An input delay of 21 trading days cut the 5% tail forecast gain from 16.2% to 7.0% in revised US market data for 2000–2025
We tested how delayed inputs change a daily US market loss forecast. The main test used 6,539 days in 2000–2025 from the revised 202608 Kenneth French data file. A fixed method scaled historical returns by an estimate of current volatility. With prior-close inputs, it reduced mean 5% quantile forecast loss by 16.2% against a 1,260-day historical baseline. With 21 extra trading days of delay, the gain was 7.0%. The absolute score gain fell from 2.502 to 1.091 basis points. The paired 95% block bootstrap interval for this decrease was 0.717 to 2.322 basis points. Against a 252-day baseline, the delayed gain had an interval that included zero. In 2016–2025, the delayed model had no clear 5% score gain. We also used 13 dated archive files to forecast 273 later days with only archived inputs. In that small selected sample, the model had 26.3% higher 5% forecast loss than the five-year baseline. A post-plan check on the same dates found little effect from revisions at the same input cutoff. Prior-close inputs reduced loss but did not make the model beat the baseline on those dates. The main test assumes access to earlier returns and is not a full historical availability test. Archive predictors are point-in-time files; outcome labels use the revised source. No direction forecast or trading strategy was tested. The package preserves code, source hashes, forecasts, resampling draws, figures, negative results and repeat instructions. Raw source files are retained separately under provider rights.
market risk · quantile forecast · input delay · filtered historical simulation · data vintage · out-of-sample test
References
7- usesdoi:10.21314/jor.1998.001
The forecast uses the established method of volatility updates in historical simulation.
- doi:10.1198/073500104000000370
Prior work on conditional quantile forecasts gives the risk forecasting context.
- doi:10.2307/2527341
Prior work distinguishes overall interval coverage from conditional coverage.
- usesdoi:10.1016/0304-405X(93)90023-5
The daily market series comes from the Kenneth French Data Library. Exact source URLs and vintage hashes are preserved separately.
- doi:10.1111/jofi.12513
Prior work studies the portfolio use of volatility estimates. This study tests risk forecast scores and makes no profit claim.
- doi:10.1016/j.jfineco.2020.04.015
Prior work gives context on out-of-sample tests of volatility-managed portfolios.
- In 2019–2026, 12-month factor momentum on 23 public equity factors gave no clear net gain over equal factor weights (gain +0.06%/month, 95% CI −0.30 to +0.44) sci:sha256:605a364268f178775b231b39cab8d4bba7e8cb3a0f6c07b7fc8934fa835b4d2e
This earlier SCI study tested factor return forecasts and data revisions at a monthly horizon. It motivated a separate daily risk test with explicit input cutoffs.
Cited by
11 publicationsPublic catalog entries on Sciscape.
- usesExtra weight on past falls cut 5% Nasdaq Composite tail forecast loss by 1.1% in 2015–2025 with a two-day input cutoff
The study gives the established symmetric FHS comparator and motivates explicit checks of input delay. We test a different series and a fixed extra response to past falls.
- An asymmetric volatility model cut 5% Nasdaq daily tail forecast loss by 2.0% against a symmetric model in a 2015–2025 test with dated inputs
Its daily tail-risk test motivates explicit input dates, historical baselines and score limits. We use a different index and a dense ALFRED vintage reconstruction.
- Online tail updates gave no clear loss reduction for daily Nasdaq 5% forecasts in a 2018–2025 ALFRED vintage test
This prior tail forecast test reports losses with delayed US market inputs and sparse archives. We test daily dated inputs for a different index and add delayed online feedback.
- supportsDownside shares reduced weekly Nasdaq Composite variance forecast loss in 505 dated tests from 2016 to 2025
Its delay test motivates careful checks of the information available before a forecast.
- VXN reduced daily 5% quantile loss by 4.8% in a 2017–2026 Nasdaq-100 archive test, with 60% of the gain from March 2020
Related daily loss-forecast study of input delay and distribution assumptions in US market returns. We use a different market series and historical data route.
- A 252-session trend input gave no clear monthly Nasdaq Composite return forecast gain in a 2018–2025 archive test
Related evidence on the effect of stale market inputs; motivates retaining source-age audits.
- Five-year return means gave no clear forecast gain with annual US market archives in 2005–2025
Related evidence on delayed inputs in daily market tail forecasts. This test instead forecasts the monthly market mean.
- usesTurn-of-month mean forecasts gave no clear error reduction for US market returns in 2016–2024
We reuse its 13 retained archived daily market files and source provenance. The mean forecast and trading simulation are separate new code.
- Recent return signs reduced daily Nasdaq risk forecast loss with dated inputs in 2015–2025
- usesVolatility scaling cut lower 5% return forecast loss by 8.8% in a 2015–2025 Nasdaq Composite vintage test
We extend its fixed historical-simulation comparison and availability audit to Nasdaq Composite returns with daily ALFRED real-time intervals. The source series, time rule and test period differ.
- supportsA fixed move toward long-run risk cut loss against held recent risk in dated Nasdaq forecasts with 21 extra sessions of input delay, 2018–2025
We extend its input-delay question from revised US market tail forecasts to dated Nasdaq second-moment forecasts with an explicit gap rule.