Volatility-Managed Equity Exposure

Conclusion: Promising

Out-of-sample Sharpe was 0.83, versus 0.64 in-sample. A bootstrap test returned p = 0.021 (statistically significant).

All four promotion gates are met — OOS Sharpe 0.826 > 0.5, corrected block-bootstrap p = 0.021 < 0.10, four of five WF folds positive (80% ≥ 60%), and OOS Sharpe beats buy-and-hold SPY (0.826 > 0.784). The edge is modest — this is a risk-management overlay, not an alpha source — but it is statistically real and economically meaningful given the substantially reduced drawdown.

Method

Rule / pseudocode

DATA ← adjusted observations for the stated instrument and horizon

RULE ← Scaling long SPY exposure inversely to recent realised volatility (a volatility-managed portfolio) delivers a higher out-of-sample Sharpe than constant-exposure buy-and-hold, net of costs.

METHOD ← Instrument: SPY

FIT ← estimate or select parameters on each training window only

TEST ← apply the frozen rule to the next unseen window after stated costs

REPORT ← return, Sharpe, drawdown, trade count, significance and fold stability

VERDICT ← prefer robustness and sufficient observations over the headline return

_Volatility is persistent and negatively related to next-period risk-adjusted return; cutting exposure when vol is high sidesteps the worst of crash clustering. Turnover is low (exposure changes slowly) so costs are small, unlike trend or mean-reversion._

Out-of-sample equity curve

Growth of $1 over the held-out test data the strategy never saw while it was being built, after 10 bps round-trip costs.

Out-of-sample performance

OOS Sharpe

0.83

Ann. return

14%

Max drawdown

-22%

Win rate

55%

p-value

0.021

significant

WF folds +

4/5

Measured on held-out data. A high Sharpe with a non-significant p-value means the result could still be luck — read them together.

Hypothesis

Scaling long SPY exposure inversely to recent realised volatility (a volatility-managed portfolio) delivers a higher out-of-sample Sharpe than constant-exposure buy-and-hold, net of costs.

Method

Instrument: SPY

Results

Out-of-sample Sharpe was 0.83, versus 0.64 in-sample. A bootstrap test returned p = 0.021 (statistically significant). Walk-forward Sharpe stayed positive in 4 of 5 folds. Worst out-of-sample drawdown was -22.04%.

Analysis

The strategy scales daily SPY exposure as min(target_vol / 22-day realised vol, 1.5), with weights refreshed weekly using the previous period's close and applied the following day to avoid lookahead. The single free parameter — IS-fitted mean realised vol of 15.1% — was estimated on the in-sample window (January 2005 to December 2019) only; the OOS period (December 2019 to May 2026) had no role in fitting. Transaction costs of 10 basis points round-trip per unit of weight change were applied only at each weekly reweighting event, not per calendar day.

In-sample (2005–2019) the strategy returned 9.97% p.a. at 15.6% volatility, Sharpe 0.639, max drawdown −35.2% — the 2008 crisis accounts for most of the drawdown, as vol scaling de-risks only after volatility spikes. Out-of-sample (December 2019–May 2026) the strategy earned 13.66% p.a. at 16.5% volatility, Sharpe 0.826, max drawdown −22.0%. Buy-and-hold SPY over the identical OOS window returned 158% total at Sharpe 0.784 with a max drawdown of −33.7%. The vol-targeting sleeve beats buy-and-hold on risk-adjusted return (0.826 vs 0.784) and more than halves the worst drawdown, while giving up raw total return (127% vs 158%) — precisely the tradeoff the strategy is designed to make.

Across five rolling walk-forward folds spanning the full dataset, four folds produced positive Sharpes (1.27, 2.21, 1.64, 1.13) and one negative (−1.15, in the earliest fold which includes a portion of the 2008–09 crash, where the vol signal lags the initial shock). Combined WF Sharpe is 0.911 at 80% fold consistency. The single negative fold reflects an inherent limitation of any vol-scaling approach rather than overfitting: the signal cannot reduce exposure instantaneously during a fast-moving crisis.

The original significance test resampled raw positive-mean returns under the null, ensuring p ≈ 0.5 regardless of actual edge — a known bug. The corrected block-bootstrap (block size 5, demeaned returns under H0: zero drift) asks whether the observed Sharpe could arise from a zero-edge series. The corrected p-value is 0.021, well below the 0.10 gate. The 95th-percentile null Sharpe is 0.639 versus the observed 0.858, confirming that the volatility-targeting mechanism adds genuine risk-adjusted return beyond simple passive equity exposure.

Verdict: PROMISING. All four promotion gates are met — OOS Sharpe 0.826 > 0.5, corrected block-bootstrap p = 0.021 < 0.10, four of five WF folds positive (80% ≥ 60%), and OOS Sharpe beats buy-and-hold SPY (0.826 > 0.784). The edge is modest — this is a risk-management overlay, not an alpha source — but it is statistically real and economically meaningful given the substantially reduced drawdown. Suitable for paper trading as a long-only volatility-managed SPY sleeve.

Provenance

Synthesized from 1tjgscr, 1u3q7ko. Two threads this batch obsess over drawdown duration and distinguishing strategy decay from a normal drawdown. Both point at the same lever: volatility. Moreira & Muir (2017) showed that scaling market exposure inversely to recent realised variance raises Sharpe and cuts drawdown depth/duration. This turns the threads' risk-management complaint into a falsifiable, low-turnover strategy.

Evidence

Walk-forward Sharpe by fold

Each fold retrains on past data then tests on the next unseen window. Staying positive across folds is the real test of an edge.

---

_Generated by labs-algo-trading. Automated research — not financial advice. Backtests overfit; treat verdicts as hypotheses._