on MNQ and MES the rule lost 0.075R per trade (169 trades). Each R figure is a trade's result after trading costs, divided by 1% of the account at the time of the trade.

The strategy

What the CPI day breakout is

CPI stands for Consumer Price Index. It is the US government's monthly report on how fast the prices of everyday goods and services rise: the main US inflation number. It comes out at 08:30 New York time, on a weekday near the middle of the month. The Federal Reserve (the US central bank) watches inflation when it sets interest rates. So when CPI comes out higher or lower than traders expected, futures on stock indexes and gold can move a lot in the first minutes. Futures are contracts that follow the price of a market, such as the Nasdaq 100. A common way to trade CPI day is to let the first 15 minutes pass. Mark the highest and the lowest price from 08:30 to 08:45, and trade the first breakout of that range. Traders use it on NQ (Nasdaq 100 futures), ES (S&P 500 futures) and gold futures. We tested their micro contracts, MNQ, MES and MGC (see 'Contracts'). A breakout is a candle that closes outside the range: a 5-minute candle on the 5-minute chart, and a 15-minute candle on the 15-minute chart. The trade opens at the next candle's open. The 09:30 US stock market open falls inside the time a breakout can happen.

Release days
We tested 94 of the 95 CPI days from 2018 to 2025, all at 08:30 New York time, from the BacktestingMax economic calendar. One report came out on Good Friday, 10 April 2020, when the US stock market was closed. Our data has no prices on that day. The US government shutdown of late 2025 moved the report on September 2025 prices to Friday 24 October 2025. There was no report on October prices. The report on November prices came out on Thursday 18 December 2025, so there was no CPI day in November 2025
Range
08:30 to 08:45 New York time, the first 15 minutes after the report. In UK time, that is 13:30 to 13:45 for most of the year
Breakout
A candle that closes above the range high (buy) or below the range low (sell), after 08:45 and before 12:00 New York time. The US stock market opens at 09:30, inside this window
Stop-loss
The range low for a buy, the range high for a sell
Take-profit
Twice as far from the breakout candle's close as the stop-loss (2R). A trade still open at 12:00 New York time closes then
Trades per day
1 at most: the first breakout. On 4 to 8 of the 94 days, depending on the market and the chart, no candle closed outside the range before 12:00, so there was no trade
Account and risk
$50,000 at the start. That is a common size for a prop firm evaluation. A prop firm is a company that lets traders trade its money, and the evaluation is the trading test you pass first. Each trade risks 1% of the balance at that time, rounded down to whole contracts. A typical (median) trade used 2 MNQ, 4 MES or 3 MGC contracts. If even 1 contract would risk more than 1%, the day is skipped. This happened on 13 days (5-minute chart) and 12 days (15-minute chart) on MNQ. It happened on 2 days on MES (13 September and 13 October 2022), and on 1 day on MGC's 15-minute chart (24 October 2025)
Contracts
MNQ, MES and MGC are the micro contracts of NQ, ES and gold futures (GC): the same prices at one tenth of the size. One MNQ contract makes or loses $2 per point, one MES contract $5 per point, and one MGC contract $10 when gold moves $1. We tested the micros because the full-size contracts are too big for 1% risk on CPI day. On the median CPI day from 2018 to 2025, the range was 87.5 points on MNQ, which follows the same price as NQ. For one full-size NQ contract, that alone was worth $1,750, 3.5% of a $50,000 account. We did not test YM (Dow futures) or RTY (Russell 2000 futures), because our data has only their full-size contracts. On 56 of the 94 CPI days, the range of one YM contract alone was worth more than $500, 1% of the account. RTY was the same on 52 of its 70 CPI days (our RTY data starts in 2020)
Fills
The trade opens at the open of the candle after the breakout candle. The stop-loss fills at its price, or at a candle's open if price opens past it. The take-profit fills at its exact price. A trade still open at 12:00 closes at the open of the 12:00 candle. When price reaches the stop-loss and the take-profit in the same candle, the backtest looks at the 1-minute candles inside it to see which came first. If both are in the same 1-minute candle, the backtest counts the stop-loss
Price data
BacktestingMax's futures data. It is built from CFD prices (a broker's price that follows the same market). It is adjusted to match the real contracts: their price, trading hours, tick (the smallest price step) and roll (the switch from one contract month to the next). It is not exchange data. Before mid-2018, our stock index data has hours with stale prices (prices that did not update) outside the stock market's hours. So we looked closely at January to June 2018. From 08:30 to 12:00, prices updated as often as in the rest of 2018 and in 2019. Without the 6 CPI days of January to June 2018, the 4 MNQ and MES tests lost more, not less. Those 6 days made +3.3R to +4.7R in each of these 4 tests
Trading costs
The spread is the gap between the buy and the sell price. We used a spread of 1 tick: $0.50 per contract on MNQ, $1.25 on MES and $1.00 on MGC. A trade pays half of it when it opens and half when it closes. A trade that closes at the take-profit pays no spread when it closes. There is also a commission (the broker's fee) of $0.62 per contract when a trade opens, and again when it closes. On average, a trade paid about $10 in costs on MNQ, $18 to $19 on MES and $11 to $12 on MGC. Each test paid $720 to $1,650 in costs from 2018 to 2025, the most on MES, which needed the most contracts. Real costs around the CPI report can be higher: spreads get wider in the first minutes after the report, and a stop-loss can fill past its price. This backtest adds no extra cost for that

The rules we tested

  1. Trade only on the days the US CPI report comes out, at 08:30 New York time. We took these days from the BacktestingMax economic calendar (see 'Release days').
  2. Mark the range: the highest and the lowest price from 08:30 to 08:45 New York time. On the 5-minute chart, that is the 3 candles from 08:30 to 08:45. On the 15-minute chart, it is the one candle from 08:30 to 08:45.
  3. After 08:45, buy when a candle closes above the range high. Sell (go short) when a candle closes below the range low. A flat candle (high equal to low, which usually means the price did not update) never counts. The breakout candle must close before 12:00 New York time. The trade opens at the next candle's open.
  4. Put the stop-loss at the other side of the range: at the range low for a buy, and at the range high for a sell.
  5. Put the take-profit twice as far from the breakout candle's close as the stop-loss, in the direction of the trade. Traders call this a 2R take-profit.
  6. If the trade is still open at 12:00 New York time, close it then.
  7. Take only the first breakout of the day, win or lose. After that trade, wait for the next CPI day.
  8. Size every trade so that the distance from the breakout candle's close to the stop-loss is worth 1% of the account balance at that time. Futures trade in whole contracts, so round down to a whole number of contracts. If even one contract would risk more than 1% of the account, skip the day. Because of the rounding, and because the trade opens at the next candle's open, a trade risked 0.86% of the account on average, between 0.5% and 1.1%.

The method

How we tested it

The test

Markets
MNQ (Micro Nasdaq 100), MES (Micro S&P 500) and MGC (Micro Gold)
Charts
5-minute and 15-minute candles
Dates
January 2018 to December 2025
Account
$50,000, risking 1% of the account on each trade
Trading costs
Included. Every trade pays that market's normal trading cost: the spread, which is the gap between the buy price and the sell price.

How a test passes

We call a test passed only if it meets all four checks. A strategy can make money in one backtest and still fail one of these checks.

  • At least 100 trades. With fewer trades, a good result could be luck.
  • It made money from 2018 to 2021 and again from 2022 to 2025. This shows whether the strategy kept working in the newer years, not only in the older ones.
  • A profit factor (the money won divided by the money lost) of at least 1.2 from 2022 to 2025, after trading costs.
  • In 95% of 1,000 random shuffles of the trades, the account must not fall more than 30% from its highest point.

Improvements

Can the CPI day breakout be improved?

We tested four changes that CPI day traders often make, one at a time, on the same 6 tests. The first waits for the US stock market to open at 09:30 before a breakout counts. The second puts the stop-loss at the middle of the range, so the stop-loss and the take-profit are closer. The third holds trades until 16:00 instead of 12:00. The fourth trades only against the move of the first 15 minutes, for traders who say the first move after the report often reverses. The table below explains each one. We kept the 2R take-profit and the 12:00 exit as traders describe the strategy, and tuned no numbers. If you pick the best change and judge it on the same years you used to pick it, it almost always looks better than it really is. So we picked the best change using 2018–2021 only: the change that made money in the most tests. In a tie, we picked the one with the better middle test (the median: half of the tests did better, half did worse). Then we tested it on 2022–2025, the newer years we kept aside.

Version Made money 2018–2021years used to pick the best version Made money 2022–2025years not used to pick it Made money in both periods Passed all four checks
Basic rules The rules under 'The rules we tested': the range from 08:30 to 08:45, the stop-loss at the other side, a 2R take-profit and a close at 12:00. 0 of 6middle test −6.8% 3 of 6middle test +0.1% 0 0
Wait for the 09:30 stock market openBest on 2018–2021 The range stays 08:30 to 08:45, but a breakout counts only on a candle that opens at 09:30 or later, when the US stock market is open. So the earliest trade opens at 09:35 on the 5-minute chart, and at 09:45 on the 15-minute chart. 1 of 6middle test −2.4% 2 of 6middle test −0.6% 0 0
Stop-loss at the middle of the range The stop-loss goes at the middle of the range, halfway between the range high and the range low, instead of at the other side. So a trade risks about half as far, and the 2R take-profit is closer. With the same 1% risk, a trade uses about twice as many contracts. 0 of 6middle test −5.5% 2 of 6middle test −3.2% 0 0
Hold until 16:00 A trade still open at 12:00 stays open until 16:00 New York time, when the US stock market closes. The take-profit stays at 2R, and a breakout still has to come before 12:00. 1 of 6middle test −6.2% 2 of 6middle test −0.7% 0 0
Only against the first 15 minutes If price rose from 08:30 to 08:45 (the close at 08:45 above the open at 08:30), only sell, on a break below the range. If price fell, only buy, on a break above the range. If it ended where it started, don't trade that day. A breakout the other way doesn't count: wait for one in the allowed direction until 12:00. 1 of 6middle test −3.8% 0 of 6middle test −1.5% 0 0

No change made the CPI day breakout work. From 2018 to 2025, no version made money in more than 1 of the 6 tests, and no test made money in both 2018–2021 and 2022–2025. On 2018–2021, the years we used to pick, three changes made money in 1 of 6 tests each. Waiting for the 09:30 open had the best middle test: −2.4%, compared with −6.8% with the basic rules. On 2022–2025, the years we kept aside, it made money in 2 of 6 tests, compared with 3 of 6 with the basic rules. Its middle test was −0.6%, compared with +0.1%. From 2018 to 2025, it lost less than the basic rules in 5 of the 6 tests, but it still lost money in all 6, between 0.6% and 5.3%. Look at the 254 trades the basic rules opened before 09:30. On those days, waiting for the open still lost 20.0R, compared with 46.7R. The table 'Every version, test by test' above shows how the other three changes did.

Questions

Questions traders ask

Does the CPI breakout strategy work?

Not in our backtest. We tested the breakout of the 08:30 to 08:45 New York range on 94 CPI days from 2018 to 2025. The markets were MNQ, MES and MGC, the micro contracts of NQ, ES and gold futures, each on 5-minute and 15-minute charts. The stop-loss was at the other side of the range. The take-profit was twice as far from the breakout candle's close as the stop-loss (2R), and every trade closed by 12:00. After trading costs, only MGC on the 5-minute chart made money: +0.9% in 8 years, from 89 trades. A result that small could be luck. We also tested four changes: waiting for the 09:30 stock market open, a stop-loss at the middle of the range, holding until 16:00, and trading only against the first 15 minutes. None of them made money in more than 1 of the 6 tests.

How much does NQ move on CPI day?

Much more than it used to. We measured the range of the first 15 minutes after the report, from 08:30 to 08:45 New York time. We used MNQ, which follows the same price as NQ (Nasdaq 100 futures). From 2018 to 2020, the typical range was 20 points. From 2021 to 2025, it was 144 points. In 2022, the year US inflation peaked, it was 273 points, 8 times the range on other weekdays. On 13 September 2022, it was 452.5 points. With ranges that wide, a stop-loss at the other side of the range is far away. On 13 of the 94 CPI days (5-minute chart), even one MNQ contract would have risked more than 1% of the account, which started at $50,000. The CPI day breakout lost money on MNQ: 9.7% on the 5-minute chart and 5.4% on the 15-minute chart, from 2018 to 2025.

Is the CPI day breakout a good gold strategy?

Not in our backtest. On MGC (micro gold futures), it made only +0.9% from 2018 to 2025 on the 5-minute chart, from 89 trades. It lost 6.4% on the 15-minute chart, after trading costs. Gold moves on CPI days too: its typical range from 08:30 to 08:45 grew from $4 in 2018–2020 to about $18 in 2021–2025. On gold, breakouts in the same direction as the first 15 minutes did better than breakouts against it. For example, if price rose from 08:30 to 08:45, a break above the range went the same way. Across the 2 MGC tests, 60.6% of the trades in the same direction made money (109 trades). Only 43.3% of the trades against it did (67 trades). On MNQ and MES, breakouts against the first move did a little better: 42.0% made money, compared with 39.2%. We found this after looking at the results, so it needs its own test on new data before anyone trades it.

Should you trade CPI right away or wait for the stock market open?

In our backtest, waiting helped, but not enough to make money. CPI comes out at 08:30 New York time (13:30 in the UK for most of the year), an hour before the US stock market opens at 09:30. With the basic rules, the trades opened before 09:30 lost money in all 6 tests. The trades opened at 09:30 or later made money in 5 of the 6. Those came on days when no candle closed outside the range before 09:30. We found this after looking at the results, so it needs its own test. We did test one rule: wait until 09:30 for the first breakout. It lost less than the basic rules in 5 of the 6 tests, but it still lost money in all 6: between 0.6% and 5.3% from 2018 to 2025.

Keep exploring

Related strategies we tested

Seasonal & calendar NFP day breakout backtest on NQ, ES and gold futures Made money in 0 of 6 tests, 0 of them in both periods. Best: MES (Micro S&P 500), 5-minute, +3% Seasonal & calendar FOMC day fade backtest on NQ, ES and gold futures Made money in 0 of 6 tests, 0 of them in both periods. Best: MGC (Micro Gold), 5-minute, +12%
The strategy code we ran

This is the exact code from the BacktestingMax strategy engine. Paste it into the strategy editor in the app to run it on any market.

class CpiDayBreakout(Strategy):
    """CPI day breakout on US futures: on the days the US inflation report comes out, trade
    the first breakout of the range of the 15 minutes after the release.
    All times are New York time (daylight saving included). The CPI report (the Consumer
    Price Index, the main US inflation number) comes out once a month at 08:30, on a
    weekday near the middle of the month. The US government shutdown of late 2025 moved
    the September 2025 report to Friday 24 October, there was no October 2025 report, and
    the November 2025 report came out on Thursday 18 December. The list below has the real
    release days, all at 08:30 New York.
    Release days: 'CPI m/m', 'CPI y/y' and 'Core CPI m/m' in the BacktestingMax economic
    calendar. One of them, Good Friday 10 April 2020, has no prices in the data (the stock
    market was closed), so there is no trade on it.
    Standard (all switches off):
      1. Trade only on release days.
      2. The range: the highest and the lowest price from 08:30 to 08:45 (3 candles on
         the 5-minute chart, 1 candle on the 15-minute chart). It is the same range on
         both charts.
      3. After 08:45, buy when a candle closes above the range high, and sell (go short)
         when a candle closes below the range low. A flat candle (high equal to low)
         never counts. The trade opens at the next candle's open. A breakout counts on
         any candle that closes before 12:00.
      4. Stop-loss at the other side of the range: the range low for a buy, the range
         high for a sell.
      5. Take-profit 2R away, R being the distance from the breakout candle's close to
         the stop-loss.
      6. A trade still open at 12:00 closes then.
      7. One trade a day: the first breakout, win or lose.
      8. Size every trade so that the stop-loss costs 1% of the account, in whole
         contracts. If one contract would risk more, skip the day.
    The changes, one at a time:
      mid_stop      = 1   stop-loss at the middle of the range instead of the other side,
                          so the trade risks about half as far and the 2R take-profit is
                          closer
      after_open    = 1   wait for the US stock market to open: a breakout counts only on a
                          candle that opens at 09:30 or later (the range stays 08:30 to 08:45)
      late_exit     = 1   a trade still open at 12:00 stays open until 16:00, when the US
                          stock market closes (breakouts still count only before 12:00)
      against_first = 1   only trade against the move of the first 15 minutes: if the range
                          closed higher than it opened (the last price at 08:45 above the
                          first price at 08:30), only sell on a break below the range; if it
                          closed lower, only buy on a break above it. A breakout the other
                          way doesn't count: wait for one in the allowed direction until 12:00
    """
    params = {"exit_rr": Param(type=float, default=2.0, min=0.5, max=10.0),
              "mid_stop": Param(type=int, default=0, min=0, max=1),
              "after_open": Param(type=int, default=0, min=0, max=1),
              "late_exit": Param(type=int, default=0, min=0, max=1),
              "against_first": Param(type=int, default=0, min=0, max=1)}
    lookback = 2
    release = 510         # 08:30 New York, in minutes after midnight
    range_end = 525       # 08:45: breakouts count from here
    stock_open = 570      # 09:30, the US stock market open
    noon = 720            # 12:00: a breakout candle must close before this; the standard exit
    stock_close = 960     # 16:00, the US stock market close: the exit with late_exit = 1
    # CPI report days (08:30 New York), from the BacktestingMax economic calendar.
    release_days = (
        "2018-01-12 2018-02-14 2018-03-13 2018-04-11 2018-05-10 2018-06-12 2018-07-12 2018-08-10 "
        "2018-09-13 2018-10-11 2018-11-14 2018-12-12 "
        "2019-01-11 2019-02-13 2019-03-12 2019-04-10 2019-05-10 2019-06-12 2019-07-11 2019-08-13 "
        "2019-09-12 2019-10-10 2019-11-13 2019-12-11 "
        "2020-01-14 2020-02-13 2020-03-11 2020-04-10 2020-05-12 2020-06-10 2020-07-14 2020-08-12 "
        "2020-09-11 2020-10-13 2020-11-12 2020-12-10 "
        "2021-01-13 2021-02-10 2021-03-10 2021-04-13 2021-05-12 2021-06-10 2021-07-13 2021-08-11 "
        "2021-09-14 2021-10-13 2021-11-10 2021-12-10 "
        "2022-01-12 2022-02-10 2022-03-10 2022-04-12 2022-05-11 2022-06-10 2022-07-13 2022-08-10 "
        "2022-09-13 2022-10-13 2022-11-10 2022-12-13 "
        "2023-01-12 2023-02-14 2023-03-14 2023-04-12 2023-05-10 2023-06-13 2023-07-12 2023-08-10 "
        "2023-09-13 2023-10-12 2023-11-14 2023-12-12 "
        "2024-01-11 2024-02-13 2024-03-12 2024-04-10 2024-05-15 2024-06-12 2024-07-11 2024-08-14 "
        "2024-09-11 2024-10-10 2024-11-13 2024-12-11 "
        "2025-01-15 2025-02-12 2025-03-12 2025-04-10 2025-05-13 2025-06-11 2025-07-15 2025-08-12 "
        "2025-09-11 2025-10-24 2025-12-18 "
        "2026-01-13 2026-02-13 2026-03-11 2026-04-10 2026-05-12 2026-06-10 2026-07-14 2026-08-12 "
        "2026-09-11 2026-10-14")
    def init(self, ctx):
        self.ny = ctx.clock("America/New_York")
        self.releases = set(self.release_days.split())
        self.day = None
        self.high = None          # the range high
        self.low = None           # the range low
        self.count = 0            # candles in the range so far
        self.has_start = False    # True once the candle that opens the range has been seen
        self.first_open = None    # the first price of the range
        self.last_close = None    # the last price of the range
        self.done = True          # True once today's trade is placed, or the day can't trade
        self.trade_day = None     # the date the open trade was placed on
    def on_bar(self, ctx):
        start = self.ny.minute_of_day              # this candle's open, New York time
        end = start + ctx.timeframe_minutes        # this candle's close
        exit_time = self.stock_close if ctx.params.late_exit == 1 else self.noon
        if self.ny.date != self.day:               # a new day: forget yesterday's range
            self.day = self.ny.date
            self.high = self.low = None
            self.count = 0
            self.has_start = False
            self.first_open = self.last_close = None
            self.done = self.day not in self.releases
        if ctx.position is not None:
            if end >= exit_time or self.ny.date != self.trade_day:
                ctx.close_position()               # fills at the open of the 12:00 (16:00) candle
            return
        if self.done:
            return
        # The range: the candles that open from 08:30 to 08:45.
        if self.release <= start < self.range_end:
            if start == self.release:
                self.has_start = True
                self.first_open = ctx.open[0]
            self.high = ctx.high[0] if self.high is None else max(self.high, ctx.high[0])
            self.low = ctx.low[0] if self.low is None else min(self.low, ctx.low[0])
            self.last_close = ctx.close[0]
            self.count += 1
            return
        if start < self.range_end:
            return
        if end >= self.noon:                       # no breakout before 12:00: no trade today
            self.done = True
            return
        full = self.has_start and self.count * ctx.timeframe_minutes == self.range_end - self.release
        if not ctx.gate("full_range", full and self.high is not None and self.high > self.low):
            self.done = True
            return
        direction = self.last_close - self.first_open   # how the first 15 minutes moved
        if ctx.params.against_first == 1 and not ctx.gate("first_15_minutes_moved", direction != 0):
            self.done = True
            return
        if ctx.params.after_open == 1 and not ctx.gate("stock_market_open", start >= self.stock_open):
            return                                 # before 09:30 a breakout doesn't count yet
        close = ctx.close[0]
        moved = ctx.high[0] > ctx.low[0]           # a flat candle never counts
        up = moved and close > self.high
        down = moved and close < self.low
        if not ctx.gate("range_break", up or down):
            return
        if ctx.params.against_first == 1 and not ctx.gate("against_first_15_minutes",
                                                           (up and direction < 0) or (down and direction > 0)):
            return                                 # a breakout with the first move: keep waiting
        self.done = True                           # one trade a day: this breakout is it
        if ctx.params.mid_stop == 1:
            stop = (self.high + self.low) / 2      # the middle of the range
        else:
            stop = self.low if up else self.high   # the other side of the range
        risk = abs(close - stop)
        size = ctx.size_by_risk_pct(0.01, stop=stop, entry=close)
        if not ctx.gate("one_contract_fits_1pct_risk", size > 0):
            return                                 # one contract would risk more than 1%
        target = close + ctx.params.exit_rr * risk if up else close - ctx.params.exit_rr * risk
        self.trade_day = self.day
        if up:
            ctx.buy(size=size, sl=stop, tp=target, tag="cpi_long")
        else:
            ctx.sell(size=size, sl=stop, tp=target, tag="cpi_short")

Your turn

Test the CPI day breakout yourself

Change the settings, the market or the stop-loss and see the result in seconds. Ask the BacktestingMax AI: Backtest the CPI day breakout on MNQ, 5-minute chart: on US CPI days only, mark the 08:30 to 08:45 New York range, buy when a candle closes above the range, sell when a candle closes below it, take only the first breakout of the day, stop-loss at the other side of the range, take-profit 2R, close at 12:00, risk 1% per trade.

Try it free in BacktestingMax

Free account. No credit card needed.

See every strategy we tested Browse strategies shared by traders

Backtest on historical data with trading costs, run by BacktestingMax. Past results do not guarantee future results, and nothing here is financial advice. Updated 9 November 2026.