DeployQuantGolden Cross (SMA 50/200) on TLT
iShares 20+ Year Treasury Bond ETF: long-duration Treasuries, one of the most rate-sensitive bond ETFs. Backtest 2021-01-04 to 2026-10-02, $10,000 starting capital, computed by the same engine that runs live DeployQuant strategies.
Golden cross on TLT lost money, and it still finished ahead of holding. The rule buys the 20+ year Treasury fund when the 50-day average closes above the 200-day average and sells when it crosses below. From 2021-01-04 to 2026-10-02 it turned $10,000 into $8,668, a CAGR of -2.46%, with a maximum drawdown of 20.58%. Holding TLT over the same days ended at $6,106, a CAGR of -8.23%, with a drawdown of 42.77%. The rule beat holding by 5.77 points of CAGR and cut the drawdown by 22.19 points while losing money in absolute terms.
All five closed round trips lost. The win rate was 0%, the average loss was 2.87%, the best trade was -1.03% and the worst was -5.44%. The rule was invested 34.6% of the time and held each position for 144.4 days on average. The gain over holding came from the days it was in cash while TLT fell.
Among the 12 templates run on TLT, golden cross ranks 6th by CAGR. Among the 59 ETFs tested with this rule, TLT ranks 50th. The golden cross strategy page gives a median CAGR of 2.05% across all funds and 0.78% across bond funds, so TLT's -2.46% sits below both.
Year by year
| Year | golden cross | buy & hold |
|---|---|---|
| 2021 | 2.8% | −4.4% |
| 2022 | −7.9% | −30.2% |
| 2023 | −2.8% | 2.7% |
| 2024 | −4.3% | −7.7% |
| 2025 | −1.8% | 4.0% |
| 2026 | 0.2% | −7.7% |
Year by year against holding
The rule returned 2.8% in 2021 against -4.4% for holding, a lead of 7.2 points. In 2022 it lost 7.9% while TLT lost 30.2%, a lead of 22.3 points, the largest of any year. In 2023 it lost 2.8% while holding gained 2.7%, in 2024 it lost 4.3% against -7.7%, in 2025 it lost 1.8% while holding gained 4%, and in 2026 so far it has made 0.2% against -7.7%.
The rule beat holding in four of the six years: 2021, 2022, 2024 and 2026. It trailed in 2023 and 2025, the two years TLT rose. The rising years were small ones for TLT, so the rule gave up little by missing them. TLT's calendar-year returns were -4.53% in 2021, -31.21% in 2022, 2.79% in 2023, -8.07% in 2024, 4.23% in 2025 and -8.01% so far in 2026. Four of six years were negative for the fund, and the rule lost money in four.
The 2021 gain came from October and November, after the first entry on 2021-10-19. The 2022 loss of 7.9% came entirely from January and February, before the exit on 2022-02-16. After that the rule was in cash for the rest of the year, while TLT fell the full 31.21% over 2022. The gap of 22.3 points in 2022 is most of the case for the rule on this fund.
Month by month
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2021 | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 2.2% | 2.7% | −2.0% |
| 2022 | −3.8% | −4.2% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% |
| 2023 | 0.0% | 0.0% | 0.0% | 0.0% | −2.0% | 0.2% | −1.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% |
| 2024 | 0.4% | −2.2% | 0.8% | −6.3% | 4.2% | −1.6% | 3.5% | 2.1% | 1.9% | −5.3% | 1.9% | −3.2% |
| 2025 | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | −0.8% | 1.4% | 0.2% | −2.6% |
| 2026 | −0.0% | 4.4% | −4.1% | −0.8% | 0.9% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | – | – |
The monthly record
The month table shows long runs of 0.0%. From January to September 2021 the rule was flat. Its first entry came on 2021-10-19, the same date as on the other funds tested with this rule. October 2021 made 2.16% and November 2.71%, and December lost 2.04%. January 2022 lost 3.81% and February 4.24%, and then every month from March 2022 to April 2023 shows 0.0%.
The rule re-entered on 2023-05-04. May 2023 lost 2.01%, June made 0.22% and July lost 1.01%, and the exit on 2023-07-26 left every month from August 2023 to December 2023 flat. TLT's best month in the holding series was November 2023 at 9.49%, a month the rule spent in cash. Missing it is part of why the rule trailed holding in 2023.
The next entry, on 2024-01-31, produced a jagged year. January made 0.41%, February lost 2.2%, March made 0.78% and April lost 6.28%, the rule's worst month. May made 4.19%, and the exit on 2024-05-21 locked in a 3.45% loss on the trade. The re-entry on 2024-06-26 gave June -1.6%, July 3.55%, August 2.09% and September 1.86%, then October -5.29%, November 1.87% and December -3.17%. The 2024 result of -4.3% is the sum of these small gains and losses, and it is the clearest case of whipsaw on this page: two trades in one year, both losing.
In 2025 the rule was flat from January to August and re-entered on 2025-09-16. September lost 0.77%, October made 1.35%, November 0.23% and December lost 2.56%. In 2026 it made -0.01% in January, 4.44% in February, -4.1% in March, -0.82% in April and 0.91% in May, with the exit on 2026-05-08. Holding's worst month was April 2022 at -9.19%, which the rule skipped.
Every trade
golden cross on TLT made 5 closed round trips, an average hold of 144 days, an average loser of −2.87%, a longest losing streak of 5. It held a position at the close on 34.6% of trading days.
| Entry | Entry price | Exit | Exit price | Return | Days held |
|---|---|---|---|---|---|
| 2021-10-19 | $120.85 | 2022-02-16 | $114.28 | −5.4% | 120 |
| 2023-05-04 | $91.24 | 2023-07-26 | $88.63 | −2.9% | 83 |
| 2024-01-31 | $85.55 | 2024-05-21 | $82.60 | −3.5% | 111 |
| 2024-06-26 | $84.31 | 2024-12-17 | $83.44 | −1.0% | 174 |
| 2025-09-16 | $85.80 | 2026-05-08 | $84.43 | −1.6% | 234 |
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 five trades
Every trade lost.
The first entered on 2021-10-19 at $120.85 and exited on 2022-02-16 at $114.28, a loss of 5.44% over 120 days. It is the largest loss and it spans the start of the 2022 decline. The fall continued for the rest of the year while the rule sat in cash, which is why it gained against holding.
The second entered on 2023-05-04 at $91.24 and exited on 2023-07-26 at $88.63, a loss of 2.86% over 83 days, the shortest hold. The third entered on 2024-01-31 at $85.55 and exited on 2024-05-21 at $82.6, a loss of 3.45% over 111 days. The fourth entered on 2024-06-26 at $84.31 and exited on 2024-12-17 at $83.44, a loss of 1.03% over 174 days, the smallest. The fifth entered on 2025-09-16 at $85.8 and exited on 2026-05-08 at $84.43, a loss of 1.6% over 234 days, the longest hold.
In each trade the exit came below the entry. A rule that buys when the 50-day average rises above the 200-day average buys after TLT has already rallied, and then sells after it has rolled over. That fits a fund that drifted and swung without a long uptrend. The median trade lost 2.86%, the median hold was 120 days, and no trade came close to a win. The five losses together cost the account 13.32%, against a fall of 38.94% for holding.
The facts show no open position at the end: the last exit on 2026-05-08 was followed by cash through 2026-10-02.
Entry and exit prices across the five trades
The prices show how the cash periods helped and hurt. After the first exit at $114.28 on 2022-02-16 the rule next bought at $91.24 on 2023-05-04, so it avoided the whole slide in between and re-entered far lower. That one gap is the source of the lead over holding in 2022. After the second exit at $88.63 the rule bought again at $85.55 on 2024-01-31, a lower price once more.
The 2024 exits went the other way. The rule sold at $82.6 on 2024-05-21 and bought back at $84.31 on 2024-06-26, higher than where it left. It sold at $83.44 on 2024-12-17 and bought again at $85.8 on 2025-09-16, also higher. Neither re-entry was far above the exit, and the fund moved within a narrow band through these trades, which is why the 2024 and 2025 trades lost between 1.03% and 3.45% each. A 50-day average that crosses a 200-day average inside a range produces signals that mean little, and the golden cross rule has no filter to skip them.
The last exit, at $84.43 on 2026-05-08, left the rule in cash for the rest of the window, which is why the facts show no open position and why the 2026 line of the year table stops at 0.2%. The test ends with the rule flat and a fund that was still in a drawdown of 42.77% from its first day.
Largest drawdowns
| Peak | Low point | Depth | Days to low | Recovered | Days to recover |
|---|---|---|---|---|---|
| 2021-12-03 | 2024-04-25 | −20.6% | 874 | not yet | – |
| 2021-11-09 | 2021-11-23 | −4.2% | 14 | 2021-11-30 | 7 |
| 2021-10-18 | 2021-10-21 | −1.3% | 3 | 2021-10-26 | 5 |
Buy-and-hold's deepest drawdown ran from 2021-01-04 to 2023-10-19 and reached −42.8%.
Drawdowns
The rule's deepest drawdown was 20.58%, from the peak on 2021-12-03 to the low on 2024-04-25, 874 days later, and it has not recovered. That is a long period for a rule that spent most of the window in cash, because the losses accumulated across all five trades while the fund drifted. The next two were small: 4.16% from 2021-11-09 to 2021-11-23, recovered in 7 days, and 1.34% from 2021-10-18 to 2021-10-21.
Holding had a single drawdown that never ended. It began on 2021-01-04, bottomed at 42.77% on 2023-10-19, 1,018 days later, and had not recovered by 2026-10-02. TLT's own price series shows a deepest drawdown of 44.06% over the same dates, and a longest drawdown of 1,443 sessions, the whole window. The rule's drawdown peaked on 2021-12-03 and bottomed on 2024-04-25, later than holding's low on 2023-10-19, and it was far shallower.
The rule's drawdown is not a result of any single event. The largest single loss was the first trade at 5.44%, and the drawdown of 20.58% comes from the sum of the 2021 to 2022 loss, the May to July 2023 loss and the 2024 losses. The April 2024 month alone lost 6.28%, which is the largest piece.
With trading costs
The headline run fills at the bar price. These runs charge slippage on every fill.
| Slippage per fill | CAGR | Max drawdown | Final value | Sharpe |
|---|---|---|---|---|
| None (headline) | −2.5% | −20.6% | $8,668 | -0.30 |
| 5 basis points | −2.5% | −20.8% | $8,623 | -0.32 |
| 10 basis points | −2.6% | −20.9% | $8,581 | -0.33 |
Slippage on ten fills
The rule made ten fills. With 5 basis points of slippage per fill the CAGR was -2.55% and the ending value $8,623. With 10 basis points the CAGR was -2.63% and the ending value $8,581. The drawdown moved from 20.58% to 20.76% and 20.94%. Each step costs less than a tenth of a point of CAGR, and the rule's lead over holding does not depend on it. TLT traded $2,159,787,473 a day on average, with a median minute volume of 35,636 shares, so a $10,000 order is small against that. The test does not measure spreads.
Changing the parameters
| Version | CAGR | Max drawdown | Round trips | Win rate | Final value |
|---|---|---|---|---|---|
| Published rules | −2.5% | −20.6% | 5 | 0% | $8,668 |
| SMA 40/200 | −2.9% | −23.5% | 6 | 17% | $8,438 |
| SMA 50/150 | −2.1% | −16.7% | 8 | 13% | $8,858 |
| SMA 60/250 | −2.3% | −16.3% | 4 | 0% | $8,741 |
The 40/200, 50/150 and 60/250 versions
The three variants change the averages. SMA 40/200 returned -2.92% with a drawdown of 23.46%, from 6 trades and 1 win, ending at $8,438. SMA 50/150 returned -2.09% with a drawdown of 16.72%, from 8 trades and 1 win, ending at $8,858. SMA 60/250 returned -2.32% with a drawdown of 16.26%, from 4 trades and no wins, ending at $8,741. The published 50/200 returned -2.46% with a drawdown of 20.58%.
None of the variants made money, and none changed the sign. The two slower variants had shallower drawdowns than the published rule. The faster 40/200 version lost slightly more, and the 50/150 version made more trades, 8, than any other and still lost less than the published rule. The range of CAGR across the four versions is narrow, from -2.92% to -2.09%, which suggests the problem is the fund and not the choice of averages. With one or no winning trade in each run, there is no setting here that found an uptrend in TLT.
Risk-adjusted view of the variants
Every variant has a negative Sharpe ratio: -0.365 for 40/200, -0.23 for 50/150 and -0.372 for 60/250. The 50/150 version is the least negative, with a CAGR of -2.09%, and it is also the one with eight trades. Moving the slow average from 200 to 150 days made the rule react earlier to the 2024 range, and that did not produce a win either, as the variant had one winner in eight trades. The headline run's costs show the same sign: at 5 and 10 basis points the Sharpe ratio was -0.315 and -0.327. A strategy that loses money before costs has nothing for costs to erode, and the cost runs only widen the loss.
How TLT behaved
| Measure | TLT |
|---|---|
| Data in this test | 2021-01-04 to 2026-10-02 (1444 sessions) |
| Total return, buy and hold | −40.5% |
| Annualized volatility | 15.3% |
| Deepest drawdown | −44.1% (2021-01-04 to 2023-10-19) |
| Up days | 49.0% |
| Average daily range | 0.94% |
| Average overnight gap | 0.54% |
| Correlation to SPY | 0.08 |
| Correlation to QQQ | 0.09 |
| Sessions above the 200-day average | 35.2% |
| Crossings of the 200-day average | 95 |
| Falls of 10% or more from a 20-day high | 3 |
What TLT does to a slow trend rule
TLT's annualized volatility was 15.33%, its average daily range 0.94% and its deepest drawdown 44.06%. It was down from the first day to 2023-10-19 and never recovered its starting price by the end of the window. Of the six calendar years, four were negative.
The fund crossed its 200-day average 95 times and spent only 35.18% of sessions above it. That is the core problem for a golden cross. A 50/200 cross is slow, but a fund that crosses its 200-day average that often does not stay in any one trend for long enough to pay for the lag. Falls of 10% or more from a 20-day high occurred only 3 times, covering 5 days. TLT's decline came without sharp 20-day drops.
The best days were 2022-11-10 at 3.9%, 2023-03-10 at 3.42%, 2022-09-28 at 3.41%, 2021-02-26 at 3.31% and 2022-12-01 at 3.24%. The worst were 2022-03-02 at -3.47%, 2022-06-13 at -3.09%, 2025-04-07 at -3.05%, 2023-05-01 at -2.83% and 2025-04-10 at -2.72%. Up days were 49% of sessions, with an average up day of 0.74% and an average down day of -0.77%. The overnight log return was -70.11% and the intraday log return 18.78%. All of the fund's loss came overnight, and intraday trading was positive. A rule that holds through the night, as every template here does, carries that loss.
After RSI(14) fell below 30, on 61 sessions, the median forward 5-day return was 0.02% against a baseline of -0.14%, and the 20-day return was -1.63% against -0.6%. RSI(2) below 10 occurred on 203 sessions, with a median 5-day return of 0.03% and a 20-day return of -0.65%. Both sets of 5-day returns are close to zero. The RSI(2) snapback turned that into the best result on the fund, 0.96%, but no signal on TLT produced a large edge.
By calendar month, November averaged 4.43% and June 1.51%, while September averaged -3.16%, October -2.63%, April -2.55% and March -1.74%. These are averages over five or six observations. TLT's correlation to SPY was 0.08 and to QQQ 0.09. It correlated 1 with TMF, 0.92 with BND and AGG, 0.91 with IEF and 0.86 with UST. The least correlated were TBF at -1, RINF at -0.4 and KMLM at -0.39.
Seasonality and the T-bill comparison
TLT's average return by calendar month was weakest in September at -3.16%, October at -2.63% and April at -2.55%, with November the standout at 4.43%. The rule's losses fall in some of the same months. April 2024 lost 6.28%, October 2024 lost 5.29%, March 2026 lost 4.1% and September 2025 lost 0.77%. Each month has five or six observations, so the overlap is a coincidence of dates in one window. November 2023, the fund's best month at 9.49%, fell in a cash period for the rule.
Monday averaged -0.19% and Wednesday 0.12%, and the lag-1 autocorrelation was -0.05. RSI(14) rose above 70 on only 21 sessions, against 61 sessions below 30. The fund was oversold far more often than overbought, which fits a decline that lasted the whole window.
The cash-like SGOV made 3.22% with the same rule and no trades. That is higher than every template on TLT, where the best was 0.96%. The comparison shows what holding long Treasuries cost in this window: the fund's own CAGR was -8.64% and no entry rule recovered it to a gain.
The rules
Hold while the 50-day average is above the 200-day; step aside when it crosses below (the death cross).
- WHEN the market opens · IF not invested AND SMA(50) > SMA(200) · THEN buy with 98% of the sleeve
- WHEN the market opens · IF invested AND SMA(50) < SMA(200) · THEN sell the whole position
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.
Other templates and other bond funds
On the same fund, the RSI(2) snapback was the best template at 0.96% with a drawdown of 22.81%. EMA 12/26 returned -0.67% with 15.92%, SMA 10/50 -0.85% with 14.24%, and momentum breakout -1.2% with 10.32%. 3-month momentum returned -2.28%, then golden cross at -2.46%, the weekly 7% target at -3.78%, the dip buyer at -4.32% and RSI mean reversion at -5.4%. The three worst were trend plus trailing stop at -7.11%, the 200-day regime filter at -7.58% and the monthly cycle at -8.08%, the last with a drawdown of 46.24%.
The faster trend rules, EMA 12/26 and SMA 10/50, did better than golden cross on TLT. The slower rules that tolerate a long drawdown, the trailing stop and the 200-day filter, did worse.
The same golden cross on other bond funds returned 1.84% on IGIB, 1.09% on IEI, 0.78% on AGG, 0.72% on BND and -0.29% on IEF, against 3.22% on SGOV with no trades. TLT is the long-duration fund in the group and had the lowest result. All of the numbers come from one window of 5.7 years with daily decisions, minute-bar fills, no margin and no fees in the headline run. Prices are adjusted for splits and dividends.
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
Did golden cross beat buy-and-hold on TLT?
Over 2021-01-04 to 2026-10-02, golden cross on TLT returned −2.5% annualized vs −8.2% for buy-and-hold: it beat buy-and-hold by 5.8% per year, with a maximum drawdown 22.2 points shallower than holding (20.6% vs 42.8%).
How many trades did it make?
5 completed round trips over 5.7 years (10 fills), with 0% of round trips closing profitably.
How often does a golden cross happen?
On a broad index, roughly every couple of years. The backtest pages show the exact trade count for each ETF over the 2021 to 2026 window.
Golden cross vs buy and hold: which does better?
It depends on the asset and the window. Every backtest page here shows the same-window buy-and-hold comparison.
Did golden cross beat buy-and-hold on TLT?
Yes, though it lost money. The rule returned -2.46% a year against -8.23% for holding, and ended at $8,668 against $6,106. Its maximum drawdown was 20.58%, against 42.77%.
How many trades did golden cross make on TLT, and did any win?
It made 5 closed round trips and none won. The losses ranged from 1.03% to 5.44%, with an average of 2.87%. There was no open position at the end.
Why did golden cross do better than holding in 2022 on TLT?
TLT lost 31.21% in 2022, and the rule lost 7.9%. The rule sold on 2022-02-16 after losing in January and February, and stayed in cash for the rest of the year.
Does changing the moving averages help golden cross on TLT?
Not in this window. SMA 40/200 returned -2.92%, SMA 50/150 returned -2.09% and SMA 60/250 returned -2.32%, against -2.46% for 50/200. None made money.
Which strategy was best on TLT?
The RSI(2) snapback had the highest CAGR at 0.96%, with a drawdown of 22.81%. Momentum breakout had the shallowest drawdown at 10.32%, with a CAGR of -1.2%. Golden cross ranked 6th of 12.
Do trading costs matter for golden cross on TLT?
Very little. With 10 basis points per fill the CAGR was -2.63% against -2.46% at no cost, since the rule made only ten fills.
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.