Golden Cross (SMA 50/200) on QID

ProShares UltraShort QQQ: -2x daily Nasdaq-100. Backtest 2021-01-04 to 2026-10-02, $10,000 starting capital, computed by the same engine that runs live DeployQuant strategies.

Golden Cross (SMA 50/200) buys when the 50-day simple average closes above the 200-day average and sells when it closes below. Applied to QID, the ProShares UltraShort QQQ fund, it turned $10,000 into $6,997.75 between 2021-01-04 and 2026-10-02. That is a CAGR of -6.03%, a total return of -30.02% and a maximum drawdown of 48.41%. The Sharpe ratio was -0.11.

QID itself lost 29.46% a year on a buy-and-hold basis, ending at $1,348 for a total return of -86.52%. The golden cross rules came out ahead of holding by 23.43% a year and $5,649.75 of end equity, and the drawdown was shallower than the 87.33% of buy and hold. The rules beat holding in five of the six calendar years. The one year they lost to holding was 2022, when QID rose.

The record is thin. There were 2 closed round trips and both lost money, so the win rate was 0%. The average loss was 16.01%. The account was invested on 18.4% of trading days and sat in cash for the rest. Because both trades were losers, the result is more a story of the days the template was out of the fund than of anything the entries did.

QID is a -2x daily fund on the Nasdaq-100. It fell for most of the window because the Nasdaq-100 rose. Over the same period the underlying index fund returned 151.07% and QID returned -89.27%. A moving-average cross on a fund that decays almost constantly will rarely fire, and when it does it tends to fire late. The test produced two trades and one year in which QID rose, and the fund's own statistics explain why a 50/200 cross is a poor fit. It is one fund over one window of 5.74 years.

Year by year

2022 was the only year the cross was invested

The year table has zeros in 2021, 2024 and 2026. In those years the 50-day average never rose above the 200-day average, so the template stayed in cash and the account value did not change. Buy and hold lost 45.2% in 2021, 31.8% in 2024 and 29.5% in 2026, which gave the template its largest yearly edges: 45.2, 31.8 and 29.5 points.

The two years with a position were 2022 and 2025, and the 2023 loss was the tail of the 2022 trade. The template gained 24.1% in 2022 while buy and hold gained 63.3%. That is the one year the template trailed by a wide margin, 39.2 points. It entered on 2022-03-11 at an adjusted price of $85.71. The fund's best months of the year, April at +28.96% and September at +22.58%, were partly captured, but the template was still holding during the declines of July (-21.97%), October (-9.41%) and November (-12.6%).

The 2023 figure is -23.4% against -55.6% for buy and hold. The template exited on 2023-02-06 at $81.39, after January's -18.28% and part of February's -6.23%. After that the account was flat for the rest of 2023. The long hold from March 2022 into February 2023 shows the usual lag of a moving-average cross: the fund's rally ended in 2022 and the cross did not turn down until February 2023. The trade closed at -5.04%, which means the gains of 2022 were mostly given back before the exit.

In 2025 the template lost 26.4% against 31.6% for buy and hold. The entry came on 2025-04-15 at $34.59, during the sell-off and rebound of April 2025, when QID had its best day of the window on 2025-04-04 at +12.12%. The exit on 2025-06-11 was at $25.26, a loss of 26.97% in 57 days. That was the worst trade. The monthly figures for the hold were -7.33% in April, -15.75% in May and -5.73% in June.

The fund's calendar returns tell the same story: -46.44% in 2021, +66.4% in 2022, -57.25% in 2023, -34.03% in 2024, -34.98% in 2025 and -34.35% in 2026 to date. QID rose in one year of six. A rule that steps aside when the trend is down will be out of the fund for most of them, and the 5 years the template beat buy and hold are the 5 years QID fell.

Month by month

Fifty-eight months of zeros

The month-by-month table is mostly zeros. The template held a position in 10 months of 2022 (March to December), January and February of 2023, and April to June of 2025. That is 15 months out of the whole test, a count consistent with the 18.4% exposure figure.

The best month was April 2022 at +28.96% and the worst was July 2022 at -21.97%. Buy and hold's best month was the same, +28.51%, and its worst was April 2026 at -22.08%. In 2022 the template's monthly path was -16.01% in March, +28.96% in April, -1.03% in May, +16.41% in June, -21.97% in July, +9.48% in August, +22.58% in September, -9.41% in October, -12.6% in November and +19.92% in December. Six of those ten months were positive. The sum of the gains was still offset by the large losing months, and the entry price of $85.71 against the exit price of $81.39 shows that the whole year produced a small loss on the trade itself.

The point to take from the monthly table is how clustered the exposure is. Every month the template held fell inside one of three windows, and between them the account sat in cash for long stretches: all of 2021, March to December 2023, all of 2024, and everything after June 2025. Over those stretches the fund fell, which is why the template's result is better than holding.

Across the fund's own calendar, May averaged -9.75% and November averaged -8.36%, with 6 and 5 observations. September averaged +4.33%. Those are six data points each, and they are consistent with nothing more than the fund's decay. The template does not use the calendar.

Every trade

golden cross on QID made 2 closed round trips, an average hold of 195 days, an average loser of −16.01%, a longest losing streak of 2. It held a position at the close on 18.4% of trading days.

Prices are adjusted for splits and dividends, so they sit below the quotes printed at the time. An open position is marked at the last close.

The two trades

The first trade ran from 2022-03-11 to 2023-02-06, 332 days, with an entry at $85.71 and an exit at $81.39. The return was -5.04%, the best trade of the two. The second ran from 2025-04-15 to 2025-06-11, 57 days, from $34.59 to $25.26, a return of -26.97%.

The two entry prices are far apart because QID's adjusted price series fell from the mid-$80s to the mid-$20s over the window. The adjusted prices are lower than the quotes printed at the time. The cross rule bought a fund that was already deep in a long decline in 2025, and it was buying near the start of a stretch of more falls. Both signals came when the 50-day average crossed above the 200-day average after QID had risen for some weeks, and both times the rise reversed.

The median hold was 332 days and the longest hold was 332 days, so the median figures from the facts are driven by one trade. The shortest hold was 57 days. The average hold was 194.5 days. The profit factor is 0 because there were no winning trades, and the longest losing streak was 2.

Two trades are not a sample. The golden cross is a template designed to trade rarely, and on a fund that rises in one year in six it traded twice. What the table shows is how late a 50/200 cross reacts: the first entry was on 2022-03-11, after QID had started to rise, and the exit was on 2023-02-06, long after its 2022 peak. The cross turned the fund's 2022 rally from a gain of 66.4% in the calendar year into a gain of 24.1% on the account, and then into a trade that ended below its entry price.

Largest drawdowns

Buy-and-hold's deepest drawdown ran from 2021-03-08 to 2026-10-02 and reached −87.3%.

A 48.41% drawdown from one October to one June

The largest drawdown on the account began on 2022-10-14 and reached its low on 2025-06-11, 971 days later. Its depth was 48.41%, and the account had not recovered by 2026-10-02. This single drawdown holds both trades: the first trade's late-2022 giveback and the second trade's loss.

The second-largest drawdown, 35.1%, ran from 2022-06-16 to a low on 2022-08-15 and recovered on 2022-10-14, after 60 days. The third, 27.15%, was in March 2022, 15 days from peak to low and 41 days to recover. Both of these came inside the first trade, where the template held through two sharp pullbacks in a fund whose daily range averaged 3.15%.

Buy and hold had one deep drawdown of 87.33%, from 2021-03-08 to 2026-10-02, and had not recovered. The golden cross drawdown was 38.92 points shallower in the gap table. The shorter exposure is the reason, since the account was in cash for most of the period in which QID lost value.

The rank table puts that in context. On QID, golden cross placed 3rd among the 12 templates at -6.03%. Ahead of it were RSI(2) snapback at -1.85% with a 59.04% drawdown and momentum breakout at -3.08% with a 45.66% drawdown. Trend plus trailing stop was next at -7.6%. At the bottom were weekly 7% target at -26.24%, dip buyer at -27.11% and monthly cycle at -28.81%. The templates that placed best all stayed out of the fund for most of the window.

With trading costs

The headline run fills at the bar price. These runs charge slippage on every fill.

Costs on two trades

With 4 fills in total, slippage has little to act on. At 5 basis points per fill the CAGR was -6.07% and the end equity was $6,980.18. At 10 basis points it was -6.09% and $6,971.99. The headline run, with no slippage, ended at $6,997.75. The drawdown went from 48.41% to 48.52% and 48.57%.

The cost of each extra basis point is a few dollars on this account. QID trades about $228,567,798 a day on average, with a median minute volume of 5,631 shares, so a position sized to a $10,000 account fills without difficulty. The test does not model market impact, and the headline run assumes fills at the bar price.

Changing the parameters

Three moving-average variants

Three variants change the averages. Each has a different lesson.

SMA 40/200 shortens the fast average to 40 days. It made 3 trades with no wins and returned -8.86% a year, with a 57.84% drawdown and $5,872.14 at the end. One extra trade and a shorter fast average made the result worse than the published 50/200. A faster average crosses earlier in a rise and earlier in a fall, and on QID the earlier entry was not rewarded.

SMA 50/150 shortens the slow average to 150 days. This was the worst variant: -16.8% a year, a 75.18% drawdown, 4 trades with 1 win and $3,479.54 at the end. A shorter slow average puts the two lines closer, so the cross happens more often and the whipsaw cost rises. The extra winning trade did not offset the losses.

SMA 60/250 lengthens both. It returned -0.83% a year with a 36.03% drawdown, 2 trades with 1 win and $9,532.03 at the end. It was the only variant with a positive Sharpe ratio, 0.088. A slower pair of averages meant the template entered later or skipped a signal, and in this window being later cost less than being earlier.

The pattern across the three is that the slower the cross, the less it lost on a fund that decays. That says something about QID in this window. It does not say that 60/250 is a better rule. With 2 to 4 trades per variant, one trade moving by a few days would reorder them.

How QID behaved

Why a 50/200 cross fits QID badly

QID is a -2x daily fund on the Nasdaq-100. Its realized beta to QQQ was -2 and the underlying returned 151.07% over the window, against -89.27% for the fund. The leverage table shows a decay gap of 212.87 percentage points between what a simple -2x of the underlying would suggest and what the fund returned. A fund that resets its leverage every day loses value in a volatile, rising market, and QID lost value in most of the years.

The long-run trend filter sees this clearly. QID spent 20% of sessions above its 200-day average and crossed it 34 times in 1,444 sessions. Of the 34 crossings, the 50/200 cross turned into a position only twice. A fund that spends most of its time under its 200-day average gives a golden cross few chances to fire, and the chances it does get are short rallies inside a downtrend.

The fund rose on 45.11% of days. The average up day was 2.16% and the average down day was -2.02%. Lag-1 autocorrelation was -0.03, which is essentially zero, so a move on one day told nothing about the next. Annualized volatility was 44.7% and the average daily range was 3.15%. The worst day was -23.53% on 2025-04-09, and the best was +12.12% on 2025-04-04. The overnight share of the log return was 42.16% and the intraday share was 57.84%.

Over 116 sessions RSI(14) was below 30, and the median 5-day return afterwards was +0.12% against a baseline of -1.17%. The median 20-day return was -3.09% against a baseline of -3.55%. RSI(2) below 10 occurred on 248 sessions with a median 5-day return of -1.17% and 20-day return of -3.93%. After oversold readings the fund did a little better than its baseline, which is why RSI(2) snapback was the best template on QID. Slow trend rules had no similar edge here.

The fund's correlation to QQQ was -1 and to SPY -0.94, with betas of -2 and -2.57. Its correlation to TLT was -0.09. TQQQ and QQQ were its least correlated funds at -1, and SQQQ and PSQ were the most correlated at 1. A golden cross on QID is therefore close to a golden cross on the inverse of QQQ, and the template spent 2022 and early 2025 holding it while the Nasdaq-100 fell.

The rules

Hold while the 50-day average is above the 200-day; step aside when it crosses below (the death cross).

A widely used trend filter. When the 50-day simple moving average is above the 200-day, the asset is in a long-term uptrend and the strategy holds. When it crosses below, the strategy moves to cash. It trades rarely, with a handful of signals per decade on an index. Its use is skipping the deepest bear markets, and it will lag some rallies.

Good for: long-horizon investors who want to hold trends but sidestep multi-year bear markets.
Watch out: crosses lag at turning points: the strategy gives back the first leg of a crash and misses the first leg of a recovery, and choppy sideways markets whipsaw it.

How the rules compare with other inverse funds

Among the 11 inverse funds, QID's -6.03% placed 7th. TBF was the best at +8.47% with a 17.5% drawdown and 4 round trips. EEV made 0.71%. SPDN lost 2.26%, SH lost 2.31%, SDS lost 2.33% and PSQ lost 4.34%. Below QID were REW at -6.71%, SQQQ at -7.41%, TECS at -21.03% and SOXS at -24.44%.

The pattern follows the leverage. The 1x funds lost the least and the 3x funds lost the most, with the 2x funds between. The median CAGR for golden cross across all 59 funds was 2.05%, and the category median across inverse funds was -4.34%. On QID the template was slightly below the category median. It also ranked 53rd of 59 funds for this template.

The rules do two things on a decaying fund: they keep the account out for most of the time, and they enter late when a rally has already started. The first helped, as the 23.43 points a year over buy and hold shows. The second produced both losing trades. Neither is a weakness of the template in general. A trend filter is designed for a fund with a long-term trend, and QID has a long-term downtrend with short rallies. The test shows the template holding up better than the fund and still losing 30% of the starting capital. It covers one window, one fund and two trades, and the cost runs use only 5 and 10 basis points.

Related

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.