Owl_SPY
Overnight hold after a weak session
Every figure on this page is a hypothetical backtest: not a live account, and not a prediction.Read the full disclosure
About the figures on this page
Any results shown are hypothetical backtests, not the performance of a live account. Backtested results have inherent limitations: they are computed with hindsight over a chosen period, they cannot fully model costs, slippage or liquidity, and they do not represent actual trading.
Nothing here is a prediction, a promise, or an indication of future results. Trading involves risk, including total loss of capital. This material is informational only. StratoQuant is not a licensed investment adviser, does not provide personalised investment advice, and does not manage money or access client accounts. See the full disclaimer.
Owl_SPY
Summary
Hypothetical backtest
- CAGR
- 8.46%
- Max drawdown
- −29.3%
- Test period
- 2005 to 2026
- Action needed
- Daily
Rules
The excerpt will be published here. Request the code below to read the full rules and program now.
Python
The excerpt will be published here. Request the code below to read the full rules and program now.
Holds a broad equity index fund overnight only, and only after a weak day: in at the close, out at the next open, never during the trading day. Across the test it grew more slowly than holding the fund, with about half its volatility and half its worst fall, and lost money in two calendar years of twenty-two. The gain per night is tiny, so it only works with closing and opening auction orders: one basis point of slippage per fill takes away about three and a half points a year.
- Direction
- Long only, overnight only
- Universe
- One broad US large-cap index fund, fixed
- Holding period
- One night
- Positions held
- One, about two nights in three
- Action needed
- Daily: decide minutes before the close, buy at the close, sell at the next open
- Tooling
- Python, end-of-day data, auction orders
| Test period | 2005-01-03 to 2026-09-30 (21.7y) |
|---|---|
| CAGR | 8.46% |
| Max drawdown | −29.3% |
| Volatility | 9.6% |
| Sharpe | 0.90 |
| Trades per year | 167 |
Simulated results only; no live account results appear anywhere in this record. A single continuously listed exchange-traded proxy for a broad US large-capitalisation index, prices as traded, with cash dividends credited only for nights held into an ex-dividend date. The test runs from January 2005. Before 2010 the minimum price step was coarse for moves this small, so the 2005 to 2009 years, which hold the deepest fall (2008), carry more uncertainty than the rest. Entries fill at the official closing price and exits at the official opening price, as auction orders do; per-share commission with a per-order minimum; no slippage. Every result is therefore highly sensitive to execution: one basis point of slippage per fill cuts CAGR to 4.9%, two basis points to 1.5%. The rule uses the session’s final high, low and close, which in practice must be read minutes before the close. Compounding single account, whole shares, no borrowing, cash earns nothing. CAGR and volatility from the daily series; maximum drawdown measured on daily marks (−29.3%); the monthly curve below shows a shallower −27.2%. Sharpe is on a zero-rate basis. The threshold was not tuned to this test (results are flat across a wide band around it), but the system was selected for publication from a larger set of researched systems. Benchmark (the same fund held throughout, dividends included): CAGR 10.86%, max drawdown −55.2%, volatility 18.9%.
How it moved with the other systems
Correlation of monthly changes, and how often both rose or fell in the same month, over the months both tests share.
| System | Correlation | Same direction | Period |
|---|---|---|---|
| Harrier_SPY | 0.20 | 60% | Jan 2005 to Sep 2026 |
| Crane_ETF | 0.26 | 64% | Jan 2005 to Sep 2026 |
| Stork_NDX | 0.18 | 63% | Jan 2005 to Sep 2026 |
Hypothetical backtests. These are facts about the past, not a suggestion to run any systems together.
Free: Python code and plain-language rules
Request one, get both. Harrier_SPY and Owl_SPY come together, each with its rules in plain language and the Python code that runs them.
Leave your email and we will send both. You run them on market data you are licensed to use, and every decision stays yours.
Thank you. We will email you the code and rules shortly. If nothing arrives within one working day, write to hello@stratoquant.com.
That did not go through. Please check your email address and try again.
We use your email only to send the code and rules and, if you tick the box, the four follow-up emails. See the privacy notice.