Tracks 3x the daily return of a US semiconductor index. Extremely volatile.
Prices as of 2026-10-23. Source: Data via FinanceDataReader · Policy rate via FRED (Federal Reserve Bank of St. Louis).
SOXL is Direxion's 3x semiconductor ETF, launched in March 2010. It targets three times the daily return of an index of roughly thirty large US-listed chip companies. Designers, equipment makers and memory producers such as Nvidia, Broadcom, AMD, Qualcomm and Micron are all in it, but the largest few names carry so much weight that a handful of stocks drive most of the move.
Semiconductors lead the business cycle, so declines run deep when the inventory cycle turns. In the 2018 trade war, the March 2020 crash, the 2022 rate-hiking cycle and the April 2025 tariff shock, the unleveraged index fell 25–45% and SOXL lost 65–90%. A 3x fund can take years to climb back to a previous high after a drop like that.
For split buying, SOXL is the reference ticker on this site. Its daily range is wide enough that a 1–2% take-profit target is often reached within days. The same width means tranches fill in clusters, so the number of tranches and their weights change results a lot. Remember too that its own history since 2010 covers only the best stretch semiconductors have had. For a test that includes the dot-com crash and 2008, see the run from 1996.
Using real closing prices only (adjusted for splits and dividends) from 2010-03-11, SOXL's worst peak-to-trough fall was -90.5%, from the 2021-12-27 high to the 2022-10-14 low over 202 trading days. It took another 873 trading days (about 3.5 years) to get back above that high, on 2026-04-10. The last close is 72.2% below the all-time high.
A fall of 90.5% needs a 954% gain to get back to even. Any strategy on SOXL has to be judged on whether you could have held it through a stretch like this. Why drawdown comes before return.
The standard deviation of daily returns is 5.79% over the whole history
and 8.61% over the last year. By the rough rule
days ≈ (target ÷ daily volatility)², a 1.5% take-profit target needs about
less than one trading day at the recent pace, so most tranches close at the target well before the time cut.
Where the rule comes from.
Three preset strategies run over 2016-01-04 to 2026-10-23 with a $10,000 seed and 0.1% fees, alongside simply holding the same instrument.
| Strategy | Return | Max drawdown | CAGR | Trades | Win rate |
|---|---|---|---|---|---|
| A2 Standard, base rules (no defence signal) | +3010.93% | -42.22% | +37.55% | 1222 | 81.6% |
| A1 Defensive, base rules | +902.99% | -33.36% | +23.84% | 1225 | 86.1% |
| A3 Aggressive, base rules (no defence signal) | +3603.08% | -52.49% | +39.79% | 1178 | 82.2% |
| Buy and hold | +4735.00% | -90.50% | +43.29% | - | - |
Read the drawdown column before the return column. Split buying leaves much of the seed in cash, which is the main reason its drawdown is smaller — treat it as lower exposure rather than protection.
Test SOXL with your own settings →The standard (diamond) preset, one calendar year at a time. A strategy worth using should have good years and bad years mixed — if one year made everything, that was the year, not the rules.
| Year | Strategy | Strategy MDD | Buy and hold | B&H MDD |
|---|---|---|---|---|
| 2026 | +2.66% | -34.57% | +76.36% | -72.67% |
| 2025 | +25.82% | -41.72% | +51.45% | -76.35% |
| 2024 | +87.70% | -13.02% | -2.80% | -62.13% |
| 2023 | +37.74% | -17.33% | +234.51% | -49.05% |
| 2022 | +64.83% | -19.60% | -86.27% | -90.04% |
| 2021 | +57.79% | -14.24% | +121.04% | -40.86% |
| 2020 | +48.14% | -21.34% | +59.31% | -80.21% |
| 2019 | +11.55% | -31.82% | +220.67% | -47.16% |
| 2018 | -15.84% | -35.23% | -44.12% | -66.48% |
| 2017 | +24.81% | -9.56% | +140.63% | -28.90% |
Across these 10 years the strategy beat buy and hold in 3 and trailed it in 7. It lost money in 1 year; buy and hold lost in 3. In SOXL's worst year here, 2022, holding returned -86.3% while the strategy returned +64.8%. In its best year, 2023, holding returned +234.5% and the strategy +37.7%. Giving up much of the best years is the price of the smaller drawdowns.
SOXL listed on 2010-03-11, and every number on this page uses real prices only. In the backtest tool you can go back to 1994, but the years before listing are index-based: SOXL listed on 2010-03-11. Prices from 1994-05-04 to the day before listing are converted from the 1x index, not real trades: the daily return of PHLX Semiconductor Index (^SOX); from July 2001 the 1x semiconductor ETF SOXX (named IGW and tracking the S&P GSTI Semiconductor Index until October 2010) multiplied by 3, minus the fund fee (0.75%/yr), borrowing cost (2x the 13-week T-bill rate) and an extra 2.3%/yr tracking cost. Treat that range as an estimate. Even close-only strategies such as A and D can shift by 1–2 points a year over this range depending on the conversion assumptions (dividends, costs, index differences). See how the index-based history is built and checked.
Only for a single day. Over longer stretches volatility drag pulls the result below 3x in choppy markets and above it in a sustained trend. The arithmetic is here.
Whole shares only. If a tranche is allocated less than one share's price, it buys nothing. Raise the seed or lower the number of tranches.
The Today's Orders page computes the next session's buy limit and each tranche's sell target from your start date. It is a calculation, not advice, and no orders are placed.