TYO Research Log
Research Log
Finished systems are the smallest part of research. This log records the rest: hypotheses, controlled changes, results and decisions — including the rejected ones. Where historical metadata was never written down, the record says so instead of inventing it.
- 26Experiments
- 16Accepted
- 1Rejected
- 3Inconclusive
- 6Archived
- 14Systems covered
Every entry follows one shape: Observation → Hypothesis → Controlled change → Backtest → Validation → Decision. "Change parameters until it looks better" is not an experiment and is not recorded here.
Backtest improvement does not imply future improvement. The more experiments run, the more likely some succeed by chance — this log exists partly so that multiple-testing risk stays visible instead of hidden.
Development protocol
New strategies developed under the evidence-by-design protocol: unseen data is sealed BEFORE development, the search budget is capped and public, and the holdout opens exactly once — after the candidate is frozen.
RES-XAU-001
EVIDENCE BY DESIGN
TYO XAU PYRAMID
Prospective quality: Process prospective
Process prospective: the developer has lived through this market and MT5 itself can read all history, so isolation is procedural (sealed manifests, guarded tooling, audit log) — not cryptographic ignorance. Stated plainly by design.
Data layers
| Development | 2016-02-01 → 2023-12-31 | |
|---|---|---|
| Internal validation | 2024-01-01 → 2024-12-31 | |
| Prospective OOS | 2025-01-01 → 2025-12-31 | |
| FINAL HOLDOUT | 2026-01-01 → 2026-06-30 | 🔒Sealed |
Search budget
Logic variants 0/12
Experiments 15/30
Optimization runs 0/10
Frozen candidates 0
The more configurations are tried, the more likely one looks good by chance — so the search count itself is published as evidence.
Milestones
Baseline: doneInternal validation: pendingMonte Carlo: pending
Status
Category
Source
No experiment matches these filters.
-
EXP-000005 ✓Accepted
Risk/return map: simple annual average vs CAGR
- Hypothesis
- CAGR compares unequal test lengths more honestly than total-return-divided-by-years.
- Decision
- The simple average produced figures no reader interprets correctly and compressed twelve systems into an unreadable band behind two outliers. CAGR puts 13 of 14 inside 1–28%.
-
EXP-000006 ✓Accepted
Correlation minimum-overlap rule
- Hypothesis
- Pairwise correlations computed over few shared months are noise and should not be displayed.
- Decision
- The 13 suppressed pairs involve short-window systems whose r estimates would swing wildly with 1–2 months of data. Surviving range: −0.56 to +0.49, median 0.07.
-
EXP-000201 ✓Accepted
COSMOS — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.22%) and its positive-month share (100%) stayed within 10pp of the early window (100%).
Positive months 0pp Mean monthly return +0.06pp Worst month +0.09ppRead entry -
EXP-000202 ✓Accepted
EVERGREEN — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.48%) and its positive-month share (84.93%) stayed within 10pp of the early window (85.71%).
Positive months -0.78pp Mean monthly return -0.37pp Worst month +17.94ppRead entry -
EXP-000203 ✓Accepted
GALOA — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.89%) and its positive-month share (96.88%) stayed within 10pp of the early window (96.6%).
Positive months +0.28pp Mean monthly return -0.36pp Worst month +10.35ppRead entry -
EXP-000204 ✓Accepted
GODSPEED — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.31%) and its positive-month share (72.15%) stayed within 10pp of the early window (82.07%).
Positive months -9.92pp Mean monthly return -0.72pp Worst month +11.16ppRead entry -
EXP-000205 ≈Inconclusive
GRANDSLAM — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window shorter than 12 months.
Positive months 0pp Mean monthly return -1.01pp Worst month -0.39ppRead entry -
EXP-000206 ≈Inconclusive
JAYRO — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window shorter than 12 months.
Positive months 0pp Mean monthly return -4.2pp Worst month -0.53ppRead entry -
EXP-000207 ✓Accepted
JOKER — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.26%) and its positive-month share (86.67%) stayed within 10pp of the early window (85.51%).
Positive months +1.16pp Mean monthly return +0.19pp Worst month +0.5ppRead entry -
EXP-000208 ✓Accepted
KOKOMO — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.49%) and its positive-month share (98.28%) stayed within 10pp of the early window (89.63%).
Positive months +8.65pp Mean monthly return -1.09pp Worst month +15.2ppRead entry -
EXP-000209 ✓Accepted
METRO — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.35%) and its positive-month share (95.95%) stayed within 10pp of the early window (84.3%).
Positive months +11.65pp Mean monthly return -0.9pp Worst month +21.54ppRead entry -
EXP-000210 ✓Accepted
MIRUMO — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.93%) and its positive-month share (90%) stayed within 10pp of the early window (88.32%).
Positive months +1.68pp Mean monthly return -0.3pp Worst month +15.35ppRead entry -
EXP-000211 ✓Accepted
NEXUS — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (1.76%) and its positive-month share (93.33%) stayed within 10pp of the early window (86.96%).
Positive months +6.37pp Mean monthly return +1.04pp Worst month +0.93ppRead entry -
EXP-000212 ≈Inconclusive
RINA — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Positive-month share fell by 14.54pp — inside the rejection bound but outside the acceptance bound.
Positive months -14.54pp Mean monthly return -4.21pp Worst month +1.53ppRead entry -
EXP-000213 ✓Accepted
SENA — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.73%) and its positive-month share (87.18%) stayed within 10pp of the early window (93.18%).
Positive months -6pp Mean monthly return -1.32pp Worst month -4.55ppRead entry -
EXP-000214 ✓Accepted
SUPREMACY — 70/30 split stability of the monthly profile
- Hypothesis
- The monthly return profile of the first 70% of the backtest (positive-month share, mean monthly return) persists in the final 30%.
- Decision
- Late window kept a positive mean month (0.93%) and its positive-month share (100%) stayed within 10pp of the early window (100%).
Positive months 0pp Mean monthly return -0.2pp Worst month +0.31ppRead entry -
EXP-000001 ✓Accepted
Equity-basis vs balance-basis drawdown across the catalogue
- Hypothesis
- Balance-basis drawdown (closed trades only) materially understates peak-to-trough risk for systems that hold losing positions, and must not be ranked against equity-basis figures silently.
- Decision
- The gap on the worst case was a factor of ~4 (12.44% vs 48.82%). Publishing the balance figure as "the" drawdown would have materially understated risk.
-
EXP-000002 ✓Accepted
Stop-loss share denominator — data-integrity check
- Hypothesis
- An observed stop-loss share above 100% indicates the denominator excludes some closed trades.
- Decision
- Share can no longer exceed 100%; extracted totals now match each report header exactly.
-
EXP-000003 ✕Rejected
Positive-month share as a consistency score input
- Hypothesis
- Share of profitable months is a usable consistency measure for these systems as-is.
- Decision
- Grid- and martingale-style exits win almost every month by construction and give it back in one. Rewarding the raw share ranks the riskiest builds highest — the opposite of what a consistency score is for. Replaced by share × worst-month factor.
-
EXP-000004 ✓Accepted
TYO SCORE: worst-month factor and drawdown cap
- Hypothesis
- Scaling consistency by the worst month, and capping the total by drawdown, restores discrimination without hand-tuning per system.
- Decision
- RINA (2nd-highest profit factor, 56% DD) lands at 55 and JAYRO at 40 — the ordering now reflects what the score claims to measure. Caps are disclosed on every page they bind.
-
EXP-000101 ▤Archived
GALOA — SUNS build vs current build, full 2003–2026 window
- Hypothesis
- Not recorded
- Decision
- Not recorded
Profit factor +5.27 Max drawdown -6.09pp Trades +86 Win rate -2.12ppRead entry -
EXP-000102 ▤Archived
GALOA — recent-window run (2024–2026)
- Hypothesis
- Not recorded
- Decision
- Not recorded
-
EXP-000103 ▤Archived
GRANDSLAM — 2022-start vs 2024-start dataset window
- Hypothesis
- Not recorded
- Decision
- Not recorded
-
EXP-000104 ▤Archived
JOKER — MT4 build vs MT5 rebuild
- Hypothesis
- Not recorded
- Decision
- Not recorded
-
EXP-000105 ▤Archived
NEXUS — two builds on the identical dataset
- Hypothesis
- Not recorded
- Decision
- Not recorded
Profit factor +0.02 Max drawdown +2.96pp Trades 0 Win rate 0ppRead entry -
EXP-000106 ▤Archived
GODSPEED — cross-market run on GOLD
- Hypothesis
- Not recorded
- Decision
- Not recorded
Systems we have shipped.
Each Expert Advisor is documented with its concept, logic, risk model and test conditions. No black boxes, no unexplained results.