SPDN trading strategies, backtested

Direxion Daily S&P 500 Bear 1X Shares: Direxion's -1x daily S&P 500 fund. Every DeployQuant template run on SPDN over 5.7 years of minute data, same engine, same window, sorted by return.

SPDN is a daily inverse fund on the S&P 500 with a leverage of negative 1. It rises when the index falls on a given day and falls when the index rises. From 2021-01-04 to 2026-10-02 the S&P 500 fund SPY gained 125.29%, and SPDN lost 46.16% on its own price series. Buy-and-hold in our test returned negative 9.68% a year, turned $10,000 into $5,575, and had a maximum drawdown of 46.2% that began on the first day of the window and never recovered.

Every one of the 12 DeployQuant templates finished with a negative CAGR, and every one beat holding. The templates did not find a way to profit from a falling index in a window where the index mostly rose. They reduced time in a losing position. The best result, RSI(2) snapback, was negative 1.45% a year. The worst, the weekly 7% target, was negative 9.04%, which is about where holding landed.

The test used $10,000 per run, minute-bar fills, no margin, and no fees in the headline numbers. A fund that loses money in a rising market gives a trading rule little to find, and the statistics on this page show how little.

The companion pages for the other inverse funds, such as SH and PSQ, show the same pattern with different index exposure.

Twelve ways to lose less on an inverse fund

The ranking runs from negative 1.45% to negative 9.04%. Time invested explains part of the ranking: the two templates that sat in the market for most of the window finished last, and they lost about as much as holding.

RSI(2) snapback was invested 46.5% of the time and made 169 round trips with a win rate of 53%. It returned negative 1.45% with a drawdown of 26.3%. The profit factor was 0.93, so gross losses were slightly larger than gross wins, and the average win of 1.18% against an average loss of 1.36% shows why. Its 2022 return was 22.4%, the best of any year, in a year when the fund itself gained 18.2%. In every other year it lost money, between 2.7% and 8.8%. The largest single loss was 8.61% on a trade from 2026-04-02 to 2026-04-22.

The dip buyer made 2 round trips and won both, at 8.46% and 8.02%, both in 2022. It returned negative 1.73% and was invested 27.2% of the time. A position opened on 2025-04-28 was down 23.01% when the data ended, and that position produced the 24.1% drawdown. This is the same pattern as on the equity funds, where the dip buyer wins its small target and then sits in a loser. Here the fund's drift makes the loser permanent.

The momentum breakout made 4 round trips and returned negative 2.21%. It was invested 9.0% of the time, the lowest exposure of any template. Its profit factor of 0.20 reflects one win of 3.32% against losses of 5.60%, 1.32%, and 8.84%. The 8.84% loss came on a two-day trade on 2025-04-07 to 2025-04-09, the days when the index rebounded sharply and SPDN had its worst single day at negative 9.38%.

Golden cross made 4 round trips and returned negative 2.26%. It held one trade for 304 days, from 2022-03-25 to 2023-01-23, for a gain of 10.75%. It then lost 15.89% over 90 days from 2025-04-04 to 2025-07-03, which is the single worst trade of the template. SMA 10/50 returned negative 2.61% with 17 round trips and a win rate of 24%, and EMA 12/26 returned negative 3.78% with 21 round trips and a win rate of 24%. For both, 2022 was the one clearly positive year, at 8.6% and 3.5%, and each lost money in each of the next four years.

The 200-day filter returned negative 4.50%. Its profit factor was 0.11, and it was invested 19.5% of the time with 24 round trips and 4 wins. SPDN was above its 200-day average on only 22.6% of sessions and crossed the line 48 times. That count is high for a fund that spent most of its time below the line, and it is the source of the 24 round trips: 13 of them came in 2022 and 9 in 2023, each one a short entry that was reversed. The average win was 0.85% and the average loss was 1.50%. The 200-day filter page shows the same rule across all 59 funds.

The trailing-stop trend rule had a drawdown of 44.6%. It lost 13.65% on a trade from 2025-05-02 to 2025-10-01 and 11.12% on a trade from 2022-12-07 to 2023-07-19. A 15% stop gives a position room to lose, and on a fund that drifts down this became a deep loss before the exit.

The monthly cycle was invested 95.2% of the time, returned negative 8.63%, and won 24 of 69 round trips. It had a 41.8% drawdown and an 8 trade losing streak. The weekly 7% target had the extreme version of the same shape: 140 round trips, 12 wins, a win rate of 9%, and a losing streak of 25. Its wins averaged 7.54% and its losses averaged 1.12%, which is the expected profile for a rule that cuts losers on Thursday and waits for a 7% move that arrives rarely on an inverse S&P fund. Its best trade of 10.13% ran from 2024-07-15 to 2024-08-05. It was ahead of holding in 2021 and 2022 by 6.1 and 1.4 points and slightly behind in the four years after that.

What each template made in the one winning year

SPDN rose 18.72% in 2022 and fell in every other calendar year, so the 2022 column is where the templates differ most. Holding made 18.2% in the template run. The RSI(2) snapback made 22.4%, the monthly cycle 21.9% and the weekly 7% target 19.6%. The dip buyer made 16.8% from its two trades, the golden cross 14.1% and RSI(14) mean reversion 12.6%. The slower trend rules made less: SMA 10/50 8.6%, trend plus trailing stop 5.1% and EMA 12/26 3.5%. Three templates beat the fund in the year it paid, and all three were the ones that traded often or stayed invested.

The other five years look different. The templates that held a position in 2023 gave back most of what they had made. The trend plus trailing stop made 5.1% in 2022 and then lost 20.8% in 2023. The 3-month momentum switch and the 200-day filter lost 14% and 13.6% in 2022 itself.

RSI(14) mean reversion shows how a single long hold can set a whole year. Its trade from 2021-01-26 to 2021-09-21 lost 13.45%, and the template's 2021 result was -17.9%. A second long loss, 12.06% from 2025-05-08 to 2025-11-18, sat inside a 2025 result of -11.3%. In the two years when the fund fell steadily, the rule bought after drops that continued.

The intraday and overnight split on SPDN is close to even. The log return was -27.3% overnight and -32.81% intraday, with 45.42% of the loss coming overnight and 54.58% during the session. A rule that buys the open and sells within a session had no edge here, because the session itself lost money. The average intraday range was 1.14% and the average overnight gap 0.44%, so the session range was wider than the overnight gap on a typical day.

How each strategy traded SPDN

Trade shapes, exposure and costs on a falling fund

The trading table separates the templates by how often they traded. At one end, the monthly cycle and weekly rule sat in the market for most of the window and gave back almost everything holding lost. At the other end, the 200-day filter, momentum breakout, and 3-month momentum template were invested 19.5%, 9.0%, and 17.7% of the time.

These low-exposure templates share a structure. Their entry signals fire on rallies in the fund, and rallies in SPDN mean falls in the index. The rules buy after SPDN has risen enough to look strong, and the fund then gives it back as the index recovers. The win rates of 13% for 3-month momentum, 17% for the 200-day filter, and 25% for momentum breakout and golden cross show that most of the entries were reversed. Profit factors of 0.11, 0.11, 0.20 and 0.46 follow from small wins and larger losses.

The 2022 trades are the exception, and they carry most of the positive results across the table. Nine of the 12 templates had a positive 2022. The window's single bear market provided a trend that the long-only rules could follow, and the trade lists show the timing: the golden cross entry on 2022-03-25, the dip buyer entries on 2022-03-30 and 2022-08-11, the SMA 10/50 gain of 7.87% from 2022-09-09 to 2022-11-03. After the index turned up at the end of 2022 and again in 2023, the same rules lost money in each following year.

The cost runs show a pattern opposite to the equity funds. The high-turnover templates took the largest hit. RSI(2) snapback went from negative 1.45% to negative 4.25% at 5 basis points and negative 6.98% at 10, with its drawdown rising from 26.3% to 38.3%. The weekly 7% target went from negative 9.04% to negative 10.69% and negative 14.66%, and its drawdown rose to 61.0%. The monthly cycle went from negative 8.63% to negative 10.78%.

The low-turnover templates barely moved. The dip buyer went from negative 1.73% to negative 1.77%. Golden cross went from negative 2.26% to negative 2.40%. Momentum breakout went from negative 2.21% to negative 2.34%.

RSI(2) snapback ranked first on the headline numbers. At 5 basis points it already trailed the dip buyer, momentum breakout, and golden cross. With 338 fills in the window, each fill pays the assumed cost, and the headline return of negative 1.45% is a figure before any cost.

SPDN trades $113,208,638 a day on average with a median minute volume of 4,782 shares. The sleeve in this test is $10,000, small against that volume. The cost runs apply flat offsets and say nothing about how wide the spread is when the index moves fast.

The cross-fund comparison on the SH page gives a second inverse S&P 500 fund over the same window, where the best template was also RSI(2) snapback.

How SPDN behaved

Calendar years

Biggest single days

Average return by calendar month

Most and least correlated funds

Leverage and decay against SPY

The last column compounds -1× SPY'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 -1× the underlying's return over a year.

Over 2021-01-04 to 2026-10-02, SPDN returned −46.2% while SPY returned 125.3% and a perfect daily-reset -1× version would have returned −62.0%. Its measured daily beta to SPY was -1.00.

Why an inverse fund lost money while it tracked correctly

SPDN did what it is designed to do. Its beta to SPY was negative 1.00, its correlation to SPY was negative 1.00, and the R-squared of daily returns against SPY was 1. The loss comes from the index. SPY returned 125.29% in the window, and a fund that gives the opposite daily return of a rising index loses value.

The leverage table shows how that loss compounds. SPY returned 125.29% over the window. A frictionless daily-reset −1× version of SPY would have returned negative 61.98%, and SPDN returned negative 46.16%, a smaller loss than that ideal. The fund was ahead of the ideal in every year except 2021, when it was slightly behind at negative 24.82% against negative 24.62%. In 2022 it was 18.72% against 15.32%, and in 2024 negative 12.91% against negative 21.17%. The ideal leaves out fees, borrowing costs and interest on cash, so the data does not identify the source of the difference.

By year, SPDN returned negative 24.82% in 2021, 18.72% in 2022, negative 15.05% in 2023, negative 12.91% in 2024, negative 11.01% in 2025, and negative 8.37% in 2026 to date. It had one winning calendar year in six. The index gained 30.46%, lost 18.16%, then gained 26.18%, 24.86%, 17.72%, and 13.77%. A trend rule that tries to catch the one down year must correctly time both the entry and the exit, because the index recovered in 2023 and the fund lost 15.05%.

Only 45.6% of days were up days for SPDN, with an average up day of 0.77% and an average down day of negative 0.77%. The averages are equal, so the loss comes from the frequency of down days. Annualized volatility was 16.4%. The average intraday range was 1.14% and the average overnight gap was 0.44%. Both the overnight and the intraday sessions lost money, with the overnight session accounting for 45.42% of the loss and the intraday session 54.58%.

The lag-1 autocorrelation was negative 0.02, so there was no day-to-day persistence for a short-term rule to use. On oversold readings, RSI(14) was below 30 on 70 sessions, with a median 5-day forward return of 0.19% and a median 20-day return of negative 0.32%, against baselines of negative 0.31% and negative 1.28%. RSI(2) was below 10 on 238 sessions, with medians of negative 0.25% and negative 1.30%, almost the same as the baselines. On a fund that drifts down, the baseline forward return is already negative, so a bounce after an oversold reading is only a smaller loss. That matches the RSI mean reversion template's result of negative 3.97% with a worse drawdown of 28.5%. Its worst trade, from 2021-01-26 to 2021-09-21, held 238 days and lost 13.45%.

The fund had 7 falls of 10% or more from a 20-day high, spread over 23 days, which is fewer than the equity funds. The dip buyer enters only after such a fall, and it completed only 2 round trips.

By calendar month, SPDN averaged 2.87% in September and 0.33% in February and December, and lost 3.48% in November, 2.55% in May, 2.49% in July, and 2.40% in October. These are 5 or 6 observations per month. The best days were 2025-04-04 at 5.99% and 2025-04-03 at 4.89%, and the worst were 2025-04-09 at negative 9.38% and 2022-11-10 at negative 5.40%. One week in April 2025 holds both the two best days and the worst day.

Other inverse funds in the test show a similar split. SQQQ returned negative 42.33% on buy-and-hold and its best template, golden cross, returned negative 7.41%. PSQ returned negative 13.76% and its best template, RSI(2) snapback, returned 0.13%. SOXS returned negative 48.27% and its best template, trend plus trailing stop, returned negative 4.93%. TBF, the inverse Treasury fund, was the exception, with buy-and-hold of 12.21%.

The test is one window of 5.74 years in which the S&P 500 more than doubled. An inverse fund is a hedge instrument, and the results show what happened to a $10,000 position with daily-decision rules over this window. They do not describe a hedging use with another position alongside, and they are not a forecast.

Other inverse 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.