Tested by BacktestingMax
Wyckoff spring backtest
Does the Wyckoff spring work?
No. On 15-minute, 1-hour and 4-hour charts, the Wyckoff spring lost money after trading costs. We backtested it on 10 markets from 2019 to 2025: 3 currency pairs, gold, silver and 5 stock indices. A spring is a candle that dips below a trading range, a stretch where price moved sideways, and then closes back inside the range. Our rules bought at the open of the next candle. The stop-loss was at the spring candle's low, and the take-profit at the range's top. R is the amount a trade risks: 1% of the account. On these three charts, the 8,218 trades lost 0.190R on average: 0.19% of the account per trade. Before trading costs, they broke even. But the stop-loss was close to the entry, so trading costs took 0.195R per trade. Of the 30 backtests, 15 had at least 100 trades, and only 2 of those made money. On daily charts, which we tested separately, the spring made money on 9 of 10 markets. But there were only 139 trades in total over 12 to 19 years, and 5 of them made most of the profit. We also tested daily charts on 5 more currency pairs, and only 2 of the 5 made money.
30 backtests 2019–2025 Trading costs included Tested 9 November 2026
Where it worked 30 backtests- Made money
- 2 of 30 tests, with the standard rules
- Made money in both periods
- 0 2019–2022 and again 2023–2025
- Best backtest
- +131% Dow Jones, 15-minute chart: account growth 2019–2025
- With a retest of the spring
- 1 tests made money in both periods, compared with 0 with the standard rules
The results
Results on every market
Each box shows one backtest of the Wyckoff spring with the standard rules, from 2019 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 2019–2022 and 2023–2025, or passed all four of our checks (see "How a test passes" further down).
| Market | 15-minute | 1-hour | 4-hour |
|---|---|---|---|
| EUR/USD | −67% Lost money | −7% Lost money | +0.1% Too few trades |
| GBP/USD | −93% Lost money. Account down 90% by Jun 2024 | −25% Lost money | −17% Too few trades |
| USD/JPY | −74% Lost money | −40% Lost money | −7% Too few trades |
| Gold | −90% Lost money | −33% Too few trades | +0.2% Too few trades |
| Silver | −100% Lost money. Account down 90% by Sept 2020 | −69% Lost money | −14% Too few trades |
| S&P 500 | −30% Lost money | −28% Too few trades | +3% Too few trades |
| Nasdaq 100 | −42% Lost money | −11% Too few trades | −13% Too few trades |
| Dow Jones | +131% Made money overall, lost in one period | −10% Too few trades | +2% Too few trades |
| DAX | −52% Lost money | +3% Too few trades | −2% Too few trades |
| FTSE 100 | −44% Lost money | +5% Made money overall, lost in one period | +3% Too few trades |
All the numbers, test by test
| Market | Chart | Trades | Win rate | Profit factor | Result | Max drawdown | 2019–2022 | 2023–2025 | Monte Carlo drawdown | Trading costs |
|---|---|---|---|---|---|---|---|---|---|---|
| EUR/USD | 15-minute | 842 | 19.1% | 0.80 | −66.7% | −70.1% | −52.6% | −29.8% | −78.7% | $7,321 |
| EUR/USD | 1-hour | 123 | 17.9% | 0.92 | −7.3% | −31.2% | −9.9% | +2.9% | −35.0% | $1,362 |
| EUR/USD | 4-hour | 31 | 16.1% | 1.00 | +0.1% | −14.3% | −11.7% | +13.4% | −19.2% | $158 |
| GBP/USD | 15-minute | 857 | 15.2% | 0.66 | −93.2% | −94.1% | −76.3% | −71.3% | −94.8% | $6,368 |
| GBP/USD | 1-hour | 144 | 20.8% | 0.76 | −24.5% | −36.0% | −30.6% | +8.8% | −47.9% | $1,381 |
| GBP/USD | 4-hour | 30 | 16.7% | 0.30 | −16.5% | −19.1% | −8.8% | −8.4% | −20.4% | $159 |
| USD/JPY | 15-minute | 651 | 19.8% | 0.73 | −74.4% | −74.9% | −55.5% | −42.5% | −80.7% | $5,860 |
| USD/JPY | 1-hour | 111 | 15.3% | 0.51 | −40.1% | −40.1% | −24.9% | −20.2% | −48.4% | $1,011 |
| USD/JPY | 4-hour | 24 | 16.7% | 0.66 | −6.9% | −8.2% | −3.2% | −3.8% | −15.3% | $104 |
| Gold | 15-minute | 752 | 19.3% | 0.64 | −89.8% | −89.8% | −84.6% | −33.4% | −92.5% | $6,642 |
| Gold | 1-hour | 99 | 19.2% | 0.55 | −33.4% | −34.9% | −15.1% | −21.5% | −41.9% | $1,036 |
| Gold | 4-hour | 23 | 21.7% | 1.01 | +0.2% | −11.2% | −6.9% | +7.7% | −13.7% | $172 |
| Silver | 15-minute | 476 | 14.7% | 0.22 | −99.5% | −99.5% | −98.4% | −70.9% | −99.6% | $8,162 |
| Silver | 1-hour | 157 | 15.3% | 0.34 | −68.8% | −70.7% | −49.2% | −38.5% | −71.1% | $3,205 |
| Silver | 4-hour | 47 | 21.3% | 0.63 | −14.0% | −22.6% | −15.5% | +1.8% | −24.1% | $728 |
| S&P 500 | 15-minute | 681 | 22.2% | 0.93 | −29.9% | −51.7% | −14.6% | −17.9% | −68.0% | $7,840 |
| S&P 500 | 1-hour | 92 | 17.4% | 0.60 | −28.3% | −28.9% | −26.3% | −2.7% | −38.6% | $729 |
| S&P 500 | 4-hour | 25 | 28% | 1.13 | +2.5% | −9.6% | +1.1% | +1.4% | −12.1% | $63 |
| Nasdaq 100 | 15-minute | 655 | 21.2% | 0.93 | −42.1% | −68.5% | +19.0% | −51.3% | −73.2% | $9,472 |
| Nasdaq 100 | 1-hour | 70 | 24.3% | 0.81 | −11.1% | −22.3% | +6.4% | −16.5% | −26.8% | $558 |
| Nasdaq 100 | 4-hour | 18 | 5.6% | 0.11 | −13.2% | −13.2% | −9.2% | −4.4% | — | $43 |
| Dow Jones | 15-minute | 741 | 23.5% | 1.20 | +131.3% | −40.9% | −9.1% | +154.6% | −49.3% | $8,147 |
| Dow Jones | 1-hour | 86 | 16.3% | 0.86 | −9.8% | −22.0% | −9.3% | −0.6% | −34.3% | $510 |
| Dow Jones | 4-hour | 20 | 30% | 1.14 | +2.0% | −6.8% | +2.1% | −0.2% | −9.8% | $33 |
| DAX | 15-minute | 554 | 21.3% | 0.82 | −51.9% | −67.7% | −56.8% | +11.5% | −70.6% | $3,127 |
| DAX | 1-hour | 95 | 22.1% | 1.03 | +2.6% | −27.9% | +12.1% | −8.5% | −27.7% | $525 |
| DAX | 4-hour | 29 | 27.6% | 0.92 | −1.6% | −7.4% | −3.2% | +1.7% | −12.1% | $67 |
| FTSE 100 | 15-minute | 583 | 22.1% | 0.91 | −44.0% | −69.0% | +1.2% | −44.7% | −69.6% | $10,594 |
| FTSE 100 | 1-hour | 152 | 24.3% | 1.04 | +5.2% | −19.5% | −4.1% | +9.7% | −34.2% | $1,605 |
| FTSE 100 | 4-hour | 50 | 20% | 1.07 | +3.1% | −17.4% | −1.5% | +4.7% | −22.4% | $343 |
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. 2019–2022 and 2023–2025 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
What we found
What the 30 backtests show
After a spring, price reached the range's top before it fell below the spring's low only 1 time in 5. Wyckoff traders expect a real spring's low to hold. Our rules found 10,880 springs on 15-minute, 1-hour and 4-hour charts from 2019 to 2025. After 2,175 out of 10,880 (20.0%), price reached the range's top first. After 8,662 out of 10,880 (79.6%), it fell below the spring's low first. In the other 43, both happened in the same candle, or the data ended first. We traded 8,218 of the springs. Most of the others came while a trade was already open. Of these trades, 3,073 (37.4%) hit the stop-loss within the first candle after the entry. The take-profit was far away: for half of the trades, more than 4.28 times as far as the stop-loss. So winning 1 trade in 5 was about enough to break even before trading costs: the average trade made +0.005R. But because the stop-loss was close to the entry, trading costs were large compared with the amount at risk. They took 0.195R per trade, and the average trade lost 0.190R.
In a separate test on daily charts, the spring made money on 9 of 10 markets, but it traded only about once a year on each market. These tests are outside the 30 backtests, with more years: 2007 to 2025 on the currency pairs, gold and silver, and 2014 to 2025 on the stock indices. Our rules found 174 springs and traded 139 of them. These trades made +0.638R on average, after trading costs, and only EUR/USD lost money. On a daily chart, the stop-loss is far from the entry compared with the spread, the gap between the buy and the sell price. So trading costs took only 0.042R per trade. But a few trades made most of the profit: the 5 best made +69.7R, and the other 134 made +0.141R on average. As a check, we also tested daily charts on 5 more currency pairs, and only 2 of the 5 made money: AUD/USD and USD/CHF. Together, the 113 trades on all 5 pairs made +0.338R on average, but AUD/USD alone made more than that total.
On the 15-minute, 1-hour and 4-hour charts, stock indices came closest to break-even, and silver did worst. On the 5 indices, the 3,851 trades lost 0.023R on average, and they made +0.098R before trading costs. On the 3 currency pairs, the 2,813 trades lost 0.205R on average, and on gold and silver, the 1,554 trades lost 0.578R. Silver's 680 trades lost 0.953R on average. On silver's 15-minute chart, the spring candle closed less than one spread above the stop-loss in 203 out of 476 trades. The account in that backtest fell 99.5%. The best backtest was the Dow Jones on the 15-minute chart: +131.3% from 2019 to 2025. But it lost 9.1% from 2019 to 2022. Its 10 best trades made +117.1R, more than all 741 trades together (+104.7R). Without those 10 trades, the other 731 lost money.
Springs after a fall lost less than other springs, as Wyckoff's idea of accumulation suggests, but they still lost money. Accumulation is a range after a fall, where big buyers slowly buy. To sort the springs, we used the 200 EMA, a moving average of the last 200 candles. We counted a spring as coming after a fall when its candle closed below the 200 EMA. On the shorter charts (15-minute to 4-hour), these springs lost 0.144R per trade, and springs above the 200 EMA lost 0.243R. They lost less in 2019–2022 and again in 2023–2025. On the 5 stock indices, springs below the 200 EMA made +0.087R per trade, and springs above it lost 0.130R. On gold and silver, springs below the 200 EMA did worse than springs above it. We also tested a version of the rules that trades only springs after a fall. Its trades differ a little: when it skips a spring, it can trade a later spring that an open trade would have blocked. On the indices, its 1,935 trades made +0.070R on average. But its 10 best trades there made more than all 1,935 together. Without those 10, the other trades lost 0.014R on average.
Springs on light volume did worse, not better. Many Wyckoff traders see light volume on a spring as a sign that few sellers are left. Volume is how much trading happened in a candle. We called a spring's volume light when its candle had less volume than the average candle in its range. On the shorter charts (15-minute to 4-hour), springs on light volume won 14.6% of the time and lost 0.296R per trade. Springs with average or heavier volume won 21.5% of the time and lost 0.156R. Light-volume springs lost more on all 3 charts, and in both 2019–2022 and 2023–2025. Only 2,013 out of 8,218 traded springs came on light volume. The candle that breaks below a range usually has more volume than the candles inside it: 1.55 times their average in a typical trade. The version 'Only on light volume' lost 0.325R per trade, more than any other version.
What Wyckoff traders look at
Springs after a fall lost less, and light volume did not help
We split the trades of the standard rules by three things Wyckoff traders look at. These are what came before the range, the spring candle's volume, and how deep the spring went. The numbers cover the 10 markets on 15-minute, 1-hour and 4-hour charts together, from 2019 to 2025. Each result is the average of one trade, in R. ATR is the average size of a candle over the last 14 candles.
| Springs | Trades | Trades that won | Average trade before trading costs | Average trade after trading costs |
|---|---|---|---|---|
| After a fall: the spring candle closed below the 200 EMA | 4,546 | 19.5% | +0.062R | −0.144R |
| Not after a fall: the spring candle closed above the 200 EMA | 3,654 | 20.2% | −0.062R | −0.243R |
| Light volume: the spring candle had less volume than the average candle in its range | 2,013 | 14.6% | −0.040R | −0.296R |
| Average volume or more | 6,205 | 21.5% | +0.020R | −0.156R |
| A shallow spring: less than 0.25 ATR below the range's low | 4,673 | 17.2% | +0.019R | −0.211R |
| A deeper spring: 0.5 to 1 ATR below the range's low | 1,367 | 27.2% | −0.004R | −0.120R |
| All springs | 8,218 | 19.8% | +0.005R | −0.190R |
18 trades came before the 200 EMA had 200 candles, so the two EMA rows add up to 8,200 trades. Springs 0.25 to 0.5 ATR deep are not shown: their 2,178 trades lost 0.190R on average. A deeper spring puts the stop-loss further from the entry. So trading costs are a smaller part of its risk. Its take-profit is also closer, when measured in stop-loss distances, so price reaches it more often. Before trading costs, shallow and deeper springs did about the same. Light volume is a different case. Light-volume springs usually had a take-profit further away, in stop-loss distances, than other springs. With the take-profit at a similar distance, they still won a little less often.
Every version
On the shorter charts, no version made money on currency pairs, gold or silver
Each number is the average result of one trade, in R, after trading costs. The columns from '3 currency pairs' to 'All 10 markets' cover 2019 to 2025, with the 15-minute, 1-hour and 4-hour charts together. The last two columns are the separate tests on daily charts: on the same 10 markets, and on 5 more currency pairs (see 'Daily charts' further down). The four changes are explained under 'Can the Wyckoff spring be improved?' below.
| Version | 3 currency pairs | Gold and silver | 5 stock indices | All 10 markets | Daily charts, 10 markets | Daily charts, 5 more currency pairs |
|---|---|---|---|---|---|---|
| Standard rules | −0.205R | −0.578R | −0.023R | −0.190R (8,218 trades) | +0.638R (139 trades) | +0.338R (113 trades) |
| With a retest of the spring | −0.171R | −0.506R | −0.015R | −0.189R (2,359 trades) | +1.073R (46 trades) | −0.154R (34 trades) |
| Only on light volume | −0.286R | −0.761R | −0.091R | −0.325R (2,278 trades) | +0.261R (52 trades) | +0.242R (54 trades) |
| A higher take-profit | −0.146R | −0.536R | +0.045R | −0.130R (7,967 trades) | +0.279R (131 trades) | +0.373R (113 trades) |
| Only after a fall | −0.139R | −0.625R | +0.070R | −0.165R (4,727 trades) | +0.373R (60 trades) | +0.276R (75 trades) |
A version's trades are not exactly the springs in the table above. When a version skips a spring, it can trade a later spring that an open trade would have blocked. On the stock indices, a few big winners made the profit of the two versions that made money there. Without the 10 best trades on the indices, both versions lost money there: −0.040R per trade with the higher take-profit, and −0.014R with only springs after a fall. On daily charts, the versions had only 34 to 139 trades each.
The strategy
What the Wyckoff spring is
Richard Wyckoff was a stock trader and writer in the early 1900s. His method treats price moves as the actions of big buyers and sellers. After a fall, price often moves sideways for a while, with a clear low and a clear top. This sideways stretch is a trading range. Wyckoff called a range after a fall accumulation: big buyers slowly buy there. A spring is a quick drop below the range's low that comes straight back inside the range. Wyckoff traders read a spring as the last move down before price rises. The drop makes some traders sell, and it hits other traders' stop-losses below the range. The big buyers buy from all of these sellers. After a real spring, price should not fall below the spring's low again. It should rise to the top of the range, or further. Many traders wait for a retest before they buy: a second dip, often on lower volume, that stays above the spring's low. Wyckoff called it a test. Traders judge a Wyckoff range by eye, so two traders can see different springs on the same chart. To backtest the spring, we wrote exact rules for a range and a spring, using the ATR. ATR (average true range) is the average size of a candle over the last 14 candles. It grows and shrinks with volatility, how far price moves, so a rule written in ATR fits every market and chart. Below are the exact settings and test details.
- Trading range
- At least 30 candles, and no taller than 4 × ATR(14) from the highest high to the lowest low. The range goes back from the candle before the spring for as long as it stays within 4 ATR, so it is as long as it can be. In a typical trade (the median), the range was 36 candles long and 3.8 ATR tall.
- Spring
- A candle whose low goes below the range's low by at most 1 × ATR(14), and that closes back inside the range. In a typical trade, its low went 0.21 ATR below the range's low, and it closed a tenth of the way up the range.
- Entry
- Buy at the open of the candle after the spring.
- Stop-loss
- At the spring candle's low. It does not move. In a typical trade, it was 0.76 × ATR(14) below the entry price. A stop-loss does not protect against a gap, a jump in price between two candles (see 'Risk per trade').
- Take-profit
- At the range's top. For half of the trades, the take-profit was more than 4.28 times as far from the entry as the stop-loss.
- Trading costs
- Included: the spread, the gap between the buy price and the sell price. The entry is a market order, an order that fills at the current price, and it pays half the spread. A trade that hits its stop-loss pays the other half. The take-profit is a limit order, an order that waits at a set price, and in our backtest engine it pays no spread. So a winning trade pays half the spread, and a losing trade pays all of it. There is no commission. No slippage is added either: slippage is a fill at a worse price than the order's price. A stop-loss fills at its price, unless price jumps past it. With the standard rules, trading costs took 0.212R per trade on 15-minute charts, 0.128R on 1-hour charts, 0.067R on 4-hour charts and 0.042R on daily charts.
- Spreads
- The spread is 0.8 pips on EUR/USD, 1.2 on GBP/USD and 0.9 on USD/JPY. A pip is 0.0001, or 0.01 on USD/JPY. The spread is $0.30 an ounce on gold and $0.020 an ounce on silver. On the indices, it is 0.4 index points on the S&P 500, 1.5 on the Nasdaq 100 and 2.0 on the Dow Jones. On the DAX and the FTSE 100, it is 1.0 point. On silver, the Nasdaq 100, the Dow Jones, the DAX and the FTSE 100, these spreads are wider than the default spreads in our backtest engine. We widened them to match a typical retail broker. On the 5 currency pairs we added on daily charts, the spread is 1.0 pips on AUD/USD, 1.3 on USD/CAD, 1.5 on USD/CHF, 1.4 on EUR/JPY and 2.5 on GBP/JPY.
- Take-profit fills
- A real take-profit may not fill when price only touches it. On 15-minute charts, 79.8% of the take-profits were reached by a candle whose high went at least one full spread above the take-profit. On 1-hour charts, 88.0% were, and on 4-hour charts, 89.8%. The rest only just reached the take-profit, and some of them might not have filled in real trading.
- Risk per trade
- 1% of the account balance, calculated from the spring candle's close. The account started at $10,000. The trade opens at the next candle's open and pays half the spread, so its real risk was a little different: 1.02% in a typical trade. On the shorter charts, 162 trades risked more than 1.5% of the account. 132 of them were on silver's 15-minute chart, where the stop-loss was often closer to the entry than the spread. On daily charts, 2 trades risked more than 1.5%. The larger one, a silver trade in 2018, risked 6.3%: the spring candle closed only 1 cent above the stop-loss, and the next candle opened about 4 cents higher. The biggest loss was 7.84R, far more than the 1R a trade risks. It was a GBP/USD 1-hour trade that stayed open over a weekend in early 2025. When the market opened again, price was far below the stop-loss.
- Stop-loss and take-profit in one candle
- When one candle reaches both the stop-loss and the take-profit, the test uses 1-minute prices to see which was reached first. If both were reached in the same minute, the test counts the trade as a stop-loss.
- Smallest trade size
- Some springs were skipped. For them, a trade that risks 1% would have been smaller than the smallest trade the market allows. The smallest trade is 1,000 units of a currency pair, 1 ounce of gold, 50 ounces of silver or 0.1 of an index contract. With the standard rules, 630 springs were skipped this way. 624 of them were on silver's 15-minute chart, from October 2020, after the account in that backtest had fallen below $1,000. The other 6 were on gold's 15-minute chart and silver's 1-hour chart, in 2025. No spring was skipped this way on 4-hour or daily charts. A stop-loss very close to the entry can also need a bigger trade than our backtest engine allows: 30 times the account balance. Then the trade uses the biggest size allowed, and it risks less than 1%. 1,739 out of 8,218 trades risked less than 0.9% of the account, most of them because of this limit. For 32 more springs, no trade opened. The trade was sized at this limit, and at the next candle's open it was just over the limit, so the engine refused it.
- Volume
- Volume is how much trading happened in a candle. We use the volume that comes with our price data for each 1-minute candle, and a longer candle's volume is the total of its 1-minute candles. This volume comes from one price source, not from an exchange: forex has no central exchange, and our prices are CFD prices (see 'Price data'). So it does not show how much the whole market traded. On most forex and CFD charts, volume also comes from one price source, often as tick volume: the number of price changes in a candle. To decide if a spring is on light volume, we compare the spring candle's volume with the average volume of the candles in its range. In a typical trade, the spring candle had 1.55 times that average.
- Overnight fees
- Not included: the trading costs on this page are the spread only. Brokers charge a fee (a swap), or sometimes pay one, for each night a trade stays open. Most spring trades closed fast. Half of them closed within 24 minutes on 15-minute charts, within 2 hours on 1-hour charts and within 12 hours on 4-hour charts. On daily charts, half were open for 2.4 days or more. So overnight fees would matter most there, and the daily results are before these fees.
- Price data
- CFD prices: the prices a broker quotes for a CFD (contract for difference), the product most brokers outside the US offer for trading these markets. The 15-minute, 1-hour and 4-hour tests start in 2019, because our index data before mid-2018 has hours with no price change on charts shorter than daily. The rules leave out two kinds of candle, and these candles don't count for the range, the ATR or the 200 EMA. The first kind is tiny candles built from a few minutes of prices after the Friday close (17:00 New York time) on some Fridays. The second is candles with no price change (the high equals the low), for example on market holidays. From 2019 to 2025, they were less than 0.5% of the candles in every test, except on the FTSE 100 (1.4% to 2.0%) and the DAX 1-hour chart (0.8%). In the daily tests on the indices from 2014, they were up to 5.7% of the candles, on the S&P 500.
- Wrong prices in our data
- They changed the results only a little. Our DAX, FTSE 100 and silver data has some wrong prices, about 1% away from the real price, at times with no trading. From 2019 to 2025, they appear on 44 days on the DAX, 9 days on the FTSE 100 and 1 day on silver. On the DAX, most come between 22:00 and 23:00 German time. A spring is a false break of a range's low, so one wrong price can make a spring, or remove one. So we ran these tests again without the wrong prices. With the standard rules, the wrong prices added 16 trades and removed 13, out of 2,143 trades on these 3 markets. The biggest change was the DAX 1-hour backtest: +10.1% without the wrong prices, instead of +2.6%. With a retest of the spring, the DAX 15-minute backtest lost 2.6% in 2019–2022 without them, instead of making 6.0%. That version would then have made money in 8 of 30 tests in 2019–2022, not 9, the same as 'Only after a fall'. Its middle test was better, so it would still be marked 'Best on 2019–2022'. The daily tests start earlier, and our data has more wrong prices in those years. They are on the S&P 500, the Nasdaq 100, the DAX and the FTSE 100 from 2014 to 2018, and on silver from 2009 to 2016. Without them, the DAX daily backtest made +1.7% instead of +7.0%, and 'Only after a fall' made no trades there instead of 3. The other daily changes were small.
- Daily charts
- We tested daily charts on their own, over more years, because a daily spring is rare. These tests are separate from the 30 backtests. We tested the currency pairs, gold and silver from 2007 to 2025, split into 2007–2018 and 2019–2025. We tested the stock indices from 2014 to 2025, split into 2014–2021 and 2022–2025. Our rules found 174 springs, about 1 a year on each market, and traded 139 of them: 7 to 20 on each market. In the first period of each split, 9 of the 10 markets made money, and in the second period, 6 did. As a check, we also tested daily charts on 5 more currency pairs from 2007 to 2025: AUD/USD, USD/CAD, USD/CHF, EUR/JPY and GBP/JPY. The AUD/USD backtest made +58.1% and the USD/CHF backtest +17.0%, and the other 3 lost money. The daily results are in the 'Every version' table above, not in the results grid.
The rules we tested
- Find a trading range: at least 30 candles in a row whose highest high is no more than 4 ATR above their lowest low. ATR(14) is measured on the candle just before the spring candle. The range starts at that candle and goes back as far as the candles stay within 4 ATR.
- A spring is a candle whose low goes below the range's lowest low by no more than 1 ATR. The candle must also close back inside the range: above the range's low and below its top.
- Buy at the open of the next candle.
- Put the stop-loss at the spring candle's low, and the take-profit at the range's top. Neither one moves.
- Hold one trade at a time. A spring that comes while a trade is open is not traded.
- Risk 1% of the account on each trade.
The method
How we tested it
The test
- Markets
- EUR/USD, GBP/USD, USD/JPY, Gold, Silver, S&P 500, Nasdaq 100, Dow Jones, DAX and FTSE 100
- Charts
- 15-minute, 1-hour and 4-hour candles
- Dates
- January 2019 to December 2025
- Account
- $10,000, 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 2019 to 2022 and again from 2023 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 2023 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 Wyckoff spring be improved?
We tested four changes to the rules, one at a time. Each set of rules is a version, and the standard rules are the first. Two changes come from what Wyckoff traders look for: a retest of the spring, and light volume on the spring. One trades only springs after a fall, as in Wyckoff's accumulation. The last one moves the take-profit further away: one range height above the range's top. Most changes trade fewer springs, so more tests have fewer than 100 trades: 15 of the 30 with the standard rules, and up to 21 with 'Only on light volume'. The two 'Made money' columns count every test, also those with fewer than 100 trades, and each period on its own. So they can be higher than the 'Made money' number at the top of the page, which counts only tests with at least 100 trades, over 2019 to 2025. A test with fewer than 100 trades can't count as 'made money in both periods' or 'passed all four checks'.
| Version | Made money 2019–2022years used to pick the best version | Made money 2023–2025years not used to pick it | Made money in both periods | Passed all four checks |
|---|---|---|---|---|
| Standard rules The rules under 'The rules we tested', with nothing changed. | 6 of 30middle test −9.6% | 11 of 30middle test −4.1% | 0 | 0 |
| With a retest of the springBest on 2019–2022 This version does not buy on the spring. It waits up to 10 candles for a retest. A retest is a candle whose low reaches the range's low but stays above the spring's low, and that closes above the range's low. Then the version buys at the next candle's open, with the same stop-loss and take-profit. If price reaches the spring's low or the range's top first, there is no trade. The retest has no volume rule. | 9 of 30middle test −2.7% | 10 of 30middle test −1.8% | 1 | 1 |
| Only on light volume This version only trades a spring whose candle had less volume than the average candle in its range. Many Wyckoff traders read a spring on light volume as a sign that few sellers are left. | 7 of 30middle test −5.3% | 9 of 30middle test −4.2% | 0 | 0 |
| A higher take-profit The take-profit is one range height above the range's top, instead of at the top. Wyckoff traders call the rise after accumulation the markup. | 6 of 30middle test −10.9% | 9 of 30middle test −4.9% | 0 | 0 |
| Only after a fall This version only trades a spring whose candle closed below the 200 EMA, a moving average of the last 200 candles that gives more weight to recent candles. A close below the 200 EMA usually means price had been falling, as in Wyckoff's accumulation. | 8 of 30middle test −5.1% | 14 of 30middle test −0.8% | 2 | 0 |
No. On the shorter charts, every version still lost money per trade across the 10 markets. The best, 'A higher take-profit', lost 0.130R per trade, and it won only 10.9% of its trades. 'With a retest of the spring' made money in the most tests in 2019–2022: 9 of 30, against 6 with the standard rules. In 2023–2025, it made money in 10 of 30, against 11. Its trades won more often, 23.1% against 19.8%, but the average trade stayed the same: −0.189R, against −0.190R. Only 1 of its 30 backtests passed all four checks: the S&P 500 on the 15-minute chart, which made +42.5% from 2019 to 2025. 'Only after a fall' made money in the most tests in 2023–2025: 14 of 30. On the stock indices, it made +0.070R per trade, but without its 10 best trades there, it lost 0.014R per trade. 'Only on light volume' did worst: −0.325R per trade. On the daily charts of the 10 markets, every version made money on average, but with only 46 to 139 trades in 12 to 19 years. That is too few to rely on, and on the 5 more currency pairs, 'With a retest of the spring' lost money.
Passed all four checks with a retest of the spring
- S&P 500, 15-minute
Questions
Questions traders ask
What is the success rate of the Wyckoff spring?
In our backtest, 19.8% of the spring trades won: 1,624 out of 8,218 on 15-minute, 1-hour and 4-hour charts from 2019 to 2025. A trade won when price reached the take-profit at the range's top before the stop-loss at the spring's low. After trading costs, a winning trade made 3.46R on average. A losing trade lost 1.09R on average: a little more than 1R, because it also paid trading costs. So the trades needed to win about 24% of the time to break even, and they won only 19.8%. On daily charts, 24.5% of the 139 trades won, and the trades made +0.638R on average.
Should you wait for a test after a Wyckoff spring?
No. In our backtest, waiting for the test changed little. We waited up to 10 candles for a retest. That is a candle that dips back to the range's low, stays above the spring's low, and closes above the range's low. On 15-minute, 1-hour and 4-hour charts, these trades won 23.1% of the time, compared with 19.8% when we bought right after the spring. But the average trade stayed the same: −0.189R, against −0.190R. Waiting also gave far fewer trades: 2,359, compared with 8,218. On daily charts, waiting gave 46 trades on 10 markets, and they made +1.073R on average, but that is too few trades to judge. On 5 more currency pairs, the 34 daily trades with a retest lost 0.154R on average.
Is a Wyckoff spring on low volume better?
No. In our backtest, springs on light (low) volume did worse than other springs. A spring was on light volume when its candle had less volume than the average candle in the range. These springs won 14.6% of the time, and lost 0.296R per trade after trading costs. Springs with average or heavier volume won 21.5% of the time, and lost 0.156R. Light-volume springs lost more on 15-minute, 1-hour and 4-hour charts, and in both 2019–2022 and 2023–2025. The volume here is the volume in our price data, from one price source, as on most forex and CFD charts. It does not show how much the whole market traded.
What is the best timeframe for the Wyckoff spring?
In our backtest, only the daily chart made money after trading costs, but it gave very few trades. On daily charts, the 139 trades on 10 markets made +0.638R on average. That is about 1 trade a year on each market, and the 5 best trades made most of the profit. On 5 more currency pairs, only 2 of the 5 made money on daily charts. On 15-minute charts, the 6,792 trades lost 0.185R on average. On 1-hour charts, the 1,129 trades lost 0.234R, and on 4-hour charts, the 297 trades lost 0.137R. Trading costs took 0.212R per trade on 15-minute charts, and only 0.042R on daily charts.
Keep exploring
Related strategies we tested
Chart patterns Cup and handle pattern backtest Made money in 1 of 40 tests, 0 of them in both periods. Best: USD/CAD, 1-hour, +19% Chart patterns Head and shoulders pattern backtest Made money in 6 of 30 tests, 4 of them in both periods. Best: Gold, 1-hour, +37% Chart patterns 7 well-known chart patterns, backtested and ranked Made money in 7 of 30 tests, 2 of them in both periods. Best: S&P 500, 4-hour, +49%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 WyckoffSpring(Strategy):
"""The Wyckoff spring, coded as a false break below a trading range, plus the changes
Wyckoff traders argue about. Wyckoff traders judge ranges and springs by eye; these rules
make them exact.
ATR is ATR(14), Wilder's average true range: the average size of a candle over the last 14
candles. It is worked out here, so that the skipped candles (below) don't count.
The trading range: count back from the candle before the spring candle. The range is the
longest run of candles whose highest high and lowest low are at most 4 ATR apart (the ATR
of the candle before the spring candle). It must be at least 30 candles long. Its lowest
low is the range low (the support), and its highest high is the range top.
The spring: a candle whose low goes below the range low, by at most 1 ATR, and that closes
back above the range low (and below the range top). Buy at the next candle's open.
Stop-loss: the spring candle's low. Take-profit: the range top. Size every trade so that a
loss at the stop-loss costs 1% of the account. One trade at a time: a spring that comes
while a trade is open is skipped. The stop-loss and the take-profit don't move.
Skipped candles: a candle that opens on a Friday at 17:00 New York time or later (a few
minutes of prices after the weekly close make a tiny extra candle in some years), and a
candle whose high equals its low (a market holiday in the price data). The strategy ignores
them: they are never part of a range or a spring, and they are left out of the ATR, the EMA
and the average volume. On a market that trades at the weekend, remove the Friday skip.
The changes, one at a time (off by default):
test = 1 wait for a test of the spring. For up to 10 candles after the spring
candle, wait for a candle whose low comes back down to the range low (or
below it, but not below the spring low) and that closes above the range
low. Buy at the next candle's open, with the same stop-loss and
take-profit. The setup is dropped if a candle goes below the spring low,
or reaches the range top, first.
low_volume = 1 only springs on light volume: the spring candle's volume is below the
average volume of the range's candles
markup = 1 hold for the markup: the take-profit is one range height (range top minus
range low) above the range top, instead of the range top
trend = -1 only after a fall: the spring candle closes below the 200 EMA, a moving
average of the last 200 closes (trend = 1: only above it)
"""
params = {"range_bars": Param(type=int, default=30, min=5, max=1000),
"range_atr": Param(type=float, default=4.0, min=0.5, max=50.0),
"depth_atr": Param(type=float, default=1.0, min=0.0, max=10.0),
"atr_period": Param(type=int, default=14, min=2, max=100),
"test": Param(type=int, default=0, min=0, max=1),
"test_bars": Param(type=int, default=10, min=1, max=100),
"low_volume": Param(type=int, default=0, min=0, max=1),
"markup": Param(type=int, default=0, min=0, max=1),
"trend": Param(type=int, default=0, min=-1, max=1),
"ema_period": Param(type=int, default=200, min=2, max=1000)}
lookback = 2
def init(self, ctx):
self.p = ctx.params
self.ny = ctx.clock("America/New_York")
self.k = -1 # number of the newest candle the strategy uses
self.hi = [] # highs, lows and volumes of those candles
self.lo = []
self.vol = []
self.prev_close = None
self.atr = None # Wilder's ATR, as the built-in one
self.atr_sum = 0.0
self.atr_n = 0
self.ema = None # the EMA of the closes
self.ema_sum = 0.0
self.ema_n = 0
self.pending = None # test = 1: a spring waiting for its test
# ---- candles, ATR, EMA, ranges ---------------------------------------------------------
def update_averages(self, high, low, close):
p = self.p
tr = high - low if self.prev_close is None else max(
high - low, abs(high - self.prev_close), abs(low - self.prev_close))
self.prev_close = close
n = p.atr_period
self.atr_n += 1
if self.atr_n < n:
self.atr_sum += tr
elif self.atr_n == n:
self.atr = (self.atr_sum + tr) / n
else:
self.atr += (tr - self.atr) / n
m = p.ema_period
self.ema_n += 1
if self.ema_n < m:
self.ema_sum += close
elif self.ema_n == m:
self.ema = (self.ema_sum + close) / m
else:
self.ema += (close - self.ema) * 2.0 / (m + 1)
def find_range(self, k):
"""The trading range just before candle k: (first candle, range low, range top), or None.
It is the longest run of candles up to candle k - 1 that fits in range_atr ATR."""
p = self.p
if self.atr is None or k < p.range_bars:
return None
limit = p.range_atr * self.atr
top = low = None
first = None
j = k - 1
while j >= 0:
t = self.hi[j] if top is None else max(top, self.hi[j])
b = self.lo[j] if low is None else min(low, self.lo[j])
if t - b > limit:
break
top, low, first = t, b, j
j -= 1
if first is None or k - first < p.range_bars:
return None
return first, low, top
def step(self, high, low, close, volume):
"""Add one candle. Returns the spring that closed on this candle, or None."""
p = self.p
k = self.k + 1
sig = None
# A spring breaks the range low, so its low must be below the lows of the last
# range_bars candles. Only then is the range worked out.
if k >= p.range_bars and low < min(self.lo[k - p.range_bars:k]):
rng = self.find_range(k)
if rng is not None:
first, r_low, r_top = rng
a = self.atr
if r_low - a * p.depth_atr <= low < r_low < close < r_top:
avg_vol = sum(self.vol[first:k]) / (k - first)
sig = {"k": k, "first": first, "range_low": r_low, "range_top": r_top,
"spring_low": low, "close": close, "atr": a,
"volume": volume, "avg_volume": avg_vol}
self.k = k
self.hi.append(high)
self.lo.append(low)
self.vol.append(volume)
self.update_averages(high, low, close)
if sig is not None:
sig["ema"] = self.ema
return sig
# ---- trading ---------------------------------------------------------------------------
def allowed(self, sig):
"""The filters of the changes, checked on the spring candle."""
p = self.p
if p.low_volume == 1 and not sig["volume"] < sig["avg_volume"]:
return False
if p.trend != 0:
e = sig["ema"]
if e is None:
return False
if p.trend == -1 and not sig["close"] < e:
return False
if p.trend == 1 and not sig["close"] > e:
return False
return True
def on_bar(self, ctx):
# A tiny candle after the Friday close, or a candle with no price movement: ignore it.
if self.ny.weekday == 4 and self.ny.hour >= 17:
return
high, low, close = ctx.high[0], ctx.low[0], ctx.close[0]
if not high > low:
return
sig = self.step(high, low, close, ctx.volume[0] or 0.0)
if ctx.position is not None:
self.pending = None
return
if sig is not None and not self.allowed(sig):
sig = None
if self.p.test == 1:
setup = self.pending
if setup is not None:
if low <= setup["spring_low"] or high >= setup["range_top"] \
or self.k - setup["k"] > self.p.test_bars:
self.pending = None # broke the spring low, reached the top, or no test in time
elif low <= setup["range_low"] < close:
self.pending = None
if ctx.gate("test_of_the_spring", True):
self.enter(ctx, setup, close)
return
if ctx.gate("spring", sig is not None):
self.pending = sig
return
if not ctx.gate("spring", sig is not None):
return
self.enter(ctx, sig, close)
def enter(self, ctx, sig, entry):
stop = sig["spring_low"]
target = sig["range_top"]
if self.p.markup == 1:
target = sig["range_top"] + (sig["range_top"] - sig["range_low"])
if not ctx.gate("room_to_target", stop < entry < target):
return None
size = ctx.size_by_risk_pct(0.01, stop=stop, entry=entry)
return ctx.buy(size=size, sl=stop, tp=target, tag="wyckoff_spring")
Your turn
Test the Wyckoff spring yourself
Change the settings, the market or the stop-loss and see the result in seconds. Ask the BacktestingMax AI: Backtest the Wyckoff spring on the S&P 500, 15-minute chart, from 2019 to 2025: a trading range of at least 30 candles no taller than 4 ATR(14), buy when a candle dips below the range's low by at most 1 ATR and closes back inside the range, stop-loss at that candle's low, take profit at the range's top, risk 1% per trade
.
Free account. No credit card needed.
See every strategy we tested Browse strategies shared by tradersBacktest 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.