RSI(14) Mean Reversion on EEM

iShares MSCI Emerging Markets ETF: emerging-market equities, on a different cycle than US indexes. Backtest 2021-01-04 to 2026-10-02, $10,000 starting capital, computed by the same engine that runs live DeployQuant strategies.

RSI(14) mean reversion on the iShares MSCI Emerging Markets ETF (EEM) turned $10,000 into $16,851 between 2021-01-04 and 2026-10-02. That is a 9.51% CAGR with a maximum drawdown of 14.01% and a Sharpe ratio of 0.8. Buy-and-hold EEM ended at $14,416 with a 6.58% CAGR, a 39.09% drawdown and a Sharpe of 0.44. The rule beat the fund by 2.93% a year and held a drawdown 25.08 points shallower, while it was invested only 35.7% of the time.

An entry happens at the following open once RSI(14) drops under 30 with nothing held. The order commits 98% of the sleeve. It sells everything at the open after RSI rises above 70. It has no profit target and no stop. The template's own caveat is that RSI can stay oversold for weeks in a persistent downtrend while the position keeps losing. On EEM the worst trade lost 2.57%.

The record is unusually clean. There were 18 completed round trips and 15 won, a win rate of 83%. The average win was 4.16% and the average loss was 2.4%, which gives a profit factor of 9.02. The best trade returned 11.91% and the worst lost 2.57%. Winners came in a run of up to 7 trades, while losers never ran past 2 in a row. Trades held for 41.2 days on average.

The rule was second of 12 templates on EEM by CAGR and eighth of 59 funds for this template. Among the 12 broad index ETFs it posted the highest CAGR. Other funds under this rule are covered on the RSI mean reversion strategy page. Every other template run on EEM sits on the EEM hub.

What was tested: 5.74 years, decisions once a day, no margin, and a headline run with zero fees and zero slippage. Eighteen trades is a small sample, and the reading below treats the numbers that way.

Year by year

Year by year against buy-and-hold

The rule finished positive in all six years. Buy-and-hold was negative in two of them, and those were the two years the rule beat it.

2021 returned 9.1% for the rule and -4.2% for the fund. The first trade ran from 2021-03-05 at 47.7 to 2021-06-03 at 49.18 for 3.1% over 90 days. Three round trips closed in the year and all three won, including a 3.98% gain from 2021-09-30 to 2021-10-20 and a 4.42% gain that began on 2021-11-29 and ended on 2022-01-13. October 2021 was the strongest month at 4.27%.

2022 is the year that stands out. The fund fell 20.2% in buy-and-hold terms, and the rule returned 7.9%. Five round trips closed in 2022 and 4 won. The rule bought on 2022-03-03 at 41.54, sold on 2022-04-04 at 41.95, bought again on 2022-04-25 at 37.57 and sold on 2022-06-03 at 38.46, then bought on 2022-07-18 at 35.76 and sold on 2022-08-16 at 37.09 for 3.72%. The one loser was the September entry on 2022-09-02 at 35.5, which sold on 2022-11-14 at 34.59 for -2.56% after 73 days. That trade gave the worst month of the test, September 2022 at -10.33%, though buy-and-hold lost 11.16% that month. A rule that buys oversold readings was in the market only a fraction of the time, so a falling fund could not take the whole loss.

2023 was almost flat at 0.4%, against 8.9% for the fund. Two trades closed, one win and one loss. The loss was -2.07% from 2023-02-13 to 2023-03-31. The win was 2.52% over 91 days from 2023-08-16 to 2023-11-15. November 2023 returned 7.97% for the rule, and September and October returned -2.95% and -3.23%.

2024 returned 2.1% against 6.2%, with 3 round trips and 2 wins. January and February gave 1.0% and 2.79%. The losing trade began on 2024-10-28 at 43.16 and sold on 2024-12-06 at 42.05 for -2.57%, the worst of the test.

2025 returned 22.9% against 33.1%. Three trades, 3 wins. The 2024-12-31 entry at 40.79 sold on 2025-02-18 at 43.47 for 6.57%. The April 2025 entry on 2025-04-07 at 37.86 sold on 2025-04-29 at 42.37 for 11.91%, the best trade in the record. April 2025 returned 11.67% for the rule, its best month.

2026 returned 13.5% against 23.9%, with 2 round trips and 2 wins, through 2026-10-02. They were 7.56% from 2026-03-13 to 2026-04-15 and 5.84% from 2026-07-30 to 2026-08-18.

That shape fits a rule which sits in cash for most sessions. It gave up return in the four years when the fund rose, 2023 through 2026, and it held up in the two years the fund fell.

Month by month

Reading the month table

The month table has more empty months than full ones, because the rule was invested 35.7% of the time. Months with 0.0% include the whole of 2023-04 to 2023-07, 2024-03 to 2024-06, 2025-05 to 2025-10, and 2026-05 and 2026-06. In those stretches no position was open.

The positive months outnumber the negative ones. The negative ones were 2021-03 at -0.96%, 2021-07 at -1.55%, 2021-11 at -0.56%, 2022-03 at -1.93%, 2022-06 at -0.05%, 2022-09 at -10.33%, 2022-10 at -1.91%, 2023-02 at -5.17%, 2023-09 at -2.95%, 2023-10 at -3.23%, 2024-07 at -0.12%, 2024-10 at -1.54%, 2024-11 at -2.65% and 2026-03 at -1.59%. Only September 2022 is deep. Most losing months are shallow because the rule is long a fund at an oversold reading and exits on recovery.

Three months account for much of the return. April 2025 returned 11.67% and November 2022 returned 10.84%, and April 2026 returned 9.13%. Those correspond to rebounds after sharp sell-offs. The fund's best days in the window included +6.91% on 2025-04-09 and +5.47% on 2026-04-08, and its worst included -5.53% on 2025-04-04 and -4.91% on 2026-03-03. The rule bought on 2025-04-07, which was between the two, and held the rebound.

Average returns by calendar month put November at 2.96%, January at 2.92% and May at 2.18%. The weak months were March at -1.06%, September at -0.98% and July at -0.93%. Each figure rests on 5 or 6 observations and describes only this window. The rule trades on RSI and not on the calendar, so they are context rather than a driver.

Where the fund's extreme days fell against the rule's positions

The fund's five biggest down days were 2026-06-05 at -6.52%, 2026-06-23 at -5.65%, 2025-04-04 at -5.53%, 2026-03-03 at -4.91% and 2022-10-24 at -3.89%. The rule was in cash for four of them. It held no position in June 2026, and it entered on 2025-04-07, the first session after the 2025-04-04 drop, and on 2026-03-13, after the 2026-03-03 drop. Only the 2022-10-24 day fell inside an open trade, the September 2022 position that closed on 2022-11-14 at 34.59, a loss of 2.56%.

The best days were mixed. The 8.12% gain on 2022-03-16 fell inside the trade that opened on 2022-03-03 at 41.54, and that trade still returned only 0.99% by 2022-04-04, which says the rally gave back much of its gain inside the hold. The 6.91% day on 2025-04-09 and the 5.47% day on 2026-04-08 were both captured, inside the 11.91% and 7.56% trades. The 4.21% gain on 2022-11-04 fell in the September 2022 trade. The 4.42% day on 2026-06-11 was missed, as the rule was in cash from mid-April to late July 2026.

November 2022 returned 10.84% for the rule against 14.98% for holding, which was the fund's best month. September 2022 shows the opposite risk, with the rule at -10.33% and the fund at -11.16%, a month in which the rule was exposed nearly as fully as the fund. The rule's drawdown figure of 14.01% largely records that month.

Every trade

RSI mean reversion on EEM made 18 closed round trips, an average hold of 41 days, an average winner of 4.16%, an average loser of −2.40%, a profit factor of 9.02, a longest losing streak of 2. It held a position at the close on 35.7% 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 trades that decided the result

The five best trades returned 11.91%, 7.56%, 6.57%, 5.84% and 4.42%. Two of them are in 2025 and two in 2026, so the recent years carried the largest wins. The median trade returned 3.1% and the median hold was 39 days. The longest hold was 91 days and the shortest was 18 days.

Three trades lost: -2.57%, -2.56% and -2.07%. Every other trade returned between 0.99% and 11.91%. This is where the profit factor of 9.02 comes from. Losses stayed under 3% while gains ran up to 11.91%, and 15 of 18 trades won. In a sample this size, one extra loser of 10% would change the picture a lot.

The three losing trades share a feature. They were all entered at an oversold reading that kept falling: the September 2022 entry at 35.5 sold lower at 34.59, the February 2023 entry at 37.26 sold at 36.49, and the October 2024 entry at 43.16 sold at 42.05. Each held for between 39 and 73 days. The exit rule waits for RSI above 70, which can take a long time when the fund drifts. The template has no stop.

Trade outcomes by year: 2021 went 3 for 3, 2022 4 for 5, 2023 1 for 2, 2024 2 for 3, 2025 3 for 3 and 2026 2 for 2. No year with more than two trades produced more than one loss. The record ends flat, with no open position on 2026-10-02: the last trade closed on 2026-08-18.

Re-entry prices after each exit

Comparing each exit with the next entry shows what the rule did with its cash. In 2022 the pattern was favourable. It sold on 2022-04-04 at 41.95 and bought again on 2022-04-25 at 37.57. It sold on 2022-06-03 at 38.46 and bought on 2022-07-18 at 35.76. It sold on 2022-08-16 at 37.09 and bought on 2022-09-02 at 35.5. Each re-entry came in below the previous exit, because RSI had to fall under 30 again before the next buy, and the fund had kept sliding in between.

The pattern held in a few later gaps. The exit on 2024-12-06 at 42.05 was followed by an entry on 2024-12-31 at 40.79, and the exit on 2025-02-18 at 43.47 was followed by the 2025-04-07 entry at 37.86. The rule left the fund with a gain of 6.57%, sat out the slide, and came back at a price well below its sale.

The gaps ran the other way after the 11.91% trade. The exit on 2025-04-29 was at 42.37 and the next entry did not come until 2025-11-24 at 52.18. From May to October 2025 every month in the table reads zero, and EEM kept rising, which is why 2025 returned 22.9% for the rule against 33.1% for holding. The exit on 2026-04-15 at 61.75 and the entry on 2026-07-30 at 62.5 were close in price, with no cash gap cost.

The first entry took place on 2021-03-05 at 47.7, two months into the window, because RSI did not fall under 30 in January or February 2021, the first two months of the table show zero. The first three trades lasted 90, 50 and 20 days. Later trades from 2024 on were shorter, at 20, 26, 39, 49, 22, 18, 33 and 19 days, with the exception of the 2023 trade that lasted 91 days.

Largest drawdowns

Buy-and-hold's deepest drawdown ran from 2021-02-17 to 2022-10-24 and reached −39.1%.

Drawdowns

The deepest drawdown for the rule was 14.01%. It began at a peak on 2022-09-12, reached its trough on 2022-10-24 after 42 days, and recovered on 2024-02-06 after another 470 days. That recovery is slow because the rule was out of the market for most of 2023 and held positions that gave little. The second was 10.03%, from a peak on 2022-01-12 to a trough on 2022-03-14, recovered on 2022-04-04 after 21 days. The third was 8.46%, from 2022-05-04 to 2022-05-12, with recovery on 2022-07-21.

All three drawdowns fall in 2022. The worst year for the fund was the worst year for the rule's equity, though the rule's losses were small by comparison. Holding EEM meant a 39.09% drawdown from 2021-02-17 to 2022-10-24, and the fund took until 2025-09-15 to recover, 1057 days after the trough. A second buy-and-hold drawdown of 14.01% ran from 2026-06-22 to 2026-07-29 and had not recovered by 2026-10-02, and a third of 13.29% ran from 2026-02-25 to 2026-03-30.

The rule's trough on 2022-10-24 is the same date as the fund's. The fund's worst day, -3.89%, also fell on 2022-10-24 in the list of its five worst days. The rule was long on that day, since the September entry was still open until 2022-11-14. The 14.01% depth of the rule's drawdown equals the depth of buy-and-hold's 2026 drawdown, and the two are unrelated events.

The rule's drawdown profile is unusually shallow for an equity fund with 19.86% annualized volatility. The main reason is time out of the market. Exposure was 35.7%, so the rest of the time the account was in cash, with no exposure to the daily swings.

With trading costs

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

Costs

The rule made 36 fills in 5.74 years, which is few. At 5 basis points the CAGR was 9.18%, the max drawdown 14.03%, the end equity $16,554.88 and the Sharpe 0.772. At 10 basis points the CAGR was 8.85%, the max drawdown 14.05%, the end equity $16,274.54 and the Sharpe 0.747. The headline was 9.51%, 14.01% and $16,850.84.

Costs reduced the result by small steps. Even at 10 basis points the rule's CAGR stayed above buy-and-hold's 6.58%, and the drawdown barely moved. EEM trades $1,332,500,625 on an average day and a median of 38,144 shares a minute. A $10,000 account sends orders far smaller than that.

Changing the parameters

Neighbouring thresholds

Four variants moved one threshold at a time.

Buying below 25 instead of 30 gave 12 trades with 10 wins, a CAGR of 9.09%, a max drawdown of 14.03% and a Sharpe of 0.913. It earned slightly less than the base 9.51% on fewer trades, and its Sharpe was higher than the base 0.8.

Buying below 35 gave 24 trades with 18 wins, a CAGR of 10.65%, a max drawdown of 14.4% and an end equity of $17,881.75. This is the highest CAGR of the variants, and the Sharpe was 0.794. A looser entry bought more dips and the fund's tendency to bounce paid for them.

Selling above 65 instead of 70 gave 19 trades with 14 wins, a CAGR of 6.4% and a max drawdown of 15.14%. Selling above 75 gave 12 trades with 8 wins, a CAGR of 5.11% and a max drawdown of 26.96%. The Sharpe was 0.583 and 0.428. The exit is the sensitive threshold here. An earlier exit cut the gains, and a later exit held the positions long enough that the drawdown rose to 26.96% with a worse return.

The entry variants stayed within a narrow band of CAGR, from 9.09% to 10.65%, and the exit variants dropped to 6.4% and 5.11%. The base setting is not the best on CAGR, since buying below 35 beat it, so 30 is not the top setting on this fund. Four variants on one fund and one window say little about robustness elsewhere.

How EEM behaved

Why EEM fit the rule

EEM tracks emerging-market equities. Over the window it returned 46.91% for a 6.93% CAGR, with annualized volatility of 19.86%. By calendar year the fund returned -4.3% (2021), -20.64% (2022), 9.17% (2023), 6.39% (2024), 33.91% (2025) and 24.37% in the 2026 part-year. It fell 39.82% from 2021-02-17 to 2022-10-24 and recovered on 2025-09-15, a stretch of 1148 sessions. Versus SPY the beta was 0.81 and the correlation 0.67.

A mean-reversion rule needs oversold readings that reverse. EEM provided them. RSI(14) fell below 30 on 27 sessions in the window. Following such readings, the median 5-day return was 2.94% and the median 20-day return 3.65%. Across every session the medians were 0.21% and 0.54%. The 27 sessions can overlap in time, so they represent fewer independent events than the count suggests, but the direction favoured the rule.

The fund also had a quirk in how it earned. 120.76% of its log return came intraday and -20.76% came overnight. The average overnight gap was 0.75% and the average intraday range was 1.04%. The rule buys and sells at the open, so it takes each overnight gap into the position at the next open, and the fund's net overnight return was negative.

Large moves were not frequent. The fund had 14 drawdown events of 10% inside 20 days, covering 34 days. Up days averaged 0.92% and down days -0.93%. RSI(14) rose above 70 on 78 sessions, which gave the exit signals. EEM closed above its 200-day average on 66.67% of sessions, with 31 crossings of that line.

Short-term RSI readings and weekday averages for EEM

EEM's RSI(2) closed below 10 on 163 sessions. The median 5-day forward return after those closes was 0.4% on 163 observations and the median 20-day return was 1.2% on 163, against baselines of 0.21% and 0.54%. The edge is positive but small, which may be why the RSI(2) snapback made 3.87% on this fund while the slower RSI(14) rule made 9.51%. The 14-day reading under 30 showed larger forward medians of 2.94% and 3.65% on 27 sessions, and the rule only trades that deeper washout.

The lag-1 autocorrelation of daily returns was -0.07, a mild sign that one day's move was partly reversed the next. The weekday averages were 0.04% on Mondays, 0.03% on Tuesdays, 0.12% on Wednesdays, 0.01% on Thursdays and -0.02% on Fridays. None is large enough to matter for a rule that trades a handful of times a year.

The fund's correlation to QQQ was 0.69 and to SPY 0.67, with a beta to TLT of 0.12. That puts EEM near a U.S. index in direction but with a much weaker link than VV or SPY have to each other, which fits the rule's results: the same RSI thresholds produced 9.51% here and a lower CAGR on SPY, VV and IWM.

The rules

Buy when the 14-day RSI drops below 30 (oversold), sell when it recovers above 70 (overbought).

A standard mean-reversion setup. The Relative Strength Index measures how stretched recent price action is. Readings under 30 have historically marked short-term washouts in uptrending assets. This template buys at the next session open and holds until RSI crosses back above 70. It has no profit target and no stop.

Good for: assets that trend up over time but overshoot on the way, such as broad index ETFs.
Watch out: in a persistent downtrend, RSI can stay oversold for weeks while the position keeps losing; there is no stop-loss in this template.

How RSI mean reversion compares on EEM

The rule ranked 2nd of 12 templates on EEM. Only the weekly 7% target earned more, at 11.62%, and it took a 27.4% drawdown. The trend plus trailing stop earned 8.85% with 16.33%. The golden cross earned 7.76% with 20.65% and the 200-day regime filter earned 7.76% with 17.74%. The dip buyer earned 7.14%.

Further down were the RSI(2) snapback at 3.87%, the monthly cycle at 3.14% with a 44.39% drawdown, the momentum breakout at 2.36% and the EMA 12/26 trend at 0.29%. The 3-month momentum lost 1.62% and the SMA 10/50 trend lost 3.89%. The SMA 10/50 trend, 3-month momentum and EMA 12/26 trend were the bottom three, and RSI mean reversion had the shallowest drawdown of all 12.

Among the broad index ETFs, EEM was first for this rule at 9.51%. VOOV earned 8.01% with 13.06%, QQQE 7.22%, VOOG 7.09% and VTV 6.8%. QQQM earned 6.46%, QQQ 6.4%, VV 5.02%, IOO 4.46%, SPY 4.37%, VOO 4.22% and IWM 0.73%. Median CAGR for the template came to 2.98% over the 59 funds and 6.46% for the broad index group. SPY and QQQ, the two best known, sat below EEM here.

The limits are the window and the sample. 5.74 years and 18 trades make a short record, and the 2022 sell-off carries much of the story. Different dates would change which trades land where.

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.