DeployQuantTLT trading strategies, backtested
iShares 20+ Year Treasury Bond ETF: long-duration Treasuries, one of the most rate-sensitive bond ETFs. Every DeployQuant template run on TLT over 5.7 years of minute data, same engine, same window, sorted by return.
TLT holds Treasury bonds with maturities of 20 years or more, and it was one of the worst places to hold money in this window. From 2021-01-04 to 2026-10-02 buy-and-hold returned negative 8.23% a year, turned $10,000 into $6,106, and fell 42.77% from the first day of the data to 2023-10-19 without ever recovering the starting level. In the profile series the fund lost 40.49% in total, with a 31.21% loss in 2022 alone.
Every template beat holding on CAGR, and only one of the 12 finished positive: RSI(2) snapback at 0.96%. Eleven had a shallower drawdown than holding. The exception was the monthly cycle, which had a 46.24% drawdown, and it still finished 0.15 points of CAGR ahead of holding.
A bond fund with a falling price gives rules very little to work with. A template that is out of the fund avoids the loss, and a template that is in the fund takes it. The headline runs have no fees or slippage, and the window is one rate cycle.
| Strategy | CAGR | max DD | Sharpe | trades | win rate | final value |
|---|---|---|---|---|---|---|
| RSI(2) Dip Snapback | 1.0% | −22.8% | 0.15 | 172 | 59% (+1 open) | $10,564 |
| EMA 12/26 Trend | −0.7% | −15.9% | -0.03 | 21 | 29% | $9,619 |
| SMA 10/50 Trend | −0.8% | −14.2% | -0.04 | 17 | 35% | $9,524 |
| 20-Day Momentum + Trailing Stop | −1.2% | −10.3% | -0.14 | 2 | 0% | $9,328 |
| 3-Month Momentum Switch | −2.3% | −18.6% | -0.33 | 7 | 29% | $8,760 |
| Golden Cross (SMA 50/200) | −2.5% | −20.6% | -0.30 | 5 | 0% | $8,668 |
| Weekly Entry + 7% Target | −3.8% | −31.0% | -0.22 | 104 | 13% | $8,017 |
| Drawdown Dip Buyer + 8% Target | −4.3% | −27.8% | -0.27 | 0 | – (+1 open) | $7,763 |
| RSI(14) Mean Reversion | −5.4% | −36.0% | -0.42 | 13 | 54% (+1 open) | $7,270 |
| SMA-200 Trend + 15% Trailing Stop | −7.1% | −38.8% | -0.64 | 3 | 0% (+1 open) | $6,546 |
| 200-Day SMA Regime Filter | −7.6% | −40.2% | -1.04 | 48 | 15% | $6,360 |
| First-to-Last Day of Month | −8.1% | −46.2% | -0.52 | 69 | 43% (+1 open) | $6,167 |
| Buy & hold TLT | −8.2% | −42.8% | -0.52 | – | – | $6,106 |
Ranking templates on a fund that fell
The ranking by CAGR was RSI(2) snapback at 0.96%, EMA 12/26 at negative 0.67%, SMA 10/50 at negative 0.85%, 20-day momentum with a trailing stop at negative 1.20%, 3-month momentum at negative 2.28%, the golden cross at negative 2.46%, the weekly 7% target at negative 3.78%, the dip buyer at negative 4.32%, RSI(14) mean reversion at negative 5.40%, the trailing-stop template at negative 7.11%, the 200-day filter at negative 7.58% and the monthly cycle at negative 8.08%. Holding was negative 8.23%.
The gaps between the top six come almost entirely from time spent out of the fund. SMA 10/50 and EMA 12/26 were each invested 42.2% of the time and had drawdowns of 14.24% and 15.92%. In 2022, when holding lost 30.2%, EMA 12/26 lost 0.2% and SMA 10/50 lost 8.3%. In 2023, holding made 2.7%, EMA made 5.4% and SMA 10/50 made 9.6%. Those two years are most of their lead. The trend templates made no money in 2024 or 2025: EMA 12/26 lost 8.3% in 2024 and 2.7% in 2025, and SMA 10/50 lost 6.9% in 2024 and 3.8% in 2026, and made 3.6% in 2025.
Both trend templates won 29% and 35% of their trades. A trend rule on a bond fund that rallies in brief bursts and falls the rest of the time has a low hit rate. EMA 12/26's best trade was 8.20% from 2023-11-10 to 2024-01-22, a 73-day hold at the turn of the 2023 rally. SMA 10/50's best was 5.70% over the same stretch. Their worst trades, 3.85% and 5.49%, were small.
RSI(2) snapback is the one positive result, and it needs context. It made 172 round trips, won 59%, and had a profit factor of 1.10, with an average win of 1.06% against an average loss of 1.35%. It was invested 45.2% of the time. It lost 17.5% in 2022 while holding lost 30.2%, then made 11.8% in 2023 and 16.0% in 2025. Its drawdown of 22.81% started 2022-01-28 and was recovered on 2025-12-11, 1,130 days after the low. A profit factor of 1.10 over 172 trades is a thin edge, and the cost runs below remove it.
The golden cross and the trailing-stop template had no winning trade at all. The golden cross made 5 round trips, all losers, from a loss of 1.03% to one of 5.44%. The trailing-stop template made 3, losing 8.56%, 14.63% and 9.22%, and holds a position entered 2025-02-26 at an adjusted 84.90 that was down 8.73% at the end of the window. Both finished ahead of holding because they were out of the fund for stretches. The golden cross was invested 34.6% of the time and the trailing-stop template 64.5%.
The 200-day filter, with 48 round trips and a win rate of 15%, returned negative 7.58%, only 0.65 points of CAGR above holding. Its profit factor was 0.14, with a longest losing streak of 12. On a fund that crossed its 200-day average 95 times in the data, a filter built on that line gets whipsawed. The monthly cycle, invested 95.2% of the time, was in effect holding with 139 fills.
The dip buyer made one entry, on 2022-05-03 at 100.55, and never reached its 8% target. It was down 22.94% on that position at the end. Its CAGR of negative 4.32% counts that open loss, and its 0 closed trades make the win rate not applicable. It is the clearest example here of a rule that buys a falling fund and waits for a bounce that does not arrive.
RSI(14) mean reversion made 13 round trips and won 54%, but its average loss was 7.95% against an average win of 3.90%, and one trade lost 27.19% over 325 days from 2022-01-04 to 2022-11-25. The rule produced that trade, and the template lost 26.5% in 2022.
The rallies of 2023 and the positions still open
The fund made 2.7% in 2023 on the template basis, and the two rally months that year, November at 9.49% and December at 8.44% in the template run, decided how each rule did. The SMA 10/50 and EMA 12/26 trend rules both had December 2023 as their best month at 8.44%, and the SMA 10/50 template made 9.6% that year against 5.4% for the EMA. The RSI(2) snapback made 11.8% and the weekly 7% target 8.8%. The 200-day filter lost 17.7% in 2023 and the trend plus trailing stop lost 13.2%, so the two rules built on the long average missed the rally and took the preceding falls.
The 2022 column shows the opposite split. Holding lost 30.2%, and the monthly cycle lost 30.7%. The RSI(14) mean reversion lost 26.5% and the weekly target 14.9%. The trend rules lost 0.2% and 8.3%, and the 3-month momentum switch made 0%. A fund that falls 31% in a year pays the templates that are mostly in cash, so the rules that were mostly in cash did best.
Three templates end the window inside a losing position. The dip buyer entered on 2022-05-03 at 100.55 and was down 22.94%, which is why its CAGR of -4.32% sits below most templates, with 76.8% exposure and no closed trade. The trend plus trailing stop entered on 2025-02-26 at 84.9 and was down 8.73%. The RSI(14) template entered on 2026-07-16 at 82.87 and was down 6.5%. In each case the rule had not yet met its exit when the data ended, and the open loss sits inside the CAGR.
The calendar-year view also shows how few good years the fund had. It made 2.79% in 2023 and 4.23% in 2025 and lost in the other four, including -8.07% in 2024 and -8.01% in 2026 to date. The worst month for holding was April 2022 at -9.19%, a month in which the RSI(14) template also lost 9.21%, within a tenth of a point of the fund.
How each strategy traded TLT
| Strategy | Time in market | Avg hold (days) | Best trade | Worst trade | Profit factor | With 10 bps slippage |
|---|---|---|---|---|---|---|
| RSI(2) snapback | 45.2% | 5 | 4.2% | −5.6% | 1.10 | −4.8% |
| EMA 12/26 trend | 42.2% | 43 | 8.2% | −3.9% | 0.85 | −1.4% |
| SMA 10/50 trend | 42.2% | 52 | 5.7% | −5.5% | 0.80 | −1.4% |
| momentum breakout | 17.7% | 188 | −1.0% | −6.0% | – | −1.3% |
| 3-month momentum | 23.1% | 69 | 2.3% | −6.0% | 0.16 | −2.5% |
| golden cross | 34.6% | 144 | −1.0% | −5.4% | – | −2.6% |
| weekly 7% target | 85.0% | 16 | 7.9% | −4.3% | 0.78 | −6.8% |
| dip buyer | 76.8% | – | – | – | – | −4.3% |
| RSI mean reversion | 57.1% | 86 | 8.4% | −27.2% | 0.47 | −5.9% |
| trend + trailing stop | 64.5% | 257 | −8.6% | −14.6% | – | −7.3% |
| 200-day regime filter | 30.3% | 14 | 2.7% | −3.3% | 0.14 | −9.1% |
| monthly cycle | 95.2% | 28 | 9.4% | −8.3% | 0.59 | −10.2% |
Costs and the few trades that set each result
RSI(2) snapback is the template most exposed to costs, with 345 fills. At 5 basis points its CAGR fell from 0.96% to negative 1.96%, and at 10 basis points to negative 4.81%, with the drawdown rising to 28.22%. Its positive result disappears with a small charge, and the result at 10 basis points of $7,537 is behind several slower templates. The weekly 7% target with 208 fills went from negative 3.78% to negative 5.44% and negative 6.84%. The 200-day filter went from negative 7.58% to negative 8.36% and negative 9.10%. The monthly cycle went from negative 8.08% to negative 10.22%, which is worse than holding.
The templates with few trades barely moved: EMA 12/26 went from negative 0.67% to negative 1.37%, SMA 10/50 from negative 0.85% to negative 1.41%, and the golden cross, the momentum template and the dip buyer by less than a tenth of a point.
At 10 basis points momentum at negative 1.27% was the best, ahead of EMA 12/26 at negative 1.37%, SMA 10/50 at negative 1.41% and 3-month momentum at negative 2.52%. No template made money once 10 basis points were charged.
Average holds show how differently the templates behaved. The trailing-stop template held 257 days on average, the golden cross 144.4, the trend templates about 42 to 52 days, RSI(2) snapback 5.3 days, and the 200-day filter 13.5 days. The 200-day filter's 48 trades won 7 times. Its best was 2.68% over 111 days from 2024-07-03 to 2024-10-22. Its worst was 3.26% over 6 days from 2025-04-02 to 2025-04-08.
The momentum template made 2 trades, both losers: 5.98% over 243 days from 2022-12-02 to 2023-08-02 and 0.98% over 133 days from 2023-11-30 to 2024-04-11. It was invested 17.7% of the time. Its small drawdown of 10.32% and CAGR of negative 1.20% say the rule rarely traded, and when it did, it lost a little. 3-month momentum made 7 trades, won 2, and was invested 23.1% of the time. It lost 5.98% over 20 days from 2025-04-02 to 2025-04-22.
For the results in this window, trade counts below 10 should be read as anecdotes. The cases with more than 100 trades, RSI(2) snapback and the weekly template, are the ones where the CAGR is a reasonable summary of the rule. Both lose money after costs. The weekly template won 13% of its 104 round trips, with a longest losing streak of 16. Its average win was 7.13% and its average loss was 1.22%. That is a pattern of many small losses and a few capped wins, and on a fund that is trending down the wins are rare.
On US equity index funds the templates mostly gave up return to cut the drawdown. On TLT in this window there was little return to give up, so every template came out ahead of holding on CAGR.
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 |
Calendar years
| Year | Return |
|---|---|
| 2021 | −4.5% |
| 2022 | −31.2% |
| 2023 | 2.8% |
| 2024 | −8.1% |
| 2025 | 4.2% |
| 2026 (part) | −8.0% |
Biggest single days
| Best day | Move |
|---|---|
| 2022-11-10 | 3.9% |
| 2023-03-10 | 3.4% |
| 2022-09-28 | 3.4% |
| 2021-02-26 | 3.3% |
| 2022-12-01 | 3.2% |
| Worst day | Move |
|---|---|
| 2022-03-02 | −3.5% |
| 2022-06-13 | −3.1% |
| 2025-04-07 | −3.0% |
| 2023-05-01 | −2.8% |
| 2025-04-10 | −2.7% |
Average return by calendar month
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.4% | −0.7% | −1.7% | −2.5% | −0.8% | 1.5% | 0.3% | −0.8% | −3.2% | −2.6% | 4.4% | −1.0% |
Most and least correlated funds
| Most correlated | Least correlated | ||
|---|---|---|---|
| TMF | 1.00 | TBF | -1.00 |
| BND | 0.92 | RINF | -0.40 |
| AGG | 0.92 | KMLM | -0.39 |
| IEF | 0.91 | CTA | -0.28 |
| UST | 0.86 | USDU | -0.24 |
How TLT behaved in this window
TLT's annualized volatility was 15.33%, its average intraday range was 0.94% and its average overnight gap was 0.54%. It rose on 49.00% of days, with an average up day of 0.74% and an average down day of negative 0.77%. Its average daily dollar volume was $2,159,787,473.
The return split is the most unusual number on the page. The overnight log return was negative 70.11% and the intraday log return was 18.78%. The overnight share of the return was 136.58% and the intraday share was negative 36.58%. Over this window the fund gained during the session and lost much more between the close and the next open. A rule that bought at the open and sold at the open was exposed to the overnight leg on every day it held. That is a feature of this fund over this window, and the data does not say why.
The calendar years were negative 4.53% in 2021, negative 31.21% in 2022, 2.79% in 2023, negative 8.07% in 2024, 4.23% in 2025 and negative 8.01% in 2026 through 2026-10-02. TLT gained in two of the six. The drawdown ran from 2021-01-04 to 2023-10-19, reaching 44.06% in the profile table and 42.77% in the buy-and-hold run. It had not recovered by the end of the data, and the longest drawdown lasted 1,443 sessions. In the buy-and-hold run the fund took 1,018 days to reach the low.
Correlation to SPY was 0.08 and to QQQ 0.09, with betas of 0.07 and 0.06. The most correlated funds were TMF at 1.00, then BND and AGG at 0.92, IEF at 0.91 and UST at 0.86. The least correlated was TBF at negative 1.00, then RINF at negative 0.40, KMLM at negative 0.39, CTA at negative 0.28 and USDU at negative 0.24. A long-bond fund did not hedge US equities in the sense of moving opposite to them. The correlation was close to zero.
The biggest days were 2022-11-10 at 3.90%, 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 negative 3.47%, 2022-06-13 at negative 3.09%, 2025-04-07 at negative 3.05%, 2023-05-01 at negative 2.83% and 2025-04-10 at negative 2.72%. TLT fell on 2025-04-07 and 2025-04-10 during the equity sell-off, and 3-month momentum, which entered on 2025-04-02, lost 5.98% over that stretch.
Average monthly returns were best in November at 4.43% and June at 1.51%, and weakest in September at negative 3.16%, October at negative 2.63%, April at negative 2.55% and March at negative 1.74%. The data has 5 or 6 observations per month.
The fund was above its 200-day average in 35.18% of sessions and crossed it 95 times. Compared with a fund that trends, that is a high number of crossings for a low share of time above, and it explains why the 200-day filter made 48 trades and won 7. RSI(14) fell below 30 in 61 sessions, and the median forward 5-day return was 0.02% against a baseline of negative 0.14%. The 20-day return was negative 1.63% against negative 0.60%, so buying oversold readings on TLT gave a worse 20-day result than the baseline. RSI(2) fell below 10 in 203 sessions, with a median 5-day return of 0.03% and a 20-day return of negative 0.65%, close to the baselines. Lag-1 autocorrelation was negative 0.05. There were 3 falls of 10% or more from a 20-day high, over 5 days.
The window is a single rate cycle. Other bond funds in the data, such as IEF, AGG and SGOV, had much smaller losses, and the results on this page should not be read as results for bonds in general.
Oversold readings, weekdays and bond peers
The oversold signals did not work on TLT. RSI(14) fell below 30 on 61 sessions, and the median forward 5-day return after them was 0.02% against a baseline of -0.14%. The median 20-day return was -1.63% against -0.6%. RSI(2) fell below 10 on 203 sessions, with forward medians of 0.03% over 5 days and -0.65% over 20, against the same baselines. A buy after a drop produced roughly the same 5-day result as any other day and a worse 20-day result. The RSI(14) template shows it: one trade lost 27.19% from 2022-01-04 to 2022-11-25, and its 13 round trips had a profit factor of 0.47. The RSI(2) rule held for about 5 days on average and made 0.96%.
The fund had only 3 falls of 10% or more from a 20-day high, over 5 days. The fund's losses came as a slow decline, and the dip buyer, which waits for a 10% fall, entered once and never reached its target. The lag-1 autocorrelation was -0.05.
The weekday averages were -0.19% on Monday, 0.02% on Tuesday, 0.12% on Wednesday, -0.08% on Thursday and -0.05% on Friday. The Monday figure is the largest, and with a 0.77% average down day it is still inside the noise. No weekday rule is supported.
The funds correlated with TLT were TMF at 1, BND and AGG at 0.92, IEF at 0.91 and UST at 0.86. The only strongly opposite fund was TBF at -1, and the least correlated others were RINF at -0.4, KMLM at -0.39 and CTA at -0.28.
The shorter bond funds give a comparison. IEI held at -0.36% and its best template, the trend plus trailing stop, made 1.43%. IGIB held at 0.21% and the EMA 12/26 trend made 2.35%. AGG held at -0.78% and the RSI(2) snapback made 1.57%. On TLT the same snapback rule made 0.96% from a hold of -8.23%. TLT traded about $2,159,787,473 a day with a median of 35,636 shares a minute.
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 TLT?
Of the 12 templates tested on TLT over 2021-01-04 to 2026-10-02, the strongest by CAGR was RSI(2) snapback at 1.0% (max drawdown 22.8%), versus −8.2% 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 TLT?
12 of 12 templates beat TLT buy-and-hold (−8.2% CAGR) on this window; 11 of 12 had a shallower maximum drawdown than holding (42.8%).
What was the best strategy for TLT?
[RSI(2) snapback](/learn/strategies/rsi2-dip-snapback/tlt/) had the highest CAGR at 0.96% and was the only template with a positive result. Holding returned negative 8.23%. At 10 basis points of slippage the RSI(2) result was negative 4.81%.
Did any strategy beat buying and holding TLT?
All 12 templates beat holding on CAGR, and 11 had a shallower drawdown than the 42.77% for holding. The monthly cycle had a deeper drawdown at 46.24%. Only one template had a positive CAGR.
Why did TLT fall so much from 2021 to 2026?
The fund returned negative 31.21% in 2022 and was negative in 2021, 2024 and 2026 as well. The data shows what the price did and does not give a cause. It fell 44.06% from 2021-01-04 to 2023-10-19 in the profile table.
Does TLT move opposite to stocks?
Not in this window. Correlation to SPY was 0.08 and to QQQ 0.09. TLT fell on 2025-04-07 at negative 3.05% and on 2025-04-10 at negative 2.72%, during the equity sell-off.
How did trend strategies do on TLT?
[SMA 10/50](/learn/strategies/sma-10-50-trend/tlt/) returned negative 0.85% and EMA 12/26 negative 0.67%, against negative 8.23% for holding. Both were invested 42.2% of the time and did best in 2022 and 2023. The golden cross lost on all 5 of its trades.
Do costs change the TLT results?
They remove the one positive result. RSI(2) snapback went from 0.96% to negative 1.96% at 5 basis points because it made 345 fills. The monthly cycle fell below holding at 10 basis points, at negative 10.22%.
Other bond 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.