DeployQuantRSI(2) Dip Snapback on QAI
NYLI Hedge Multi-Strategy Tracker ETF: replicates a basket of hedge-fund strategies. Backtest 2021-01-04 to 2026-10-02, $10,000 starting capital, computed by the same engine that runs live DeployQuant strategies.
QAI is a fund that tries to replicate a basket of hedge-fund strategies, and it moves slowly. Its annualized volatility over the test window was 6.52% and its average daily range was 0.53%. The RSI(2) snapback rule buys QAI when the 2-day RSI is below 10 and sells when it recovers above 70. Between 2021-01-04 and 2026-10-02 it turned $10,000 into $12,140, a CAGR of 3.43%. Holding QAI from the first day turned $10,000 into $12,451, a CAGR of 3.89%. The rule trailed the fund by 0.46% a year and finished $311 behind.
What the rule bought for that small shortfall was a shallower fall. Its maximum drawdown was 6.84%, and holding the fund went through 14.65%. The Sharpe ratio was 0.78 for the rule and 0.64 for buy-and-hold. On a risk-adjusted basis the rule did better than the fund even though it earned less in dollars.
The rule made 152 closed round trips and held one more open at the end, which comes to 305 fills. The win rate was 65%. The average winner was 0.53% and the average loser was -0.61%, with a profit factor of 1.65. It held a position at the close on 36.8% of trading days. The edge per trade is small, and the cost runs below show how quickly slippage can erase it.
Among the 12 templates run on QAI this one ranked fourth by CAGR. Across the 59 funds it ranked 33rd for this template, and the template's median CAGR across all funds was 4.74%. Within the six alternative-strategy funds, QAI came fifth by CAGR, ahead of only ALTY.
All of this is one 5.74 year test with daily decisions and no fees in the headline run.
Year by year
| Year | RSI(2) snapback | buy & hold |
|---|---|---|
| 2021 | 0.4% | −0.2% |
| 2022 | −0.4% | −8.6% |
| 2023 | 4.5% | 10.0% |
| 2024 | 3.1% | 6.6% |
| 2025 | 0.7% | 8.0% |
| 2026 | 11.7% | 8.0% |
Year by year against the fund
The rule had a positive return in five of the six calendar years and a loss of 0.4% in 2022. Buy-and-hold had losses in 2021 and 2022. The rule finished ahead of the fund in three years, 2021, 2022 and 2026, and behind in 2023, 2024 and 2025.
2021 was flat for both: 0.4% for the rule and -0.2% for the fund, a gap of 0.6 points. The rule made 21 round trips that year and won 14.
2022 was the year that mattered. QAI lost 8.6% in the buy-and-hold column, and 8.77% on the raw fund. The rule lost 0.4%, a gap of 8.2 points in its favour, the widest gap of any year in either direction. It made 32 round trips in 2022, the most of any year, and won 19. The fund's deepest drawdown ran from 2021-02-17 to 2022-10-14. The rule was invested on only about a third of days, and many of its entries in 2022 were short trades that closed within days at gains of 1% or more: 1.4% from 2022-01-11, 1.5% from 2022-03-14 and 1.4% from 2022-11-29. These are three of the ten best trades of the whole test.
2023 went to the fund. QAI returned 10.0% and the rule 4.5%, a gap of 5.5 points. It made 29 round trips and won 20.
2024 was 3.1% for the rule and 6.6% for the fund, 25 round trips with 15 wins. In 2025 the rule returned 0.7% and the fund 8.0%, a gap of 7.3 points, the largest in the fund's favour. The rule made 23 round trips and won 17, so its trades were mostly profitable but small. The year includes its second-worst trade, a 2.8% loss from 2025-02-21 to 2025-03-13.
2026 is a partial year and the rule did better: 11.7% against 8.0%, a gap of 3.7 points in the rule's favour. It made 22 round trips and won 14. The best trade of the test was in this year, 3.0% from 2026-06-11 to 2026-06-15, and so were three others among the best ten.
The pattern is that the rule gave up return in the years the fund rose steadily and gained ground in the year it fell. Over the whole window those roughly cancelled, leaving the rule 0.46 points a year short.
Month by month
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2021 | −0.4% | −0.4% | 0.7% | 0.8% | −0.0% | 0.6% | 0.6% | 0.3% | −0.6% | 0.3% | −1.4% | −0.2% |
| 2022 | 0.5% | 0.3% | 0.2% | 1.3% | −0.3% | −3.0% | 1.2% | −1.5% | −2.2% | 0.5% | 2.0% | 0.7% |
| 2023 | 0.4% | −0.8% | 0.7% | 1.1% | −0.1% | −0.1% | 0.9% | 0.3% | 0.6% | −0.0% | 0.9% | 0.6% |
| 2024 | 0.4% | 1.0% | 1.0% | 0.2% | 0.4% | 0.8% | 0.9% | 1.0% | −0.7% | −0.5% | −0.1% | −1.1% |
| 2025 | 0.6% | −0.2% | −2.4% | −0.3% | 0.7% | 1.0% | 0.0% | 0.6% | 1.1% | 0.8% | −1.6% | 0.5% |
| 2026 | 2.5% | 1.5% | 1.4% | 1.3% | 1.4% | 2.7% | −0.1% | 0.2% | −0.0% | 0.4% | – | – |
Months: small numbers either side of zero
No month moved the account by more than 3.02%. The best month was June 2026 at 2.73%, helped by the 3.0% trade of 2026-06-11. The worst was June 2022 at -3.02%. The rule entered on 2022-06-10 at $26.61 and exited on 2022-06-16 at $25.86, a loss of 2.8%, tying for the worst trade of the test. The fund's own worst day of June was 2022-06-13 at -2.27%, which fell between that entry and exit.
For the fund held outright, April 2026 at 4.52% was the best month and September 2022 at -3.91% the worst. The rule's September 2022 was -2.2%, a smaller loss. In that month the rule made a trade from 2022-09-19 to 2022-09-29 that lost 2.4%.
The month table shows how often the rule returned only a few tenths of a percent. In 2023 the rule's monthly results ran from -0.8% to 1.1%, and in 2024 from -1.1% to 1.0%. Compare that with the fund's own volatility of 6.52% a year. A rule that holds for 36.8% of days takes only a share of that movement.
2026 stands out for its string of gains: 2.5% in January, 1.5% in February, 1.4% in March, 1.3% in April, 1.4% in May and 2.7% in June, six months in a row of 1.3% or more. After June the monthly figures dropped to -0.1%, 0.2%, -0.0% and 0.4%. The first half of 2026 produced most of the year's 11.7%.
QAI's own month-of-year averages are small. November averaged 1.58% over five years, January 0.93% and May 0.79%, while September averaged -0.63% and March -0.46%. With five or six observations per month, none of this is a seasonal rule.
Every trade
RSI(2) snapback on QAI made 152 closed round trips and one position still open at the end of the test, an average hold of 5 days, an average winner of 0.53%, an average loser of −0.61%, a profit factor of 1.65, a longest losing streak of 4. It held a position at the close on 36.8% of trading days.
Best 10 round trips
| Entry | Entry price | Exit | Exit price | Return | Days held |
|---|---|---|---|---|---|
| 2026-06-11 | $35.76 | 2026-06-15 | $36.84 | 3.0% | 4 |
| 2026-03-30 | $33.60 | 2026-04-01 | $34.34 | 2.2% | 2 |
| 2026-02-06 | $34.16 | 2026-02-10 | $34.74 | 1.7% | 4 |
| 2026-05-19 | $35.57 | 2026-05-22 | $36.16 | 1.7% | 3 |
| 2022-03-14 | $27.31 | 2022-03-17 | $27.73 | 1.5% | 3 |
| 2022-01-11 | $28.55 | 2022-01-13 | $28.96 | 1.4% | 2 |
| 2022-11-29 | $26.26 | 2022-12-01 | $26.63 | 1.4% | 2 |
| 2024-08-05 | $29.16 | 2024-08-08 | $29.51 | 1.2% | 3 |
| 2024-05-01 | $29.33 | 2024-05-06 | $29.67 | 1.2% | 5 |
| 2023-01-31 | $27.11 | 2023-02-02 | $27.41 | 1.1% | 2 |
Worst 10 round trips
| Entry | Entry price | Exit | Exit price | Return | Days held |
|---|---|---|---|---|---|
| 2022-06-10 | $26.61 | 2022-06-16 | $25.86 | −2.8% | 6 |
| 2025-02-21 | $31.68 | 2025-03-13 | $30.80 | −2.8% | 20 |
| 2022-09-19 | $25.88 | 2022-09-29 | $25.27 | −2.4% | 10 |
| 2021-11-18 | $29.17 | 2021-12-08 | $28.68 | −1.7% | 20 |
| 2025-11-13 | $33.22 | 2025-11-25 | $32.77 | −1.4% | 12 |
| 2022-03-02 | $28.08 | 2022-03-10 | $27.71 | −1.3% | 8 |
| 2024-12-11 | $31.47 | 2024-12-23 | $31.08 | −1.2% | 12 |
| 2021-02-17 | $29.54 | 2021-03-02 | $29.22 | −1.1% | 13 |
| 2023-02-16 | $26.93 | 2023-02-24 | $26.65 | −1.0% | 8 |
| 2022-01-19 | $28.54 | 2022-02-01 | $28.25 | −1.0% | 13 |
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 trades: many, short and mostly small
There were 153 trades including the one open at the end. The 152 closed trades had a median return of 0.23% and a median hold of 4 days. The longest hold was 20 days and the shortest was 1. The best trade returned 3.02%. The worst lost 2.82%. Winners averaged 0.53% and losers averaged -0.61%, so each loser was slightly larger than each winner. The profit factor of 1.65 comes from the win rate of 65%, since 99 of the 152 closed trades made money.
The longest winning streak was 11 trades and the longest losing streak was 4. The open position was bought on 2026-09-25 at $36.25 and was marked at a loss of 0.03% on the last close.
The best trades cluster in a few episodes. 2026 provided four of the top ten: 3.0% on 2026-06-11, 2.2% on 2026-03-30, 1.7% on 2026-02-06 and 1.7% on 2026-05-19. 2022 provided three: 1.5%, 1.4% and 1.4%. These came after drops of the fund that the 2-day RSI picked up, and each was held for 2 to 4 days.
The worst trades tell a different story. The two worst, at -2.8%, came on 2022-06-10 and 2025-02-21. The second was held for 20 days, against the median of 4. When a dip keeps dipping, the rule waits for RSI(2) to recover above 70 and has no stop. The trade from 2021-11-18 to 2021-12-08 also lasted 20 days and lost 1.7%. All ten of the worst trades were held for 6 days or more, against a median hold of 4 days, which fits the strategy's caveat about a crash that keeps crashing.
The rule's expected profit per trade is small. A winner of 0.53% on a fund with an average daily range of 0.53% means the typical win is about one day's range. That makes the strategy sensitive to anything that moves the entry or exit price, which the next section tests.
Largest drawdowns
| Peak | Low point | Depth | Days to low | Recovered | Days to recover |
|---|---|---|---|---|---|
| 2024-08-06 | 2025-04-08 | −6.8% | 245 | 2026-01-23 | 290 |
| 2022-05-04 | 2022-09-27 | −6.5% | 146 | 2023-11-14 | 413 |
| 2021-11-17 | 2022-03-07 | −3.3% | 110 | 2022-04-07 | 31 |
Buy-and-hold's deepest drawdown ran from 2021-02-17 to 2022-10-14 and reached −14.7%.
Drawdowns
The largest drawdown was 6.84%, from a peak on 2024-08-06 to a low on 2025-04-08 after 245 days. It recovered on 2026-01-23, 290 days later. The fund's own drawdown in the same stretch, from 2025-02-18 to 2025-04-08, was 7.63%, with recovery on 2025-06-26. The rule's drawdown began earlier, in August 2024, while the fund's began in February 2025, and the rule took longer to climb out.
The second drawdown, 6.55%, ran from 2022-05-04 to 2022-09-27 and took until 2023-11-14 to recover, 413 days. This is the rule's worst stretch in terms of time. It overlaps the June 2022 loss and the September 2022 trade, and the fund's own bad period. The third, 3.28%, ran from 2021-11-17 to 2022-03-07 and recovered by 2022-04-07, in 31 days.
Buy-and-hold's deepest drawdown ran from 2021-02-17 to 2022-10-14, a depth of 14.65%, and took until 2024-03-12 to recover. The raw fund's longest drawdown lasted 770 sessions. The rule's 6.84% is much shallower than the fund's, which is the benefit it delivered. The fund's own record shows no fall of 10% or more from a 20-day high, over the whole window, so QAI never gave the rule a crash to sit through.
A shallow drawdown can still be long. The first of the three took 245 days to reach its low and the second 146 days, and recoveries took 290 and 413 days. For a rule that makes small gains each trade, recovery is slow, since each trade adds little.
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) | 3.4% | −6.8% | $12,140 | 0.78 |
| 5 basis points | 0.8% | −8.8% | $10,456 | 0.20 |
| 10 basis points | −1.8% | −18.4% | $9,005 | -0.38 |
Slippage erased the edge
This is the most important table on the page. The headline run fills at the bar price. At 5 basis points of slippage per fill, the CAGR fell from 3.43% to 0.78%, the final value from $12,140 to $10,456, and the Sharpe ratio from 0.78 to 0.196. The maximum drawdown rose from 6.84% to 8.75%. At 10 basis points the CAGR was -1.81%, the final value $9,005 and the Sharpe ratio -0.383. The maximum drawdown was 18.41%.
The rule made 305 fills. Each fill pays the slippage once, so each round trip pays it twice. The average winner is 0.53%, small enough that slippage takes a large share of it. This rule is the highest-turnover template in the library, in the template's own words, and on a fund whose daily moves are this small, the headline return and the cost-adjusted return are very different numbers.
Buy-and-hold made one fill and so paid the slippage once. At 5 basis points the rule's CAGR of 0.78% is far under the fund's 3.89%, and at 10 basis points the rule loses money. The headline run therefore flatters the rule against the fund, and the cost runs reverse the result.
The drawdown at 10 basis points, 18.41%, is larger than the fund's own 14.65%. The cost runs show a fragile edge in practice. QAI traded $3,152,202 a day on average, with a median of 317 shares per minute bar. Trading that thin means actual fills could differ from the bar price by more than 5 or 10 basis points, and the test does not model that.
Changing the parameters
| Version | CAGR | Max drawdown | Round trips | Win rate | Final value |
|---|---|---|---|---|---|
| Published rules | 3.4% | −6.8% | 152 | 65% | $12,140 |
| RSI(2) < 5 / > 70 | 3.9% | −5.7% | 148 | 66% | $12,462 |
| RSI(2) < 15 / > 70 | 3.0% | −7.5% | 165 | 63% | $11,839 |
| RSI(2) < 10 / > 60 | 3.5% | −7.5% | 161 | 64% | $12,189 |
| RSI(2) < 10 / > 80 | 4.2% | −6.4% | 149 | 65% | $12,633 |
Changing the thresholds
Four variants moved the buy or sell level. Buying below 5 instead of 10 returned 3.91% with a 5.7% drawdown over 148 trades, 98 of them winners, and ended at $12,462. Buying below 15 returned 2.98% with a 7.53% drawdown over 165 trades and ended at $11,839. Selling above 60 returned 3.51% with a 7.45% drawdown over 161 trades, ending at $12,189. Selling above 80 returned 4.16% with a 6.41% drawdown over 149 trades, ending at $12,633.
Three of the four beat the published rule's 3.43% on return, and the stricter entry of 5 and the later exit of 80 also had shallower drawdowns. The Sharpe ratios were 0.896 and 0.918 for those two, against 0.78 for the published rules. The looser entry of 15 was the only variant to do worse than the published rule on CAGR and Sharpe, at 0.665.
The trend across variants is mild. A tighter entry means fewer, more extreme dips, and a later exit means holding for more of the recovery. Both produced fewer trades than the published rule's 152 and a better result. That fits the cost picture: with fewer fills, less slippage is paid. These runs have no slippage, so the gain here comes from the rule itself, though the effect would be larger once costs are counted.
The range of CAGR across the five rows is 2.98% to 4.16%, and the buy-and-hold CAGR is 3.89%. Two variants edged above the fund. Four variants on one window do not show which setting is right, only that the result did not collapse when the thresholds moved.
How QAI behaved
| Measure | QAI |
|---|---|
| Data in this test | 2021-01-04 to 2026-10-02 (1444 sessions) |
| Total return, buy and hold | 25.3% |
| Annualized volatility | 6.5% |
| Deepest drawdown | −14.9% (2021-02-17 to 2022-10-14) |
| Up days | 51.4% |
| Average daily range | 0.53% |
| Average overnight gap | 0.27% |
| Correlation to SPY | 0.82 |
| Correlation to QQQ | 0.80 |
| Correlation to TLT | 0.18 |
| Sessions above the 200-day average | 75.7% |
| Crossings of the 200-day average | 15 |
| Falls of 10% or more from a 20-day high | 0 |
How QAI behaved and why dips came and went
QAI returned 25.27% over 1,444 sessions, a CAGR of 4.0% on the raw data. It was up on 51.35% of days, and both the average up day and the average down day were about 0.3%, 0.31% and -0.32%. The deepest drawdown was 14.94%, from 2021-02-17 to 2022-10-14, recovering on 2024-03-12. Its correlation to SPY was 0.82 and to QQQ 0.8, with a beta to SPY of 0.33. The correlation to TLT was 0.18.
The 2-day RSI fell below 10 on 154 sessions. After those sessions the median return over the next 5 days was 0.29% against a baseline of 0.16%, and over 20 days it was 0.66% against 0.45%. That is a modest edge, and it explains the rule: the rebound exists, and it is small. The 14-day RSI fell below 30 on only 24 sessions, with a median 5-day return of 0.67% and a 20-day return of 0.86%.
A median 5-day return of 0.29% is not much to work with, and the average winner of 0.53% matches that. The trade-by-trade picture and the forward-return picture agree: this fund snaps back a little after a washout.
QAI closed above its 200-day average on 75.74% of sessions and crossed it 15 times. It never fell 10% or more from a 20-day high. For a fund with annualized volatility of 6.52%, the dips that trigger RSI(2) are small ones.
One oddity in the profile is the split of returns between the overnight and intraday parts. The overnight log return was -48.34% and the intraday log return was 70.66%, so the fund gained during sessions and lost between close and open. The rule buys and sells at the open, so when it is invested it holds through the overnight stretch where the fund tended to lose, which is a possible drag on the rule's return.
Among the alternative-strategy funds, CLSE returned 16.58% with this rule and RINF 8.52%. CTA returned 5.47%, KMLM 3.95% and QAI 3.43%. ALTY returned 3.33%. QAI had the shallowest drawdown of the six, 6.84% against 7.52% for CLSE.
The rules
Buy extreme 2-day RSI washouts under 10, exit as soon as RSI(2) recovers above 70.
- WHEN the market opens · IF not invested AND RSI(2) < 10 · THEN buy with 98% of the sleeve
- WHEN the market opens · IF invested AND RSI(2) > 70 · THEN sell the whole position
A short-horizon mean-reversion template popularized by Larry Connors' RSI-2 research. A 2-period RSI under 10 flags a sharp multi-day selloff. In assets with a persistent upward drift, those selloffs have tended to snap back within days. Trades are frequent and short. This is the highest-turnover template in the library.
Good for: liquid index ETFs with strong long-term drift; turnover is high so per-trade edges are small.
Watch out: high trade counts make results sensitive to execution quality; a crash that keeps crashing will hand this template several losing entries in a row.
Other templates on QAI
Three templates beat this rule's 3.43% on QAI. The golden cross returned 4.28% with a 7.63% drawdown. The 200-day regime filter returned 4.26% with a 6.02% drawdown, the best risk profile of the group. The weekly 7% target returned 4.21% with a 15.8% drawdown.
Below the snapback came the monthly cycle at 3.28% with a 16.07% drawdown, the EMA 12/26 trend at 2.76% and the trend with a trailing stop at 2.61%. RSI mean reversion managed 2.15%. Close behind were the SMA 10/50 trend and the 3-month momentum, each a little above 2.1%. Two templates, the dip buyer and the momentum breakout, show 0% because they made no trades on QAI.
This rule's 6.84% drawdown was the third shallowest among the templates that traded, after the 3-month momentum's 5% and the 200-day filter's 6.02%. It is also the highest-turnover template, with 152 round trips here. The QAI overview page has all 12, and the RSI(2) template page shows this rule on every fund.
These comparisons use the headline runs, which have no fees.
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 RSI(2) snapback beat buy-and-hold on QAI?
Over 2021-01-04 to 2026-10-02, RSI(2) snapback on QAI returned 3.4% annualized vs 3.9% for buy-and-hold: it trailed buy-and-hold by 0.5% per year, with a maximum drawdown 7.8 points shallower than holding (6.8% vs 14.7%).
How many trades did it make?
152 completed round trips over 5.7 years (305 fills), with 65% of round trips closing profitably.
How often does RSI(2) trade?
Far more than RSI(14), with dozens of round trips per year on a volatile ETF. The backtest table on each page shows the exact count over the test window.
Is RSI(2) too fast for daily bars?
It is designed for daily bars. The 2-day window catches short, sharp washouts rather than long regimes.
Did RSI(2) snapback beat buy-and-hold on QAI?
No, by a small margin. It returned 3.43% a year against 3.89% for buy-and-hold and ended at $12,140 against $12,451. Its maximum drawdown was 6.84%, against 14.65% for holding.
How many trades did it make?
It made 152 closed round trips plus one open position, 305 fills in all. Of the closed trades, 99 were winners, a win rate of 65%. The average hold was 5 days.
How much did trading costs matter?
A great deal. At 5 basis points of slippage per fill the CAGR fell from 3.43% to 0.78%, and at 10 basis points it was -1.81%. The average winning trade was only 0.53%.
What was the worst drawdown?
6.84%, from the peak on 2024-08-06 to the low on 2025-04-08. It recovered on 2026-01-23. A second drawdown of 6.55% in 2022 took 413 days to recover.
Do different RSI thresholds change the result?
Somewhat. Buying below 5 returned 3.91%, selling above 80 returned 4.16%, and buying below 15 returned 2.98%. Those runs have no slippage, so they say little about costs.
Why is the profit per trade so small on QAI?
QAI has annualized volatility of 6.52% and an average daily range of 0.53%. After RSI(2) fell below 10, its median 5-day return was 0.29%. A rebound that small leaves little room for a rule that trades often.
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.