Empirical tails gave no clear 5% joint risk forecast gain for Nasdaq in a 2018–2025 dated-input test
We tested daily loss limits and mean tail losses for the Nasdaq Composite on 2,011 sessions in 2018–2025. Each forecast used an ALFRED version dated before the session at which its return interval started. Four fixed models used historical returns, normal tails, Student-t tails, or empirical standardized tails. The primary 5% joint score difference for empirical tails minus normal tails was -0.03615. Its 95% paired 21-session block interval was -0.08147 to +0.00550. This did not give a clear gain. The secondary 1% difference was -0.28784, with interval -0.57724 to -0.03748. That interval included zero after 2020 was removed. Empirical breach rates were closer to their targets, but quantile-only loss gains remained uncertain. The source has a known reporting gap. Fresh-input controls retained the main conclusions. The package preserves the local plan, source requests and hashes, code, 22,128 forecasts, all scores, 15,000 resampling rows, figures, and repeat checks. This retrospective index test does not establish return-direction skill or trading profit.
expected shortfall · value at risk · point-in-time data · Nasdaq Composite · forecast evaluation · filtered historical simulation · negative result
References
6- usesdoi:10.21314/jor.1998.001
The fixed volatility-scaled historical forecasts follow the established volatility updating method.
- usesdoi:10.1214/16-AOS1439
Joint elicitability provides the basis for quantile and expected-shortfall evaluation.
- usesdoi:10.1016/j.jeconom.2018.10.008
We use its FZ0 joint loss equation, checked against arXiv:1707.05108v1.
- usesVolatility scaling cut lower 5% return forecast loss by 8.8% in a 2015–2025 Nasdaq Composite vintage test sci:sha256:a8d414223facf768878a01e0307db214555fb5c69e6f24f643b076010c981d6d
We reuse its retained ALFRED source and compare a related fixed volatility-scaling method, with a new joint risk target and timing rule.
- Extra weight on past falls cut 5% Nasdaq Composite tail forecast loss by 1.1% in 2015–2025 with a two-day input cutoff sci:sha256:31b0a8b00f45fe3341bbdc1456bad373997c39805ebd67fe6d01a9378374576a
Prior dated Nasdaq quantile tests motivate checking mean tail loss as well as the loss limit.
- usesOnline tail updates gave no clear loss reduction for daily Nasdaq 5% forecasts in a 2018–2025 ALFRED vintage test sci:sha256:4362565a5793464cf2fc7a3446df0da0bd2251491d46810773461ceb5dc1fedb
Its source calendar audit identified the Good Friday row that our fixed plan removes from all inputs and outcomes.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesA fixed Student t shape gave no clear weekly Nasdaq distribution score gain over normal at equal variance in 2016–2025
We reuse MIT interval parsing and EWMA initialization. The earlier study tested daily joint tail scores; this study tests full weekly return distributions.
- usesFitted Student-t models cut daily Nasdaq Composite log loss in a 2018–2025 dated-input test
We adapt its source reconstruction, calendar filter, EWMA calculation, and conservative input cutoff. We extend its tail-score benchmark to full densities and monthly distribution fits.