Strategies · Classic rules

Overbought, oversold

The pattern behind many Composer community strategies, without the leverage: step aside when the Nasdaq-100 is very overbought, buy it when it has been sold hard, and otherwise follow the trend.

Pre-registered Sep 26, 2026, before it was tested. Results computed Sep 26, 2026.

Open in the builder

Hypothetical backtest; not investment advice; paper trading only.Past results, real or simulated, do not predict future ones. ARCANE places no live orders from this page.

The rule

Trades every trading day

If the 10-day RSI of QQQ is above 80, hold T-bills (BIL); if it is below 30, hold QQQ; otherwise hold QQQ while the S&P 500 is above its 200-day average and 7–10 year Treasuries (IEF) when it is below. Checked daily.

Where the numbers came from. RSI above 80 and below 30 are the textbook extremes, 10 days is the community’s usual window, 200 days the usual trend line. Nothing was fitted; the leveraged funds the community uses were left out.

Hold what is below
If the 10-day RSI of QQQ is greater than 80
then: Hold BIL
Otherwise
If the 10-day RSI of QQQ is less than 30
then: Hold QQQ
Otherwise
If the price of SPY is greater than the 200-day moving average of SPY
then: Hold QQQ
Otherwise
Hold IEF

As text

symphony "Overbought, oversold"
description "Step aside when overbought, buy hard dips, otherwise follow the trend."
rebalance daily
benchmark SPY
weight equal
  if relative-strength-index(QQQ, 10) > 80
    asset BIL
  else
    if relative-strength-index(QQQ, 10) < 30
      asset QQQ
    else
      if current-price(SPY) > moving-average-price(SPY, 200)
        asset QQQ
      else
        asset IEF

How it would have done

Jun 3, 2008 to Sep 25, 2026, against SPY

The whole history

258 trades

A year
+13.4%
Annualized return
In all
+893%
Jun 3, 2008 to Sep 25, 2026
Sharpe
0.81
Return per unit of risk
Worst fall
−31%
Max drawdown
Swings
17%
Volatility, a year
Beta
0.51
To the benchmark

Each decision reads closing prices up to that day and trades at the next day’s close. It trades toward its targets every day. Costs: 5 bp of slippage on every trade and 0.3 bp of fees on sales. Uninvested cash earns nothing. Hypothetical: no real orders were placed.

$0$2.50$5$7.50$10200920122015201820212024Held out

Growth of $1
SPY, held
Held-out period

−40%−20%0%

What it held: QQQ 79% of the time; IEF 18% of the time; BIL 3% of the time.

In sample and the sealed holdout

The holdout was set before the preset was tested, and opened once

Measure	In sample Jun 3, 2008 – Sep 24, 2021	Sealed holdout Sep 27, 2021 – Sep 25, 2026
Annualized return	+13.2%	+13.9%
Sharpe ratio	0.82	0.79
Sortino ratio	1.15	1.16
Max drawdown	27.7%	30.9%
Calmar ratio	0.48	0.45
Volatility	16.8%	18.7%
Alpha against the benchmark	+8.2%	+3.9%
Beta	0.43	0.78
Correlation	0.53	0.72
Winning days	54%	54%
Turnover a year	13.9×	14.5×
Chance the Sharpe is above zero	100%	96%

Is it real?

Checked on the in-sample window only

Overfitting: Likely overfit
No clear edge over its benchmark

The result depends on the exact numbers chosen, or does not beat what luck would produce.

After counting 11 backtests of this idea, the out-of-sample Sharpe still clears the bar luck would set (deflated Sharpe 0.99).
The best version in one half of history usually lands in the bottom half of the other (probability of overfitting 54%).
Against SPY the difference (+0.9% a year) is within what luck produces (57% chance it truly beats it).

Backtests counted	11
Deflated Sharpe ratio (0.95 is the house bar)	0.99
Probability of backtest overfitting (20% is the house ceiling)	54%
Walk-forward Sharpe (choosing, then testing)	0.92
Walk-forward Sharpe with costs doubled	0.85
Extra return a year against SPY	+0.9%

The nudged versions (10)

Change	Sharpe
10-day RSI of QQQ: window → 8	0.80
10-day RSI of QQQ: window → 13	0.89
threshold 80 on 10-day RSI of QQQ → 72	0.65
threshold 80 on 10-day RSI of QQQ → 88	0.85
10-day RSI of QQQ: window → 8	0.80
10-day RSI of QQQ: window → 13	0.89
threshold 30 on 10-day RSI of QQQ → 27	0.92
threshold 30 on 10-day RSI of QQQ → 33	0.96
200-day moving average of SPY: window → 150	0.85
200-day moving average of SPY: window → 250	0.77

The house backtest gate

It does not pass. A strategy needs every one of these before ARCANE would paper-trade it.

Check	Value	Needs	Result
Evaluation ran without problems	clean	no problems	Passed
Report belongs to this pre-registration	overbought-oversold	overbought-oversold	Passed
Evaluated only inside the pre-registered data window (holdout untouched)	2008-06-03..2021-09-24	within 2008-06-02..2021-09-24	Passed
Walk-forward folds	10.000	>= 3	Passed
Out-of-sample observations	2520.000	>= 252	Passed
Deflated Sharpe ratio (11 trials counted)	0.992	>= 0.95	Passed
Probability of backtest overfitting	0.538	<= 0.2	Failed
Out-of-sample Sharpe (annual, after costs)	0.919	>= 0.5	Passed
Out-of-sample maximum drawdown	0.278	<= 0.25	Failed
Sharpe with doubled costs	0.853	> 0 (needs a turnover series)	Passed
Sealed holdout: positive and not clearly worse than the backtest	mean 0.000586, consistent	mean > 0 and consistent	Passed

The pre-registration

Written before the backtest; changing any of it makes a new variant

Claim: The overbought/oversold rotation over QQQ, BIL and IEF beats SPY buy-and-hold on Sharpe after 5 bp costs over the sealed holdout.
Why the edge should exist: Short-term extremes in a large index partly reverse as liquidity providers are paid to absorb one-sided flow; the trend filter keeps the dip-buying out of long declines.
Where it comes from: liquidity provision
Who is on the other side: Traders who chase short bursts and sell into sharp declines.
Why it is not arbitraged away: Short-horizon reversal is crowded and thin; what is left is small, noisy and costly to trade, which is why it is tested here without leverage.
Universe: BIL, IEF, QQQ, SPY
In-sample window: Jun 2, 2008 to Sep 24, 2021
Sealed holdout: Sep 27, 2021 to Sep 25, 2026
Evaluation: walk-forward, on sharpe, against SPY buy and hold; 13 variants planned
Costs: 5 bp slippage a trade, no commission
Capacity: $1,000,000: QQQ, BIL and IEF are deep; the rule can trade often, so costs, not size, bind.
What stops it: Paper Sharpe significantly below the backtest after 60 sessions. Costs in paper trading exceed twice the modelled 5 bp.

The data

Prices: Yahoo Finance (unofficial chart API): adjusted daily closes for BIL, IEF, QQQ, SPY. Unofficial and not licensed for redistribution. A licensed vendor (Alpaca, Tiingo or Polygon) is a founder decision before the builder is sold.