SPB Git
spboucher.ai

spb/wp41_uqo

Public

WP41: held-out, ablation, final-period and copula results; full manuscript; phase 6-8 reports

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Simon-Pierre Boucher committed (Nov 3, 2026) parent e976cfd Browse files

218 changed files +18,151 −1,484

218 changed files
Load all 218 files Unified Split
modified README.md +15 −3
@@ -36,9 +36,21 @@ calibration under stress; H4 coherent joint paths (low-rank Gaussian copula over
36 independent marginals; H5 (exploratory) economic content after assumed costs. Targets are return distributions 36 independent marginals; H5 (exploratory) economic content after assumed costs. Targets are return distributions
37 (per-minute Student-t marginals, cumulative returns, realised variance, tail events), never price levels. 37 (per-minute Student-t marginals, cumulative returns, realised variance, tail events), never price levels.
38 38
39−## 🏆 Headline results 39+## 🏆 Headline results (scaled CRPS of the cumulative return; 2024 development test, 986 468 origins; 2025 untouched, 1 008 425)
40− 40+
41−Filled from `results/metrics/main`, `ablations`, `final` by `scripts/evaluate.py` — see `docs/PHASE_7_REPORT.md`. 41+| Object | 2024 (H=5 / H=60) | 2025 (H=5 / H=60) |
42+|---|---|---|
43+| Random walk (trailing scale) | 0.4699 / 0.4347 | 0.4759 / 0.4409 |
44+| HAR-RV (minute scale, Student-t) | 0.4615 / **0.4166** | 0.4674 / **0.4232** |
45+| MRPT-S multi-resolution (3.5 M params) | **0.4531** / 0.4187 | **0.4594** / 0.4256 |
46+| MRPT-S one-minute only | 0.4533 / 0.4192 | 0.4594 / 0.4257 |
47+| MRPT-B (20.7 M params) | 0.4532 / 0.4189 | 0.4595 / 0.4257 |
48+| H1 multi-res vs 1m-only | −0.04 % / −0.10 % (t −5) | +0.01 % / +0.01 % (Holm p ≥ 0.36) → **rejected in substance** |
49+| H2 zero-shot on 60 unseen instruments vs random walk | −4.1 % / −4.3 % (pooled −4.05 % / −4.27 %) → **no transfer penalty** | same in 2025 |
50+| H3 regime conditioning | +0.02 … +0.03 % CRPS, identical stress-day coverage → **rejected** | |
51+| H4 copula paths vs independent | energy −0.003 %, variogram −0.002 % (CIs ∋ 0) → **rejected** | |
52+| H5 economic content | corr(median, realised) ≤ 0.025; net 0.0 bp [−1.7, +1.8] at 1 bp cost → **rejected** | |
53+| Time-only inputs (no returns/volumes) | +0.2 … +0.3 % vs full model: ≈ 90 % of the skill is trailing volatility + clock | |
42 54
43 ## 🗂 Repository layout 55 ## 🗂 Repository layout
44 56
modified data/manifests/universe_synthetic.json +8 −23
@@ -14,7 +14,7 @@
14 "delisted": false, 14 "delisted": false,
15 "status": "active", 15 "status": "active",
16 "first_date": "2024-01-02", 16 "first_date": "2024-01-02",
17− "last_date": "2024-05-31" 17+ "last_date": "2024-04-30"
18 }, 18 },
19 { 19 {
20 "symbol": "SYN01", 20 "symbol": "SYN01",
@@ -27,12 +27,12 @@
27 "delisted": false, 27 "delisted": false,
28 "status": "active", 28 "status": "active",
29 "first_date": "2024-01-02", 29 "first_date": "2024-01-02",
30− "last_date": "2024-05-31" 30+ "last_date": "2024-04-30"
31 }, 31 },
32 { 32 {
33 "symbol": "SYN02", 33 "symbol": "SYN02",
34 "asset_class": "stock", 34 "asset_class": "stock",
35− "role": "pooled", 35+ "role": "heldout_instrument",
36 "adv_usd": 30000000.0, 36 "adv_usd": 30000000.0,
37 "adv_tercile": "high", 37 "adv_tercile": "high",
38 "vol_tercile": "low", 38 "vol_tercile": "low",
@@ -40,32 +40,17 @@
40 "delisted": false, 40 "delisted": false,
41 "status": "active", 41 "status": "active",
42 "first_date": "2024-01-02", 42 "first_date": "2024-01-02",
43− "last_date": "2024-05-31" 43+ "last_date": "2024-04-30"
44− },
45− {
46− "symbol": "SYN03",
47− "asset_class": "etf",
48− "role": "heldout_instrument",
49− "adv_usd": 40000000.0,
50− "adv_tercile": "low",
51− "vol_tercile": "mid",
52− "group": "synthetic",
53− "delisted": false,
54− "status": "active",
55− "first_date": "2024-01-02",
56− "last_date": "2024-05-31"
57 } 44 }
58 ], 45 ],
59 "pilot_symbols": [ 46 "pilot_symbols": [
60 "SYN00", 47 "SYN00",
61− "SYN01", 48+ "SYN01"
62− "SYN02"
63 ], 49 ],
64 "feasibility_symbols": [ 50 "feasibility_symbols": [
65 "SYN00", 51 "SYN00",
66− "SYN01", 52+ "SYN01"
67− "SYN02"
68 ], 53 ],
69− "n_selected": 4, 54+ "n_selected": 3,
70− "sha256": "73171c0475e7985c635dacb931910fe08ca977a507057a3bcee01fa345817a39" 55+ "sha256": "98e37bc71e1aaf0028661ccd61d6b4509f869d83d84f128f19a7506e2c3a0980"
71 } 56 }
\ No newline at end of file
added docs/PHASE_6_REPORT.md +34 −0
@@ -0,0 +1,34 @@
1+# PHASE 6 — Joint paths (Stage B) — 2026-11-03
2+
3+## What was run (`configs/ablations.yaml`, model `mrpt_small_copula`, M3U96b)
4+Stage A: MRPT-S with the Student-t head, 4 000 steps (same budget and seed as the ablation reference `mrpt_small_b4k`),
5+best validation NLL −6.4509. Stage B: low-rank (k = 4) + diagonal Gaussian copula head over the 60 future minutes,
6+trained for 600 steps on the copula log-likelihood of the PITs with the marginals frozen (64 s); best validation copula
7+NLL −2.5e-5 per observed dimension, i.e. **indistinguishable from independence** (the likelihood at R = I is exactly 0).
8+Prediction on the 986 468 development-test origins with 200 correlated draws (`joint`) and with 200 independent draws
9+of the same marginals (`indep`); path scores computed per origin and bootstrapped by trading day
10+(`results/metrics/ablations/path_scores_mrpt_small_copula_test.json`).
11+
12+## Results (H4)
13+| Score (per origin, 60-minute paths) | Joint (copula) | Independent | Joint − independent (95 % CI) |
14+|---|---|---|---|
15+| Energy score | 3.4903e-3 | 3.4904e-3 | −9.2e-8 [−2.3e-7, +5.0e-8] (−0.003 %) |
16+| Variogram score (p = 0.5) | 1.51132e-4 | 1.51136e-4 | −3.8e-9 [−1.0e-8, +2.2e-9] (−0.002 %) |
17+| CRPS of maximum drawdown | 1.4632e-3 | 1.4633e-3 | −1.8e-7 [−6.2e-7, +2.8e-7] (−0.01 %) |
18+| CRPS of realised variance | 1.1881e-5 | 1.1869e-5 | +1.2e-8 [−2.8e-9, +3.0e-8] (+0.10 %) |
19+Cumulative-return CRPS at H = 60: joint 0.4194, independent 0.4191 (scaled), i.e. no gain either.
20+
21+**H4 is rejected**: once the Student-t marginals are conditioned on the 390-bar history, the probability-integral
22+transforms of consecutive future minutes are essentially uncorrelated, the learned factor loadings stay near zero,
23+and coherent sampling changes no path score by more than a hundredth of a percent. Diffusion decoders were therefore
24+not attempted (the brief asked for them only if a dependence-aware decoder demonstrably helped).
25+
26+## Validation of the machinery (so that the null is not an implementation artifact)
27+- `tests/numerical/test_copula.py`: sampling reproduces the marginals; the sampled correlation matches the implied
28+ correlation under strong loadings; the copula NLL is minimised at independence for independent data; PIT round trip.
29+- `tests/numerical/test_scores.py::test_energy_score_prefers_true_dependence`: the energy score does reward correct
30+ dependence when it exists.
31+- The synthetic smoke run (independent t returns by construction) also learned ≈ 0 correlation, as it should.
32+
33+## Go / no-go
34+Stage B remains in the package as a documented null; the paper reports it in the ablation section.
added docs/PHASE_7_REPORT.md +38 −0
@@ -0,0 +1,38 @@
1+# PHASE 7 — Inference, calibration, subgroups, falsification, economics (2026-11-03)
2+
3+## Runs (M3U96b, `rest_chain`, 23:06 → ~02:50 ET; all exit 0)
4+| Run | Config / flags | Origins | Wall |
5+|---|---|---|---|
6+| Held-out 2024 | `main.yaml --heldout` (baselines refit on the 60 instruments' own training histories; neural checkpoints from the main study, zero-shot) | 250 322 | ≈ 15 min |
7+| Ablations 2024 | `ablations.yaml` (15 Small variants × 4 000 steps + reference + 2 seeds + copula) | 986 468 | ≈ 6 h 10 |
8+| Final 2025 pooled | `final.yaml --eval-split final --models-from results/metrics/main` | 1 008 425 | ≈ 1 h 25 |
9+| Final 2025 held-out | `final.yaml --eval-split final --heldout --models-from …` | 254 334 | ≈ 25 min |
10+Post-processing on the laptop: `scripts/pairwise.py` (DM-HAC, day-cluster bootstrap, Holm/BH, subgroups, stress-day
11+calibration) for `main/test`, `main/test_heldout`, `final/final`, `final/final_heldout`, `ablations/test`;
12+`scripts/economic.py` on the node (H5); `scripts/evaluate.py`, `make_figures.py`, `make_static_tables.py`.
13+
14+## Hypothesis verdicts (relative CRPS of the cumulative return; Holm-adjusted DM p-values)
15+| Hypothesis | Evidence | Verdict |
16+|---|---|---|
17+| H1 multi-resolution > 1m-only (equal params) | 2024 dev: −0.04 % (H5) … −0.10 % (H60), t −4.5 to −5.3; 4k-step ablation: −0.12 … −0.16 %; **2025 final: +0.01 … +0.03 %, p ≥ 0.36; 2025 held-out: p = 1** | statistically detectable in sample, economically nil, absent out of sample → **rejected in substance** |
18+| H2 transfer to unseen instruments | held-out vs random walk −4.1 … −4.3 % (pooled −4.05 … −4.27 %); beats GARCH-t (−0.9 … −1.6 %), EWMA, TCN, MLP; ties HAR/ridge/LightGBM from 30 min (p = 1); 2025 held-out identical | **supported** (no transfer penalty; parity with refitted HAR beyond 15 min) |
19+| H3 regime conditioning improves stress calibration | no-regime vs stats-regime: +0.02 … +0.03 % CRPS, p > 0.3 at 30/60; stress-day coverage identical (0.57/0.85/0.97 vs 0.57/0.85/0.97) | **rejected** (learned codebook: −0.03 … −0.15 %, small) |
20+| H4 coherent paths > independent marginals | copula NLL → 0 (independence); energy −0.003 %, variogram −0.002 %, MDD CRPS −0.01 %, RV CRPS +0.10 %, all CIs contain 0 | **rejected** |
21+| H5 economic content | median–realised correlation 0.003–0.025; net return at 1 bp cost 0.0 bp [−1.7, +1.8]; traded share < 4 % above 2.5 bp | **rejected** |
22+Supplementary findings: model size irrelevant (Base vs Small p ≥ 0.1 everywhere); neural best at H = 5 (−1.2 … −1.9 % vs HAR),
23+HAR/ridge/LightGBM best at H ≥ 15 (+0.2 … +0.7 % against MRPT-S, 2024 and 2025); time-only inputs recover ≈ 90 % of the gain
24+over the random walk; shuffled minutes cost +0.13 … +0.21 %; ticker identity hurts (+0.05 … +0.31 %).
25+
26+## Calibration
27+Neural + GARCH/EWMA: nominal at 95 %, over-dispersed centre (50 % coverage 0.55–0.59). HAR/ridge/LightGBM: nominal from
28+15 min, under-dispersed at 5 min (50 % coverage 0.37). 2025 stress days: HAR 0.46/0.77/0.95 vs MRPT-S 0.55/0.84/0.95.
29+
30+## Multiplicity and dependence
31+All intervals are day-cluster bootstraps (251–252 clusters per year); DM lags = overlap × instruments (240 … 960);
32+Holm and BH columns in every `pairwise_*.csv`. Overlapping origins are never counted as independent.
33+
34+## Artifacts
35+`results/metrics/{main,final,ablations}/pairwise_*.csv`, `subgroups_*.csv`, `stress_calibration_*.csv`, `economic_test.csv`,
36+`path_scores_mrpt_small_copula_test.json`, leaderboards and comparisons per split; `paper/tables/*.tex` (53), `paper/figures/*` (27),
37+`paper/results_macros.tex` (3 468 macros). Gates: leakage (tests), numerical (tests + copula null reproduced on synthetic data),
38+provenance (dataset versions 17 375 / 5 162 in every run_meta) — PASS.
added docs/PHASE_8_REPORT.md +38 −0
@@ -0,0 +1,38 @@
1+# PHASE 8 — Paper (2026-11-03)
2+
3+## Manuscript
4+`paper/main.tex` → `paper/main.pdf` (latexmk, pdflatex + bibtex, style `aer`, UQO template: `newtxtext`/`newtxmath`,
5+title page macros). Sections: title page + abstract, introduction, hypotheses (5 formal statements), related work,
6+data, model, experimental design, results (pilot, development test 2024, calibration, subgroups, economics, zero-shot
7+and untouched 2025), ablations and falsification (incl. Stage-B paths), limitations, reproducibility and release,
8+conclusion, bibliography, appendices (baseline specifications, additional tables, universe). 36 pages.
9+
10+## Provenance of every number
11+- `paper/results_macros.tex`: 3 468 macros generated by `scripts/evaluate.py` from the saved leaderboards, comparisons
12+ and pairwise files (`\crps<exp><model>H<h>`, `\rel…`, `\pw<exp><model>vs<ref>H<h>`, `\tpw…`, `\ppw…`, coverage macros);
13+ digits in macro names are spelt out (`Hsixzero` = H = 60).
14+- `paper/tables/*.tex` (53 files): leaderboards (scaled CRPS, WIS, QLIKE, calibration) and relative-change tables for
15+ pilot / main / main held-out / ablations / final / final held-out, universe strata, groups and ticker lists, baseline
16+ specifications (from `configs/baselines.yaml`), measured benchmark (from `results/metrics/benchmark_M3U96b.json`).
17+- `paper/figures/*.pdf` (27): scores by horizon, coverage, PIT histograms, forest plots, training curves, drawn only
18+ from result files by `src/chronos_x/eval/figures.py` (Times/STIX, monochrome + one accent, panel labels).
19+- Hand-typed numbers in the prose (subgroups, ablation ranges, copula scores, economic results) are copied from
20+ `results/metrics/*/pairwise_*.csv`, `subgroups_*.csv`, `stress_calibration_*.csv`, `path_scores_*.json`, `economic_test.csv`
21+ and listed in `docs/PHASE_5..7_REPORT.md`.
22+- Bibliography: 54 entries, every DOI/arXiv id fetched and title-matched (`scripts/build_references.py`,
23+ `results/metrics/references_verification.json`); four entries without DOI are typed by hand (LightGBM, Holm, Sklar, López de Prado).
24+
25+## Claims → evidence (central ones)
26+| Claim | Evidence |
27+|---|---|
28+| All models beat the random walk by 3.5–4.9 % | `main/compare_crps_vs_naive_rw_test.csv`, `final/…_final.csv` |
29+| H1 effect ≤ 0.1 % in sample, absent in 2025 | `main/pairwise_test.csv`, `final/pairwise_final.csv`, `ablations/pairwise_test.csv` |
30+| HAR-RV better from 15 min, neural better at 5 min | same files, `tables/main_crps_scaled.tex`, `final_crps_scaled.tex` |
31+| Transfer without penalty | `main/pairwise_test_heldout.csv`, `final/pairwise_final_heldout.csv` |
32+| H3, H4, H5 nulls | `ablations/pairwise_test.csv`, `stress_calibration_test.csv`, `path_scores_mrpt_small_copula_test.json`, `main/economic_test.csv` |
33+| Time-only ≈ 90 % of the skill | `ablations/pairwise_test.csv` (mrpt_small_time_only vs mrpt_small_b4k; b4k vs naive_rw) |
34+
35+## Checks
36+- `latexmk` exit 0; 0 undefined references/citations after the final pass; 1 remaining overfull box (appendix table, cosmetic).
37+- Visual check of the results pages (figures and tables render; legends below panels).
38+- No proprietary bars, grids or per-origin returns appear in the manuscript, tables or figures (scores and tickers only).
modified paper/appendix/appendix.tex +4 −3
@@ -9,14 +9,15 @@
9 \IfFileExists{tables/main_qlike.tex}{\input{tables/main_qlike}}{} 9 \IfFileExists{tables/main_qlike.tex}{\input{tables/main_qlike}}{}
10 \IfFileExists{tables/main_compare_wis.tex}{\input{tables/main_compare_wis}}{} 10 \IfFileExists{tables/main_compare_wis.tex}{\input{tables/main_compare_wis}}{}
11 \IfFileExists{tables/main_compare_qlike.tex}{\input{tables/main_compare_qlike}}{} 11 \IfFileExists{tables/main_compare_qlike.tex}{\input{tables/main_compare_qlike}}{}
12−\IfFileExists{tables/main_crps_scaled_heldout.tex}{\input{tables/main_crps_scaled_heldout}}{}
13 \IfFileExists{tables/main_calibration_heldout.tex}{\input{tables/main_calibration_heldout}}{} 12 \IfFileExists{tables/main_calibration_heldout.tex}{\input{tables/main_calibration_heldout}}{}
14−\IfFileExists{tables/final_crps_scaled.tex}{\input{tables/final_crps_scaled}}{} 13+\IfFileExists{tables/final_compare_crps.tex}{\input{tables/final_compare_crps}}{}
15−\IfFileExists{tables/final_calibration.tex}{\input{tables/final_calibration}}{} 14+\IfFileExists{tables/final_crps_scaled_heldout.tex}{\input{tables/final_crps_scaled_heldout}}{}
15+\IfFileExists{tables/ablations_calibration.tex}{\input{tables/ablations_calibration}}{}
16 \IfFileExists{tables/ablations_wis.tex}{\input{tables/ablations_wis}}{} 16 \IfFileExists{tables/ablations_wis.tex}{\input{tables/ablations_wis}}{}
17 \IfFileExists{tables/pilot_crps_scaled.tex}{\input{tables/pilot_crps_scaled}}{} 17 \IfFileExists{tables/pilot_crps_scaled.tex}{\input{tables/pilot_crps_scaled}}{}
18 \IfFileExists{tables/pilot_calibration.tex}{\input{tables/pilot_calibration}}{} 18 \IfFileExists{tables/pilot_calibration.tex}{\input{tables/pilot_calibration}}{}
19 19
20+\IfFileExists{tables/universe_groups.tex}{\input{tables/universe_groups}}{}
20 \section{Universe} 21 \section{Universe}
21 \label{app:universe} 22 \label{app:universe}
22 \IfFileExists{tables/universe_strata.tex}{\input{tables/universe_strata}}{} 23 \IfFileExists{tables/universe_strata.tex}{\input{tables/universe_strata}}{}
added paper/figures/fig_ablations_calibration.pdf +0 −0

Binary file not shown.

added paper/figures/fig_ablations_calibration.png +0 −0

Binary file not shown.

added paper/figures/fig_ablations_crps.pdf +0 −0

Binary file not shown.

added paper/figures/fig_ablations_crps.png +0 −0

Binary file not shown.

added paper/figures/fig_ablations_forest.pdf +0 −0

Binary file not shown.

added paper/figures/fig_ablations_forest.png +0 −0

Binary file not shown.

added paper/figures/fig_ablations_pit.pdf +0 −0

Binary file not shown.

added paper/figures/fig_ablations_pit.png +0 −0

Binary file not shown.

added paper/figures/fig_ablations_wis.pdf +0 −0

Binary file not shown.

added paper/figures/fig_ablations_wis.png +0 −0

Binary file not shown.

added paper/figures/fig_final_calibration.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_calibration.png +0 −0

Binary file not shown.

added paper/figures/fig_final_calibration_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_calibration_heldout.png +0 −0

Binary file not shown.

added paper/figures/fig_final_crps.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_crps.png +0 −0

Binary file not shown.

added paper/figures/fig_final_crps_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_crps_heldout.png +0 −0

Binary file not shown.

added paper/figures/fig_final_forest.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_forest.png +0 −0

Binary file not shown.

added paper/figures/fig_final_forest_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_forest_heldout.png +0 −0

Binary file not shown.

added paper/figures/fig_final_pit.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_pit.png +0 −0

Binary file not shown.

added paper/figures/fig_final_pit_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_pit_heldout.png +0 −0

Binary file not shown.

added paper/figures/fig_final_wis.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_wis.png +0 −0

Binary file not shown.

added paper/figures/fig_final_wis_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_final_wis_heldout.png +0 −0

Binary file not shown.

modified paper/figures/fig_main_calibration.pdf +0 −0

Binary file not shown.

added paper/figures/fig_main_calibration_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_main_calibration_heldout.png +0 −0

Binary file not shown.

modified paper/figures/fig_main_crps.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_main_crps.png +0 −0

Binary file not shown.

added paper/figures/fig_main_crps_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_main_crps_heldout.png +0 −0

Binary file not shown.

modified paper/figures/fig_main_forest.pdf +0 −0

Binary file not shown.

added paper/figures/fig_main_forest_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_main_forest_heldout.png +0 −0

Binary file not shown.

modified paper/figures/fig_main_pit.pdf +0 −0

Binary file not shown.

added paper/figures/fig_main_pit_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_main_pit_heldout.png +0 −0

Binary file not shown.

modified paper/figures/fig_main_training.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_main_wis.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_main_wis.png +0 −0

Binary file not shown.

added paper/figures/fig_main_wis_heldout.pdf +0 −0

Binary file not shown.

added paper/figures/fig_main_wis_heldout.png +0 −0

Binary file not shown.

modified paper/main.pdf +0 −0

Binary file not shown.

modified paper/results_macros.tex +2508 −0
Large diff (2,511 lines) is not loaded by default to keep the page fast.Load diff
modified paper/sections/ablations.tex +66 −1
Large diff (67 lines) is not loaded by default to keep the page fast.Load diff
modified paper/sections/abstract.tex +17 −1
Large diff (18 lines) is not loaded by default to keep the page fast.Load diff
modified paper/sections/conclusion.tex +25 −1
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
modified paper/sections/introduction.tex +55 −1
Large diff (56 lines) is not loaded by default to keep the page fast.Load diff
modified paper/sections/model.tex +2 −0
Large diff (5 lines) is not loaded by default to keep the page fast.Load diff
modified paper/sections/results.tex +7 −2
Large diff (28 lines) is not loaded by default to keep the page fast.Load diff
added paper/sections/results_heldout.tex +30 −0
Large diff (30 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/ablations_calibration.tex +32 −0
Large diff (33 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/ablations_compare_crps.tex +31 −0
Large diff (32 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/ablations_compare_qlike.tex +31 −0
Large diff (32 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/ablations_compare_wis.tex +31 −0
Large diff (32 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/ablations_crps_scaled.tex +32 −0
Large diff (33 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/ablations_logscore.tex +32 −0
Large diff (33 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/ablations_qlike.tex +32 −0
Large diff (33 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/ablations_wis.tex +32 −0
Large diff (33 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_calibration.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_calibration_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_compare_crps.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_compare_crps_heldout.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_compare_qlike.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_compare_qlike_heldout.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_compare_wis.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_compare_wis_heldout.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_crps_scaled.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_crps_scaled_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_logscore.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_logscore_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_qlike.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_qlike_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_wis.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/final_wis_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/main_calibration.tex +2 −0
Large diff (13 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/main_calibration_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/main_compare_crps.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/main_compare_crps_heldout.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/main_compare_qlike.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/main_compare_qlike_heldout.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/main_compare_wis.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/main_compare_wis_heldout.tex +25 −0
Large diff (26 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/main_crps_scaled.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/main_crps_scaled_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/main_logscore.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/main_logscore_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/main_qlike.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/main_qlike_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/main_wis.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
added paper/tables/main_wis_heldout.tex +26 −0
Large diff (27 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/pilot_calibration.tex +2 −0
Large diff (13 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/pilot_compare_crps.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/pilot_compare_qlike.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/pilot_compare_wis.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/pilot_crps_scaled.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/pilot_logscore.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/pilot_qlike.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
modified paper/tables/pilot_wis.tex +2 −0
Large diff (14 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/baselines_meta.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/compare_crps_vs_naive_rw_test.csv +73 −0
Large diff (73 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/compare_qlike_vs_naive_rw_test.csv +73 −0
Large diff (73 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/compare_wis_vs_naive_rw_test.csv +73 −0
Large diff (73 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/leaderboard_test.csv +77 −0
Large diff (77 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_1m_5m/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_1m_5m/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_1m_only_b4k/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_1m_only_b4k/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_b4k/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_b4k/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_copula/copula/train_log.jsonl +37 −0
Large diff (37 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_copula/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_copula/train_summary.json +20 −0
Large diff (21 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_cross/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_cross/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_gaussian/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_gaussian/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_learnedregime/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_learnedregime/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_noregime/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_noregime/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_ohlcv/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_ohlcv/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_returns_only/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_returns_only/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_seed2/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_seed2/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_seed3/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_seed3/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_shuffled/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_shuffled/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_ticker/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_ticker/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_time_only/train_log.jsonl +209 −0
Large diff (209 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/mrpt_small_time_only/train_summary.json +10 −0
Large diff (11 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/pairwise_test.csv +193 −0
Large diff (193 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/path_scores_mrpt_small_copula_test.json +104 −0
Large diff (105 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/run_meta_test.json +435 −0
Large diff (436 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/stress_calibration_test.csv +137 −0
Large diff (137 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_ewma_test.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_1m_5m_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_1m_only_b4k_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_b4k_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_copula_indep_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_copula_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_cross_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_gaussian_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_learnedregime_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_noregime_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_ohlcv_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_returns_only_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_seed2_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_seed3_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_shuffled_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_ticker_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_mrpt_small_time_only_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_naive_rw_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/ablations/summary_seasonal_vol_test.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/baselines_meta.json +15 −0
Large diff (16 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/compare_crps_vs_naive_rw_final.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/compare_crps_vs_naive_rw_final_heldout.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/compare_qlike_vs_naive_rw_final.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/compare_qlike_vs_naive_rw_final_heldout.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/compare_wis_vs_naive_rw_final.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/compare_wis_vs_naive_rw_final_heldout.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/garch_params.json +422 −0
Large diff (423 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/leaderboard_final.csv +53 −0
Large diff (53 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/leaderboard_final_heldout.csv +53 −0
Large diff (53 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/pairwise_final.csv +121 −0
Large diff (121 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/pairwise_final_heldout.csv +97 −0
Large diff (97 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/run_meta_final.json +351 −0
Large diff (352 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/run_meta_final_heldout.json +171 −0
Large diff (172 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/stress_calibration_final.csv +81 −0
Large diff (81 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/stress_calibration_final_heldout.csv +73 −0
Large diff (73 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/subgroups_mrpt_small_vs_har_rv_final.csv +125 −0
Large diff (125 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_elastic_net_final.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_elastic_net_final_heldout.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_ewma_final.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_ewma_final_heldout.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_garch_t_final.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_garch_t_final_heldout.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_har_rv_final.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_har_rv_final_heldout.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_lightgbm_final.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_lightgbm_final_heldout.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_mlp_final.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_mlp_final_heldout.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_mrpt_base_final.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_mrpt_base_final_heldout.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_mrpt_small_1m_only_final.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_mrpt_small_1m_only_final_heldout.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_mrpt_small_final.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_mrpt_small_final_heldout.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_naive_rw_final.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_naive_rw_final_heldout.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_ridge_final.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_ridge_final_heldout.json +186 −0
Large diff (187 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_seasonal_vol_final.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_seasonal_vol_final_heldout.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_tcn_final.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/final/summary_tcn_final_heldout.json +162 −0
Large diff (163 lines) is not loaded by default to keep the page fast.Load diff
modified results/metrics/main/baselines_meta.json +3 −3
Large diff (10 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/main/compare_crps_vs_naive_rw_test_heldout.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/main/compare_qlike_vs_naive_rw_test_heldout.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
added results/metrics/main/compare_wis_vs_naive_rw_test_heldout.csv +49 −0
Large diff (49 lines) is not loaded by default to keep the page fast.Load diff
modified results/metrics/main/garch_params.json +289 −1446
Large diff (1,853 lines) is not loaded by default to keep the page fast.Load diff

Diff truncated — file too large.

Menu

Theme
↑↓ navigate ↵ open esc close / or ⌘K anywhere