DeployQuantTECS trading strategies, backtested
Direxion Daily Technology Bear 3X Shares: -3x daily technology sector. Every DeployQuant template run on TECS over 5.7 years of minute data, same engine, same window, sorted by return.
TECS is the Direxion Daily Technology Bear 3X fund, which aims for minus three times the daily return of a technology sector index. The fund's own price history from 2021-01-04 to 2026-10-02 shows a total return of -99.38%, a CAGR of -58.75% and annualized volatility of 75.18%. The maximum drawdown was 99.4%, from the peak on 2021-01-06 to the trough on 2026-10-02, with no recovery. The 12 templates all used one engine, one window and a $10,000 starting balance.
All 12 templates had a negative CAGR. Trend plus trailing stop lost the least at -11.04% a year and the monthly cycle lost the most at -55.62%. Buy-and-hold in the engine returned -46.67% a year, ended at $270.72 with a drawdown of 97.52% and a Sharpe ratio of -0.72. Eleven templates beat that CAGR and 11 had a shallower drawdown. The monthly cycle was the only one to do worse than holding.
The engine's holding line is close to the fund's own price series in the early years and below it later. Holding in the engine returned -68%, 42.5%, -71.4%, -42.3%, -46.3% and -36.9% for 2021 to 2026. The fund's own calendar years were -69.35%, 45.25%, -74.54%, -49.74%, -62.41% and -71.1%. Over the whole window the engine's holding return was -97.29% against -99.38% for the fund. The facts do not say what causes the difference. Template results below are compared with the engine's holding line, and the behaviour tables use the fund's own prices.
| Strategy | CAGR | max DD | Sharpe | trades | win rate | final value |
|---|---|---|---|---|---|---|
| SMA-200 Trend + 15% Trailing Stop | −11.0% | −54.1% | -0.13 | 29 | 28% | $5,107 |
| 200-Day SMA Regime Filter | −13.9% | −62.4% | -0.15 | 17 | 18% | $4,247 |
| 20-Day Momentum + Trailing Stop | −18.3% | −72.4% | -0.32 | 83 | 29% | $3,127 |
| 3-Month Momentum Switch | −20.7% | −77.9% | -0.28 | 18 | 17% | $2,638 |
| Golden Cross (SMA 50/200) | −21.0% | −80.0% | -0.44 | 3 | 0% | $2,578 |
| RSI(2) Dip Snapback | −26.6% | −90.6% | -0.46 | 165 | 51% (+1 open) | $1,694 |
| SMA 10/50 Trend | −27.1% | −87.2% | -0.38 | 21 | 24% | $1,635 |
| RSI(14) Mean Reversion | −29.1% | −89.5% | -0.41 | 18 | 50% (+1 open) | $1,384 |
| EMA 12/26 Trend | −29.4% | −87.8% | -0.45 | 21 | 14% | $1,355 |
| Weekly Entry + 7% Target | −41.6% | −96.5% | -0.69 | 264 | 40% | $456 |
| Drawdown Dip Buyer + 8% Target | −45.3% | −97.4% | -0.67 | 2 | 100% (+1 open) | $313 |
| First-to-Last Day of Month | −55.6% | −99.1% | -0.78 | 69 | 32% (+1 open) | $94 |
| Buy & hold TECS | −46.7% | −97.5% | -0.72 | – | – | $271 |
Trend plus trailing stop, then the other trend rules
The best template made 29 round trips and won 8 of them, a 28% win rate. The average win was 14.51% and the average loss was 7.79%. The profit factor was 0.6. The rule was invested 11% of the time and held for 8.2 days on average. The CAGR was -11.04%, the Sharpe ratio -0.13 and the maximum drawdown 54.1%.
Only one calendar year was positive. In 2025 the rule made 27.9%, on 6 round trips of which 3 won. In 2022, the year TECS itself rose 45.25%, the rule lost 43.6% on 20 round trips and 5 wins. It returned 0% in 2021 and 2024, lost 7.1% in 2023 and lost 23.8% in 2026 so far. Against the engine's holding line it was ahead in 2021, 2023, 2024, 2025 and 2026 and behind in 2022, when holding returned 42.5%.
The best trade, 47.01%, ran 18 days from 2025-03-20 to 2025-04-07. Next were 16.63% over 27 days from 2022-05-27 and 16% over 20 days from 2022-09-14. The worst trade was -18.51% over 7 days from 2026-04-01. The 15% trail did its job on individual losses, and the median trade still lost 5.21%. The drawdown of 54.1% began on 2022-03-14, reached its trough on 2025-03-25 and had not recovered by the end. At 10 basis points per trade the CAGR was -11.93%.
The 200-day regime filter made 17 round trips and won 3, but the wins were tiny: an average win of 0.66% against an average loss of 6.14%, a profit factor of 0.02. Its CAGR was -13.86% and its drawdown 62.37%. The best trade made 1.37% over 47 days from 2022-09-14 and the worst lost 10.72%. TECS closed above its 200-day average on 13.17% of sessions and crossed it 34 times, which is how a rule that needs a sustained uptrend ended up in and out of the fund so often.
SMA 10/50 made 21 round trips and won 5. Its two best trades were 28.68% over 55 days from 2022-09-09 and 26.76% over 50 days from 2022-04-21, both in 2022, and it made 27.6% for that year. Its average win was 12.62% and its average loss 13.85%. The CAGR was -27.05% and the drawdown 87.19%. Its worst trade, -23.68%, lasted 68 days from 2026-02-06. EMA 12/26 made 21 round trips and won 3, with a profit factor of 0.12 and a longest losing streak of 14. Its CAGR was -29.4% and it lost money in every calendar year, including -2.1% in 2022.
Golden cross made 3 round trips and all 3 lost. The first, -15.5%, lasted 125 days from 2022-05-10. The second, -31.29%, lasted 71 days from 2022-10-12. The third, -57.33%, lasted 58 days from 2025-04-08. The CAGR was -21.03% and the drawdown 80.03%. Its exposure was 12.1% and its yearly returns were 0% except -41.2% in 2022 and -56.2% in 2025. The third trade began on 2025-04-08, the session before the 39.97% fall on 2025-04-09.
Mean reversion and the high-turnover rules
RSI(14) mean reversion made 18 round trips and won 9, a 50% win rate. The average win was 16.8% and the average loss was 28%, and the profit factor was 0.48. It held for 74 days on average and was invested 65.7% of the time. The CAGR was -29.14% and the drawdown 89.51%. It made 35.4% over 110 days from 2025-12-10 and lost 61.72% over 203 days from 2025-04-30, then lost 42.49% over 105 days from 2026-04-16. It made 34.4% in 2022 and 26.3% in 2024, and lost in the other four years. A position opened on 2026-08-18 was open at the end, down 19.55%.
The RSI(2) snapback made 165 round trips with 84 wins, a 51% win rate. The average win was 5.5% and the average loss was 7.07%, giving a profit factor of 0.83. The CAGR was -26.6% and the drawdown 90.59%. In 2022 it made 74.2%, the best calendar year of any template on this fund, and in 2024 it lost 56.4% and in 2026 so far 50.3%. Its worst trade, -42.51%, ran 27 days from 2026-04-02, and its best was 22.21% from 2025-04-02 to 2025-04-04. The median trade made 0.23%. A position opened on 2026-10-02 was open at the end, up 0.66%.
Momentum breakout made 83 round trips and won 24, a 29% win rate. The average win was 10.47% and the average loss was 5.71%, and the profit factor was 0.64. It was invested 18.1% of the time with a 4.6-day average hold. The CAGR was -18.33% and the drawdown 72.36%. All 9 of its round trips that ended in 2021 lost, and in 2022 it won 14 of 32 and made 12.4%. Its best trade was 33.58% from 2025-04-03 to 2025-04-07. The 3-month momentum switch made 18 round trips and won 3, with a profit factor of 0.09, a longest losing streak of 12, a CAGR of -20.71% and a drawdown of 77.86%. Its worst trade was -27.95% over 52 days from 2022-10-07.
The weekly 7% target made 264 round trips and won 105, a 40% win rate. The average win was 7.4% and the average loss 6.33%, and the profit factor was 0.78. The CAGR was -41.59%, with a drawdown of 96.49% and a final value of $456.42. It was invested 56.2% of the time. The target fills often on a fund with an average intraday range of 5.28%, but the win rate sits below half, and the Thursday exit realized the rest at small losses. Its worst trade was -36.08% from 2025-04-07 to 2025-04-10. The yearly returns were -35.7%, 17.7%, -48.6%, -59.3%, -58.2% and -35.3%. Its median trade lost 1.78%.
The dip buyer completed 2 round trips, both winners at the 8% target, in February and March 2021. It then bought on 2021-03-05 at an adjusted price of $7,444.91 and held that position to the end, down 99.34%. The CAGR was -45.31%, a little better than holding's -46.67%, because the two early wins came before the decline. It beat holding in 2021 and stayed within 0.1 points of it in the later years.
The monthly cycle made 69 round trips and won 22, a 32% win rate. The average win was 15.27% and the average loss 14.66%, a profit factor of 0.5. The CAGR was -55.62%, the drawdown 99.14% and the final value $94.23. It was invested 95.2% of the time. Its best months were 33.89% in March 2025 and 31.91% in September 2022, and its worst were -41.33% in April 2026 and -40.39% in May 2026. It made 62.2% in 2022. A position opened on 2026-10-01 was open at the end, down 4.41%.
How each strategy traded TECS
| Strategy | Time in market | Avg hold (days) | Best trade | Worst trade | Profit factor | With 10 bps slippage |
|---|---|---|---|---|---|---|
| trend + trailing stop | 11.0% | 8 | 47.0% | −18.5% | 0.60 | −11.9% |
| 200-day regime filter | 11.4% | 14 | 1.4% | −10.7% | 0.02 | −14.4% |
| momentum breakout | 18.1% | 5 | 33.6% | −15.1% | 0.64 | −20.6% |
| 3-month momentum | 19.9% | 23 | 8.7% | −27.9% | 0.09 | −21.2% |
| golden cross | 12.1% | 85 | −15.5% | −57.3% | – | −21.1% |
| RSI(2) snapback | 48.3% | 6 | 22.2% | −42.5% | 0.83 | −30.7% |
| SMA 10/50 trend | 28.3% | 28 | 28.7% | −23.7% | 0.38 | −27.6% |
| RSI mean reversion | 65.7% | 74 | 35.4% | −61.7% | 0.48 | −29.6% |
| EMA 12/26 trend | 26.5% | 26 | 16.6% | −18.6% | 0.12 | −29.9% |
| weekly 7% target | 56.2% | 3 | 14.6% | −36.1% | 0.78 | −49.5% |
| dip buyer | 98.5% | 15 | 8.0% | 8.0% | – | −45.6% |
| monthly cycle | 95.2% | 28 | 33.9% | −41.3% | 0.50 | −56.7% |
Exposure, profit factors and costs
Exposure ranged from 11% for trend plus trailing stop and 11.4% for the 200-day filter to 95.2% for the monthly cycle and 98.5% for the dip buyer. The two lowest-exposure templates held first and second place by CAGR. Average holds ran from 3.4 days for the weekly target and 4.6 for momentum breakout to 74 days for RSI(14) mean reversion and 84.7 for golden cross. Profit factors were below 1 for every template with losses: 0.83 for the RSI(2) snapback, 0.78 for the weekly target, 0.64 for momentum breakout, 0.6 for trend plus trailing stop, 0.5 for the monthly cycle, 0.48 for RSI(14) mean reversion, 0.38 for SMA 10/50, 0.12 for EMA 12/26, 0.09 for the 3-month switch and 0.02 for the 200-day filter.
Costs separated the high-turnover rules from the rest. The weekly target went from -41.59% to -47.96% at 5 basis points and -49.53% at 10, a larger move than any other template made. The RSI(2) snapback went from -26.6% to -28.66% and -30.66%. Momentum breakout went from -18.33% to -19.48% and -20.63%. The rules with few trades barely moved: golden cross went from -21.03% to -21.09% and -21.14%, and trend plus trailing stop from -11.04% to -11.49% and -11.93%. A cost per fill compounds on a position that is already losing.
TECS traded an average of $52,984,363 a day and the median minute volume was 70 shares. The backtest fills on minute bars at no cost in the headline run, and a live fill on a minute with 70 shares trading could differ from the bar's price.
Best and worst months, streaks and drawdown length
Every template on TECS carries a worst month, and the dates cluster. The best month for trend plus trailing stop was June 2022 at 29.41% and the worst was July 2022 at -29.24%, one month apart. The 200-day filter had almost the same pair, 29.44% in June 2022 and -30.27% in July 2022. Holding itself made 40.34% in September 2022 and lost 32.19% in July 2022. A rule that caught the June 2022 rally and stayed in gave much of it back the next month, which fits a fund whose average intraday range is 5.28%.
April 2026 was the worst month for four templates: the RSI(2) snapback at -41.66%, the monthly cycle at -40.48%, SMA 10/50 at -27.73% and EMA 12/26 at -18.73%. The RSI(14) rule had its worst month in May 2026 at -42.19% and its best in July 2026 at 28.59%. September 2022 was the best month for momentum breakout, SMA 10/50, EMA 12/26 and the dip buyer, with SMA 10/50 reaching 42.96%. March 2025 was the best month for the 3-month switch at 25.37% and for the monthly cycle at 33.2%.
Losing streaks show how the rules lose. The golden cross had 3 losing trades in a row and no wins. EMA 12/26 lost 14 in a row and the 3-month switch 12. Momentum breakout lost 10 in a row and its longest winning run was 4. The RSI(2) snapback had the longest winning streak, 8, and a losing streak of 6, which fits its 51% win rate. The RSI(14) rule had streaks of only 2 and 2 across 18 trades, so its result came from the size of the losses.
Drawdown duration separates the templates as much as depth. The monthly cycle fell from 2021-01-06 and was still below that peak on 2026-10-02, 2,095 days later. The dip buyer's 97.43% drawdown ran 2,034 days to the same date. Trend plus trailing stop reached its trough 1,107 days after its March 2022 peak and the 200-day filter 1,479 days after the same peak. The only drawdowns that recovered were short early ones, such as the 11.11% fall for trend plus trailing stop in March 2022 that took 5 days to recover.
Yearly gaps to holding show that the exit rules paid for themselves in the declining years. In 2023 trend plus trailing stop was ahead of the engine's holding line by 64.3 points and the golden cross by 71.4. In 2022 every trend rule was behind: -86.1 points for trend plus trailing stop, -92.5 for the 3-month switch and -91 for the 200-day filter. The weekly target, the monthly cycle and the RSI(2) snapback were behind in 2024 and 2026, years when the fund kept falling and these rules kept re-entering.
How TECS behaved
| Measure | TECS |
|---|---|
| Data in this test | 2021-01-04 to 2026-10-02 (1444 sessions) |
| Total return, buy and hold | −99.4% |
| Annualized volatility | 75.2% |
| Deepest drawdown | −99.4% (2021-01-06 to 2026-10-02) |
| Up days | 44.6% |
| Average daily range | 5.28% |
| Average overnight gap | 2.08% |
| Correlation to SPY | -0.91 |
| Correlation to QQQ | -0.97 |
| Correlation to TLT | -0.07 |
| Sessions above the 200-day average | 13.2% |
| Crossings of the 200-day average | 34 |
| Falls of 10% or more from a 20-day high | 92 |
Calendar years
| Year | Return |
|---|---|
| 2021 | −69.3% |
| 2022 | 45.3% |
| 2023 | −74.5% |
| 2024 | −49.7% |
| 2025 | −62.4% |
| 2026 (part) | −71.1% |
Biggest single days
| Best day | Move |
|---|---|
| 2025-04-03 | 20.2% |
| 2026-06-05 | 19.9% |
| 2025-04-04 | 19.7% |
| 2022-09-13 | 16.0% |
| 2025-01-27 | 14.8% |
| Worst day | Move |
|---|---|
| 2025-04-09 | −40.0% |
| 2022-11-10 | −24.6% |
| 2026-07-30 | −16.2% |
| 2022-11-30 | −15.0% |
| 2026-08-04 | −14.8% |
Average return by calendar month
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|
| −2.8% | 0.8% | −2.9% | −5.0% | −19.3% | −10.3% | −5.2% | −2.1% | 6.6% | −12.2% | −13.6% | 0.3% |
Most and least correlated funds
| Most correlated | Least correlated | ||
|---|---|---|---|
| REW | 0.98 | TECL | -1.00 |
| QID | 0.97 | XLK | -1.00 |
| SQQQ | 0.97 | ROM | -0.99 |
| PSQ | 0.97 | TQQQ | -0.97 |
| SDS | 0.91 | QLD | -0.97 |
Leverage and decay against XLK
| Year | TECS | XLK | -3× XLK, reset daily |
|---|---|---|---|
| 2021 | −69.3% | 37.0% | −68.9% |
| 2022 | 45.3% | −27.7% | 38.6% |
| 2023 | −74.5% | 56.0% | −78.7% |
| 2024 | −49.7% | 21.6% | −58.6% |
| 2025 | −62.4% | 24.6% | −67.9% |
| 2026 | −71.1% | 39.3% | −74.1% |
The last column compounds -3× XLK'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, TECS returned −99.4% while XLK returned 226.1% and a perfect daily-reset -3× version would have returned −99.7%. Its measured daily beta to XLK was -2.99.
How TECS moves against XLK
TECS tracks XLK with a daily multiple of -3. Measured against XLK over the full window, its realized beta was -2.99 with an R-squared of 1. XLK returned 226.13% and TECS returned -99.38%, against -99.68% for a frictionless fund that resets to -3 times XLK every day. The yearly comparison in the table above shows the fund close to the ideal each year. In 2021 the fund returned -69.35% and the ideal was -68.92%. In 2022 XLK fell 27.73% and TECS rose 45.25%, against an ideal of 38.65%. In 2023 the fund returned -74.54% against -78.71%, in 2024 -49.74% against -58.63%, in 2025 -62.41% against -67.88%, and in 2026 so far -71.1% against -74.11%. The fund finished above the ideal in five of the six years. The caption under the table says fees, borrowing costs and interest on cash put a real fund below or above the ideal line.
XLK rose in every year except 2022, and TECS lost money in every year except 2022. The three-times multiple applies to each day, and the compounding of volatile days is what separates the yearly result from a simple tripling. The 5.28% average intraday range, the 75.18% annualized volatility and the 92 episodes of falls of 10% or more from a 20-day high, covering 952 days, are the inputs.
Beta to SPY was -4.17 and to QQQ -3.24, and beta to TLT was -0.33. The correlation to QQQ was -0.97 and to SPY -0.91, and the correlation to TLT was -0.07. The most correlated funds were REW at 0.98, QID, SQQQ and PSQ at 0.97 and SDS at 0.91. The least correlated were TECL and XLK at -1, ROM at -0.99, and TQQQ and QLD at -0.97.
Daily behaviour, seasonality and RSI
TECS rose on 44.56% of sessions. The average up day was 3.68% and the average down day was -3.43%. The average intraday range was 5.28% and the average overnight gap 2.08%. The intraday log return was -329.71% and the overnight log return was -172.65%, so about two thirds of the decay, 65.63% against 34.37%, came during the trading day. The lag-1 autocorrelation was -0.03.
The worst day was 2025-04-09 at -39.97%, followed by 2022-11-10 at -24.55%, 2026-07-30 at -16.2%, 2022-11-30 at -15.01% and 2026-08-04 at -14.8%. The best days were 2025-04-03 at 20.18%, 2026-06-05 at 19.93%, 2025-04-04 at 19.74%, 2022-09-13 at 16.03% and 2025-01-27 at 14.82%. The two best sessions of April 2025 came right before the worst one. A fund that gains 20% on two consecutive days and loses 39.97% a few sessions later rewards a rule that exits in time and hurts one that confirms the move first. Golden cross, entered on 2025-04-08 and held for 58 days, lost 57.33% in that stretch.
By calendar month, May averaged -19.33%, November -13.59%, October -12.19% and June -10.27%. September averaged 6.63%, February 0.76% and December 0.33%. Each month has five or six observations. These describe this window and give no reason to expect the same pattern again.
The fund closed above its 200-day average on 13.17% of sessions. RSI(14) was below 30 on 130 sessions and above 70 on 16. After an RSI(14) reading below 30 the median forward 5-day return was -1.02% against a baseline of -1.76%, and the 20-day figure was -7.95% against -6.49%. After RSI(2) fell below 10, on 255 sessions, the median 5-day return was -1.77% and the 20-day return -7.43%. A low RSI reading slightly improved the 5-day outcome and made the 20-day outcome worse than the baseline. That fits the RSI(14) template's long holds ending in large losses.
Among other inverse funds, the best template differs by fund. On SQQQ golden cross led with -7.41% against -42.33% for holding. On QID the RSI(2) snapback returned -1.85% against -29.46%, and on PSQ it returned 0.13% against -13.76%. On SDS golden cross returned -2.33% against -21.57%, on SOXS trend plus trailing stop returned -4.93% against -48.27%, and on REW golden cross returned -6.71% against -36.12%. EEV had the one clearly positive result, RSI(14) mean reversion at 5% against -16.07%, and TBF, which held at 12.21%, was led by the monthly cycle at 11.93%. TECS's own best, trend plus trailing stop at -11.04%, matches SOXS's winner. The test is one window with daily decisions and minute-bar fills, and prices in trade lists are adjusted for splits, so entry prices such as $7,444.91 are adjusted figures.
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 TECS?
Of the 12 templates tested on TECS over 2021-01-04 to 2026-10-02, the strongest by CAGR was trend + trailing stop at −11.0% (max drawdown 54.1%), versus −46.7% 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 TECS?
11 of 12 templates beat TECS buy-and-hold (−46.7% CAGR) on this window; 11 of 12 had a shallower maximum drawdown than holding (97.5%).
What was the best strategy for TECS?
Trend plus trailing stop, at -11.04% a year against -46.67% for holding, with a maximum drawdown of 54.1%. It made 29 round trips and won 8. It was positive only in 2025, when it made 27.9%, and every template had a negative CAGR.
Did any strategy beat buying and holding TECS?
Eleven of the 12 templates beat holding on CAGR and 11 had a shallower drawdown. The monthly cycle was the exception at -55.62% against -46.67%. All 12 lost money, so beating holding means losing less.
Why did every strategy lose money on TECS?
TECS is a -3x daily fund and its own total return was -99.38% from 2021-01-04 to 2026-10-02 while XLK returned 226.13%. It rose only in 2022, by 45.25%, and fell in the other five calendar years. A rule that is invested loses in those years, and a rule that waits for confirmation enters after much of a rise has passed.
Does the weekly 7% target work on TECS?
It returned -41.59% a year from 264 round trips, with a 40% win rate and a 96.49% drawdown. The average win was 7.4% and the average loss was 6.33%. Costs hurt it more than other templates: the CAGR was -49.53% at 10 basis points.
What was the worst day for TECS?
2025-04-09, when it fell 39.97%. The two best days of April 2025 were 20.18% on the 3rd and 19.74% on the 4th. Golden cross entered on 2025-04-08, held 58 days and lost 57.33%.
Is RSI a useful signal for TECS?
Not clearly. After RSI(14) fell below 30 the median 20-day return was -7.95% against a baseline of -6.49%. RSI(14) mean reversion returned -29.14% and the RSI(2) snapback returned -26.6%, though the RSI(2) rule made 74.2% in 2022.
Does holding TECS in the engine match the fund's own return?
Close in the early years and not later. Holding in the engine returned -97.29% over the window against -99.38% for the fund, and the yearly figures differ most in 2024 to 2026. The facts do not explain the gap.
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.