DeployQuantSOXL trading strategies, backtested
Direxion Daily Semiconductor Bull 3X Shares: 3x daily leveraged semiconductors, one of the more volatile ETFs listed. Every DeployQuant template run on SOXL over 5.7 years of minute data, same engine, same window, sorted by return.
SOXL is a 3x daily leveraged fund on the semiconductor index that SOXX holds. From 2021-01-04 to 2026-10-02 it returned 33.32% a year in the test and turned $10,000 into $52,126. The path was extreme. It fell 89.62% from 2021-12-27 to 2022-10-14, took 1,230 days from the low to get back to the old high, and then rose 289.88% in the part of 2026 in the data before dropping 69.27% from 2026-06-22 to 2026-07-29.
Twelve templates were run on it. Two beat buy-and-hold on CAGR: golden cross at 45.93% and RSI(2) snapback at 39.24%. Eleven of the twelve had a shallower maximum drawdown than holding, though the shallowest was still 58.51%. Golden cross ended with $87,566 against $52,126 for holding, and it did that with 3 closed round trips and one open position. Three trades are a small sample, and a few of them carry the result.
On SOXL a rule that spends time in cash can win on return as well as risk, because the fund's own loss in 2022 was so large that sidestepping part of it paid for missed rallies. The window is one 5.7-year stretch with a single large bear market in it. Headline numbers have no fees, and the cost runs near the end of the trading section show how the busy templates change when they are added.
| Strategy | CAGR | max DD | Sharpe | trades | win rate | final value |
|---|---|---|---|---|---|---|
| Golden Cross (SMA 50/200) | 45.9% | −69.3% | 0.88 | 3 | 67% (+1 open) | $87,566 |
| RSI(2) Dip Snapback | 39.2% | −58.5% | 0.81 | 167 | 58% | $66,897 |
| First-to-Last Day of Month | 32.9% | −90.2% | 0.81 | 69 | 57% (+1 open) | $51,092 |
| Weekly Entry + 7% Target | 29.5% | −71.4% | 0.73 | 282 | 60% | $44,050 |
| EMA 12/26 Trend | 26.2% | −65.4% | 0.69 | 22 | 36% (+1 open) | $38,043 |
| 200-Day SMA Regime Filter | 23.4% | −76.7% | 0.66 | 22 | 36% (+1 open) | $33,501 |
| 20-Day Momentum + Trailing Stop | 19.1% | −67.5% | 0.59 | 138 | 37% (+1 open) | $27,225 |
| SMA 10/50 Trend | 18.2% | −72.2% | 0.60 | 17 | 47% (+1 open) | $26,110 |
| SMA-200 Trend + 15% Trailing Stop | 17.8% | −82.0% | 0.59 | 97 | 35% (+1 open) | $25,636 |
| 3-Month Momentum Switch | 16.4% | −74.2% | 0.58 | 26 | 42% | $23,950 |
| Drawdown Dip Buyer + 8% Target | 16.4% | −88.5% | 0.67 | 17 | 100% (+1 open) | $23,941 |
| RSI(14) Mean Reversion | 14.4% | −81.0% | 0.54 | 14 | 71% | $21,682 |
| Buy & hold SOXL | 33.3% | −89.6% | 0.82 | – | – | $52,126 |
The two templates that beat holding, and the ten that did not
Holding gained 119% in 2021, lost 84.8% in 2022, gained 211.2% in 2023, lost 12.1% in 2024, then gained 53.6% in 2025 and 285.2% in 2026 so far. The 2022 loss is the number that sets the terms. A fund that loses 84.8% needs a gain of several hundred percent to recover, and any template that kept some capital out of that fall started 2023 from a much higher base.
Golden cross is the clearest case. Its first trade, from 2021-10-19 to 2022-03-11, lost 20%, and then it stayed out. Its second, from 2023-03-13 to 2023-11-09, made 41.67%, and its third, from 2023-12-12 to 2024-09-18, made 27.46%. It bought again on 2025-08-11 at an adjusted price of 26.43, and that position was still open at the end with a gain of 519.56%. The profit factor was 3.24 on those three closed trades. The drawdown was 69.3% and it began on 2026-06-22. In 2022 it lost 48.8% against 84.8% for the fund, and it beat holding in 2022, 2024, 2025 and 2026. It trailed badly in 2023, 139.9 points behind, after sitting in cash until 2023-03-13.
RSI(2) snapback made 167 round trips with a 58% win rate and a profit factor of 1.3. It returned 39.24%, with a 58.51% drawdown. The average win was 8.04% and the average loss was 7.14%, so the edge was small per trade. Its best trade was a gain of 36.11% over 4 days in June 2026 and its worst a loss of 32.6% over 10 days from 2025-02-24 to 2025-03-06. In 2022 it lost 33% against 84.8%. In 2026 it gained only 3.2% against 285.2%, which was 282 points behind and the largest single-year gap in the table. It made 20 round trips that year and won 9.
Monthly cycle returned 32.86% and had a 90.24% drawdown, a little deeper than holding. It held 95.2% of the time, so it copies the fund with small differences. Its worst trade was June 2022, a loss of 48.67%, and its best was April 2026 at 143.73%. It beat holding in 2024, 2025 and 2026 and fell short in 2021 and 2023.
Weekly 7% target returned 29.47% with 282 round trips and a 60% win rate. Its average win was 7.58% and its average loss was 8.93%, so the 60% win rate was needed to keep it profitable. Its worst trade was 3 days in March 2021, a loss of 30.27%. It was in the market 45.4% of the time and had a 71.36% drawdown. The 2021 year was a loss of 34.6% against a gain of 119%.
EMA 12/26 trend returned 26.2% with 22 round trips and a 36% win rate, which paid because winners averaged 52.61% and losers 11.93%. Its best trade, from 2026-04-10 to 2026-07-08, gained 119.26%. It lost on all 5 round trips closed in 2021. The 200-day filter returned 23.44% with a 36% win rate and an unusually deep drawdown of 76.74% that began on 2026-06-22 and had not recovered by the end of the data. The 10 round trips it closed in 2023 won 3. Momentum breakout returned 19.06% from 138 round trips, SMA 10/50 returned 18.19%, trend plus trailing stop returned 17.82% from 97 round trips and a profit factor of 1.05, and 3-month momentum returned 16.43%.
Trend plus trailing stop made 39 round trips in 2026 alone on a fund with a 7.77% average intraday range, and its drawdown of 81.96% was the third deepest of the twelve.
The drawdown dip buyer returned 16.42% from 17 round trips, every one a winner. Its exposure was 96.8% and its drawdown was 88.53%. A rule that buys a 10% drop and waits for 8% needs the fund to come back, and in 2022 it did not for a long time. The longest hold in the trade list was 1,584 days and the open position at the end, entered on 2026-06-22 at 299.42, was down 45.31%. A 100% win rate here describes closed trades only.
RSI(14) mean reversion returned 14.43% with a 71% win rate. Its worst trade, from 2022-01-21 to 2022-07-21, lost 59.44% over 181 days, and its drawdown of 81.01% took 1,399 days to recover.
Other leveraged funds give a useful comparison. TQQQ held at 25.35% and RSI(2) snapback made 39.31% there. QLD held at 23.55% and RSI(2) made 26.17%. SSO held at 21.92% and RSI(2) made 24.56%, and SPUU held at 22.48% against 24.12%. On TECL the fund held at 38.16% and monthly cycle made 35.97%, and on FAS the dip buyer made 24.15% against 18.34% for holding. On TMF holding lost 31.16% a year and the best template lost 4%. RSI(2) snapback was the best template on five of these nine funds, which fits the mean reversion statistics for SOXL further down.
How each strategy traded SOXL
| Strategy | Time in market | Avg hold (days) | Best trade | Worst trade | Profit factor | With 10 bps slippage |
|---|---|---|---|---|---|---|
| golden cross | 51.8% | 222 | 41.7% | −20.0% | 3.24 | 45.8% |
| RSI(2) snapback | 38.2% | 5 | 36.1% | −32.6% | 1.30 | 31.5% |
| monthly cycle | 95.2% | 28 | 143.7% | −48.7% | 1.58 | 29.7% |
| weekly 7% target | 45.4% | 3 | 21.5% | −30.3% | 1.31 | 20.1% |
| EMA 12/26 trend | 54.6% | 52 | 119.3% | −23.1% | 2.73 | 25.3% |
| 200-day regime filter | 50.4% | 48 | 115.7% | −28.8% | 1.62 | 22.4% |
| momentum breakout | 37.0% | 6 | 100.7% | −11.8% | 1.20 | 13.6% |
| SMA 10/50 trend | 53.7% | 66 | 122.8% | −35.0% | 2.04 | 17.5% |
| trend + trailing stop | 47.1% | 10 | 117.6% | −18.2% | 1.05 | 14.1% |
| 3-month momentum | 52.7% | 43 | 124.8% | −44.5% | 1.78 | 15.3% |
| dip buyer | 96.8% | 114 | 15.3% | 8.0% | – | 21.3% |
| RSI mean reversion | 32.5% | 49 | 44.4% | −59.4% | 1.95 | 13.8% |
Exposure, costs and what the trade lists show
Time in market ran from 96.8% for the dip buyer and 95.2% for monthly cycle to 32.5% for RSI(14). Golden cross was in 51.8% of the time and still beat holding by 12.61 points of CAGR, which shows how much of the fund's return came in a few stretches. The 200-day filter was in 50.4% of the time, trend plus trailing stop 47.1%, weekly 7% target 45.4%, RSI(2) 38.2% and momentum breakout 37%.
Holding periods split into two groups. Weekly 7% target held a median of 2 days and RSI(2) a median of 4, and the trend templates held a median of 241 days for golden cross, 35 for EMA 12/26 and 65 for SMA 10/50. The average holding period for trend plus trailing stop was 10.1 days, which is short for a template meant to ride a trend, and shows that the 15% trail was hit often on a fund this volatile.
Costs matter most for the busy templates. RSI(2) snapback went from 39.24% to 35.34% at 5 basis points and 31.52% at 10, and still finished ahead of holding. Weekly 7% target dropped from 29.47% to 22.97% and 20.11%, which puts it behind holding. Momentum breakout dropped from 19.06% to 16.2% and 13.61%. Trend plus trailing stop dropped from 17.82% to 15.86% and 14.06%. Golden cross barely moved, 45.83% and 45.78%. Monthly cycle moved to 31.13% and 29.69%.
One result in the cost table looks wrong. The dip buyer returned 16.42% with no costs, 16.35% with 5 basis points and 21.31% with 10 basis points, with end equity of $30,316 at 10 basis points against $23,941 with no costs. Higher costs should not produce a better result. With 17 closed trades and one long open position the result is set by a few entries. The figures are reported as the engine produced them.
SOXL's liquidity is high, with $2,243,524,454 of average daily dollar volume and a median minute volume of 101,543.5 shares. The 5 and 10 basis point runs are the more realistic guide for templates that make hundreds of trades in a fund that moves 7.77% in an average intraday range.
April and July 2026 in the trade lists
Two months in 2026 dominate the month-by-month results. Buy-and-hold SOXL made 162.61% in April 2026 and lost 56.9% in July 2026. Almost every template shows the same pair as its best and worst month, so much of the window's return came from one rally and one reversal. The golden cross made 163.1% in April and lost 56.93% in July, so it was fully invested through both. The monthly cycle made 143.73% on its first-to-last-day trade in April 2026 and 87.87% in May, and its worst trades were June 2022 at -48.67%, July 2026 at -45.68% and January 2022 at -43.64%.
The trend templates built their best trades inside the April rally. The 200-day filter made 115.68% from 2026-04-01 to 2026-07-30, which held through the July fall and still closed higher. SMA 10/50 made 122.76% from 2026-04-14 to 2026-07-14, and the 3-month momentum switch made 124.83% from 2026-04-01 to 2026-07-28. The shorter-term EMA 12/26 left on 2026-07-08. All four exited in July, while the fund's own drawdown from 2026-06-22 to 2026-07-29 reached 69.27% and was still open at the end of the data.
The short-hold rules did not catch April the same way. The weekly 7% target made 49.38% in its best month and its best single trade was 21.48% from 2026-04-06 to 2026-04-08, a short hold inside a move that kept running. The momentum breakout made 100.74% from 2026-04-07 to 2026-04-28, its one outsized trade, with the next best at 37.76%. Its 138 round trips and 37% win rate produced a 1.2 profit factor, so one month carried a large share of its result.
The 2022 losses were also concentrated. Buy-and-hold had a 89.62% drawdown from 2021-12-27 to 2022-10-14, and the monthly cycle lost 43.64% in January 2022 and 48.67% in June 2022. The RSI(14) rule had its worst trade, -59.44%, in the 181 days from 2022-01-21 to 2022-07-21.
How SOXL behaved
| Measure | SOXL |
|---|---|
| Data in this test | 2021-01-04 to 2026-10-02 (1444 sessions) |
| Total return, buy and hold | 451.0% |
| Annualized volatility | 113.3% |
| Deepest drawdown | −90.5% (2021-12-27 to 2022-10-14) |
| Up days | 53.4% |
| Average daily range | 7.77% |
| Average overnight gap | 3.17% |
| Correlation to SPY | 0.80 |
| Correlation to QQQ | 0.87 |
| Correlation to TLT | 0.05 |
| Sessions above the 200-day average | 58.5% |
| Crossings of the 200-day average | 44 |
| Falls of 10% or more from a 20-day high | 102 |
Calendar years
| Year | Return |
|---|---|
| 2021 | 121.7% |
| 2022 | −85.6% |
| 2023 | 226.5% |
| 2024 | −12.3% |
| 2025 | 55.0% |
| 2026 (part) | 289.9% |
Biggest single days
| Best day | Move |
|---|---|
| 2025-04-09 | 56.0% |
| 2022-11-10 | 31.5% |
| 2026-07-30 | 25.1% |
| 2026-06-11 | 24.4% |
| 2025-05-12 | 21.7% |
| Worst day | Move |
|---|---|
| 2026-06-05 | −30.7% |
| 2025-04-03 | −29.8% |
| 2025-04-10 | −25.1% |
| 2025-04-04 | −23.5% |
| 2025-01-27 | −23.5% |
Average return by calendar month
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 12.8% | 4.6% | −4.4% | 9.2% | 33.7% | 12.4% | −1.8% | −8.4% | −2.1% | 3.6% | 24.8% | 2.6% |
Most and least correlated funds
| Most correlated | Least correlated | ||
|---|---|---|---|
| SOXX | 1.00 | SOXS | -1.00 |
| XLK | 0.90 | TECS | -0.90 |
| TECL | 0.90 | REW | -0.89 |
| ROM | 0.90 | QID | -0.87 |
| QQQM | 0.88 | SQQQ | -0.87 |
Leverage and decay against SOXX
| Year | SOXL | SOXX | 3× SOXX, reset daily |
|---|---|---|---|
| 2021 | 121.7% | 44.7% | 129.2% |
| 2022 | −85.6% | −35.1% | −84.8% |
| 2023 | 226.5% | 67.0% | 267.0% |
| 2024 | −12.3% | 13.0% | −0.4% |
| 2025 | 55.0% | 40.7% | 73.7% |
| 2026 | 289.9% | 95.9% | 330.2% |
The last column compounds 3× SOXX's daily return with no fees or financing: the return a perfect daily-reset fund would have had. It leaves out the fund's fees, its borrowing costs and the interest it earns on cash, which put a real fund below or above this line. A daily-reset fund does not aim for 3× the underlying's return over a year.
Over 2021-01-04 to 2026-10-02, SOXL returned 451.0% while SOXX returned 388.4% and a perfect daily-reset 3× version would have returned 854.1%. Its measured daily beta to SOXX was 2.97.
What leverage did to SOXL
SOXL's annualized volatility was 113.25% and its beta to SPY was 5.49, against 4.42 to QQQ. Its realized beta to SOXX over the window was 2.97 with an R-squared of 1, so it tracked its target closely day by day. The fund's average up day was 5.31% and its average down day was minus 5.29%, with 53.36% of days up. The average intraday range was 7.77% and the average overnight gap was 3.17%.
The leverage section of the page is the most useful measure here. SOXX returned 388.35% over the window. Three times that, held as a simple multiple, would be 1,165.05%. SOXL actually returned 450.98%. A fund rebalanced daily at exactly 3x, with no costs, would have returned 854.1%. The fund finished below both figures. The gap by year shows where it fell short. In 2024 the underlying gained 13% and SOXL lost 12.33%, a shortfall of 51.32 points against three times the underlying. In 2025 the underlying gained 40.68% and SOXL gained 54.98%, with a shortfall of 67.06 points. In 2022 the fund lost 85.64% against a simple multiple of minus 105.2%, so decay helped there by 19.56 points, since a fund cannot lose more than everything. In 2023 the fund gained 226.54% against 200.87% for the simple multiple.
In 2023 and 2026 the fund came out close to or above its simple multiple, and in 2024 and 2025 it fell well short of it.
Overnight returns made up 156.28% of the fund's return and intraday returns were minus 56.28%. In log terms the overnight return was 261.62% and the intraday return was minus 94.21%. Over this window the fund made all its gains between the close and the next open and gave some back during the session. A strategy that held only during the session would have lost money on this fund over the window.
The best day was 2025-04-09 at 55.98%, followed by 2022-11-10 at 31.49%, 2026-07-30 at 25.07% and 2026-06-11 at 24.43%. The worst was 2026-06-05 at minus 30.71%, then 2025-04-03 at minus 29.79% and 2025-04-10 at minus 25.07%. April 2025 holds both a 29.79% fall and a 55.98% rebound inside a week. The fund fell 10% or more from a 20-day high 102 times, covering 769 days.
The mean reversion numbers explain the strong RSI(2) result. RSI(2) fell under 10 on 140 sessions, with a median forward 5-day return of 1.7% against 1.22% for all days, and a forward 20-day return of 6.65% against 2.96%. RSI(14) fell under 30 on only 7 sessions, with forward returns of 11.92% over 5 days and 16.81% over 20. Seven sessions is a small sample. The lag-1 autocorrelation was minus 0.08, so the day after a large move leaned slightly the other way.
The fund spent 58.55% of sessions above its 200-day average and crossed it 44 times. That fits the 10 round trips the 200-day filter closed in 2023.
Seasonally, May averaged 33.67% and November 24.84%, with August at minus 8.44% and March at minus 4.43%. These come from five or six observations per month, so they describe this window and not a calendar rule. The best month for holding was April 2026 at 162.61% and the worst was July 2026 at minus 56.9%.
Drawdown recovery and the funds that moved with SOXL
The fund's three deepest drawdowns set the shape of the test. The first ran from 2021-12-27 to the 2022-10-14 trough for 89.62%, 291 days down, and took 1,230 days to recover, until 2026-02-25. The second, from 2026-02-25 to 2026-03-30, was 43.04% and recovered in 10 days. The third started on 2026-06-22, reached 69.27% on 2026-07-29 and was open at the end of the data. A fund that returned 450.98% in total spent 1,043 sessions in its longest drawdown. The daily history also records 102 separate falls of 10% or more from a 20-day high, spread over 769 days.
The volatility behind that is a 7.77% average intraday range and an average overnight gap of 3.17%. Up days were 53.36% of sessions, with an average gain of 5.31% against an average loss of 5.29% on down days. The lag-1 autocorrelation was -0.08, a small negative figure that points to mild reversal after large moves. That is consistent with the RSI(2) snapback doing well on this fund.
The oversold statistics make the same point from a different side. RSI(14) fell below 30 on only 7 sessions, and the median forward 5-day return after them was 11.92% against 1.22% for all days. The 20-day median was 16.81% against 2.96%. Seven observations are too few for a rule, which agrees with the RSI(14) template making only 14 round trips. The RSI(2) signal, with 140 observations, had a more modest edge.
On weekdays, Monday averaged 0.92% and Wednesday 0.68%, against 0.06% on Tuesday, 0.33% on Thursday and -0.07% on Friday. With average daily moves above 5%, those averages are within noise, and no weekday rule is supported.
The funds that moved with SOXL are mostly technology and semiconductor funds. SOXX had a correlation of 1, and XLK, TECL and ROM were at 0.90. The inverse funds SOXS at -1, TECS at -0.90 and REW at -0.89 sit at the other end. Among the same-category peers, TECL held at 38.16% and its best template, the monthly cycle, made 35.97%. ROM held at 30.19% and the weekly 7% target made 30.27%. FAS held at 18.34% and the dip buyer made 24.15%.
Build it from blocks (or type it in English), backtest it on 5.7 years of minute data in seconds, tweak any parameter, then paper trade it on live data. No card, no broker needed to start.
Frequently asked questions
What was the best strategy for SOXL?
Of the 12 templates tested on SOXL over 2021-01-04 to 2026-10-02, the strongest by CAGR was golden cross at 45.9% (max drawdown 69.3%), versus 33.3% for buy-and-hold. The best result in hindsight is not a forecast. Check drawdowns and trade counts before drawing conclusions.
Did any strategy beat buying and holding SOXL?
2 of 12 templates beat SOXL buy-and-hold (33.3% CAGR) on this window; 11 of 12 had a shallower maximum drawdown than holding (89.6%).
Which strategies beat buy-and-hold on SOXL?
Two of 12. Golden cross made 45.93% a year against 33.32% for holding, and RSI(2) snapback made 39.24%. Golden cross had only 3 closed round trips plus one open position, so the result depends on a few trades.
How deep was the SOXL drawdown?
Buy-and-hold fell 89.62% from 2021-12-27 to 2022-10-14 and did not recover until 2026-02-25, 1,230 days after the trough was passed. A second drop of 69.27% ran from 2026-06-22 to 2026-07-29 and had not recovered by 2026-10-02.
Did any template cut the SOXL drawdown a lot?
RSI(2) snapback had the shallowest at 58.51%, followed by EMA 12/26 trend at 65.38%. Eleven of the 12 templates were shallower than holding, and none got below 58%.
Why did RSI(2) snapback lose so much ground in 2026?
The fund gained 285.2% in 2026 and the template gained 3.2%, a gap of 282 points. It made 20 round trips that year and won 9.
What does 3x leverage do to SOXL over time?
SOXX returned 388.35% over the window and SOXL returned 450.98%. Three times the underlying return would have been 1,165.05%, and an ideal daily 3x fund would have returned 854.1%. The shortfall was largest in 2024 and 2025.
Do trading costs change the SOXL results?
They change the high-turnover templates. Weekly 7% target fell from 29.47% to 20.11% at 10 basis points per trade, which put it behind holding. Golden cross moved less than a tenth of a point.
Other leveraged etfs
Backtests are hypothetical, computed by DeployQuant's engine on minute-resolution consolidated US market data (2021-01-04 to 2026-10-02, $10,000 starting capital, no margin, no fees or slippage in the headline run; buy-and-hold puts 98% of the account in at the first open, as the templates do) and do not guarantee future results. Nothing on this page is investment advice. Live trading involves risk of loss.