Robot branch: EURJPY V1
Version reviewed: EURJPY V1.28 M15 Directional Half Filter
Observation period: Week 28
Week 28 result: -1.9448R
Research decision: Stop the broad V1.28 structure
Next research version: EURJPY V1.29 Selective Groups
Current priority: Tested Logic
Mode: PAPER_SIM_ONLY
orders_sent: NO
Live orders: No live orders were sent
Executive Summary
EURJPY V1.28 completed Week 28 with 47 closed paper trades and a final result of -1.9448R.
The version generated +22.3115R in Gross +R, but this was outweighed by -24.2563R in Gross -R. The M15 Directional Half Filter therefore did not make the broad EURJPY V1 structure selective enough.
A subsequent full-depth review combined and deduplicated the available EURJPY V1 trade history from Weeks 27 and 28. The resulting baseline contained 104 trades and produced:
- 45 winning trades
- 58 losing trades
- -2.2487R Net R
- -0.0216R average per trade
The review identified three smaller positive groups: one BUY group and two SELL groups.
The combined retrospective selection contained 13 trades and produced:
- 12 winning trades
- 1 losing trade
- 92.31% win rate
- +12.7112R Net R
This result was used to create the next paper-only research version: EURJPY V1.29 Selective Groups.
V1.29 is not a Demo Candidate. The selected conditions were identified from only two weeks of existing research data and require fresh forward paper observation during Week 29.
Purpose of This Report
This report documents the Week 28 result of EURJPY V1.28 and explains why the EURJPY V1 branch moved from broad signal admission to an EURJPY-specific selective model.
The report covers two separate stages:
- the Week 28 paper result produced by EURJPY V1.28; and
- the retrospective review used to define EURJPY V1.29.
The selected V1.29 groups are a new research hypothesis. Their historical result must not be interpreted as independent forward performance.
EURJPY V1.28 Research Logic
EURJPY V1.28 used the M15 Directional Half Filter.
The filter was designed to admit a candidate entry only when the latest M15 candle close was positioned inside the directional half of that candle.
The purpose of this controlled change was to reduce entries that lacked sufficient short-term directional confirmation without replacing the broader EURJPY V1 structure.
Week 28 tested whether this filter could improve the balance between positive and negative trades.
Week 28 Results
| Metric | EURJPY V1.28 |
|---|---|
| Closed trades | 47 |
| Winning trades | 20 |
| Losing trades | 27 |
| Zero-result trades | 0 |
| Win rate | 42.55% |
| Gross +R | +22.3115R |
| Gross -R | -24.2563R |
| Net R | -1.9448R |
| Average R per trade | -0.0414R |
| TP exits | 15 |
| SL exits | 20 |
| MAX_HOLD exits | 12 |
Internal weekly target: +5R
Target status: Not reached
Interpretation of the Week 28 Result
EURJPY V1.28 continued to generate substantial positive trade material. Gross +R exceeded +22R during Week 28.
However, Gross -R was larger than Gross +R. The negative contribution reached -24.2563R, leaving the final weekly result at -1.9448R.
The problem was therefore not an absence of profitable signals. The problem was that the broad V1.28 structure admitted too many weak or losing trades alongside them.
The exit distribution supported the same conclusion:
- 15 trades reached TP;
- 20 trades reached SL;
- 12 trades closed through MAX_HOLD.
The M15 Directional Half Filter alone did not sufficiently isolate the strongest parts of the EURJPY V1 signal population.
Full-Depth EURJPY V1 Review
After Week 28, the available EURJPY V1 closed-trade files were combined and duplicate trades were removed.
The review examined the full available depth of the branch rather than relying only on the latest weekly result.
The analysis covered:
- the overall multiweek result;
- weekly stability;
- BUY and SELL directions;
- sessions;
- corridor types;
- position-percentage areas;
- entry-body classifications;
- and combinations that could form a simple auditable allowlist.
After deduplication, the EURJPY V1 baseline contained 104 trades.
Full Deduplicated EURJPY V1 Baseline
| Metric | EURJPY V1 Baseline |
|---|---|
| Trades | 104 |
| Winning trades | 45 |
| Losing trades | 58 |
| Zero-result trades | 1 |
| Win rate | 43.27% |
| Gross +R | +52.1166R |
| Gross -R | -54.3653R |
| Net R | -2.2487R |
| Average R per trade | -0.0216R |
| TP exits | 32 |
| SL exits | 47 |
| MAX_HOLD exits | 25 |
The broad EURJPY V1 baseline was negative overall.
The branch produced more than +52R in Gross +R, but this positive material was accompanied by more than -54R in Gross -R.
The average result per trade remained below zero, and SL exits substantially exceeded TP exits.
EURJPY V1 Weekly Baseline
| Week | Trades | Wins | Losses | Gross +R | Gross -R | Net R |
|---|---|---|---|---|---|---|
| Week 27 | 57 | 25 | 31 | +29.8050R | -30.1090R | -0.3040R |
| Week 28 | 47 | 20 | 27 | +22.3115R | -24.2563R | -1.9448R |
Both observed weeks were negative.
Week 27 was close to flat, while Week 28 showed a larger negative result. Neither week reached the internal +5R target.
The sum of the displayed rounded weekly values may differ from the full baseline by 0.0001R because the complete baseline was calculated from unrounded trade-level results.
Main Negative Areas
The full-depth audit identified several areas with large negative contributions.
| Negative area | Trades | Net R | SL exits |
|---|---|---|---|
| BODY_ZERO | 33 | -9.5639R | 19 |
| SELL + BODY_ZERO | 22 | -7.5420R | 13 |
| SELL + PARALLEL_HORIZONTAL + ASIA_00_05 | 11 | -6.9238R | 8 |
| SELL + PARALLEL_HORIZONTAL | 13 | -5.2765R | 8 |
| SELL + PCT_90_100 + BODY_ZERO | 8 | -4.9863R | 5 |
The strongest negative concentration appeared around BODY_ZERO, particularly on the SELL side.
SELL trades in PARALLEL_HORIZONTAL corridors also performed poorly, especially during the ASIA_00_05 session.
This evidence did not support continuation of the broad EURJPY V1 model. The more useful signal groups needed to be isolated directly.
Positive Selection Review
Three positive groups were selected for further review.
Unlike the EURGBP V4 and EURGBP V5 branches, the EURJPY analysis did not point toward a SELL-only model. It identified one BUY group and two SELL groups.
Selected Group A
BUY + ASIA_00_05 + PCT_10_30
| Metric | Group A |
|---|---|
| Trades | 4 |
| Winning trades | 4 |
| Losing trades | 0 |
| Win rate | 100.00% |
| Gross +R | +4.9907R |
| Gross -R | 0.0000R |
| Net R | +4.9907R |
| Average R per trade | +1.2477R |
| TP exits | 4 |
| SL exits | 0 |
| MAX_HOLD exits | 0 |
Group A was the cleanest BUY group in the available EURJPY V1 review.
All four selected trades reached TP. However, the sample is too small to support a strong conclusion.
Selected Group B
SELL + PARALLEL_ASCENDING + ASIA_00_05 + PCT_70_90 + BODY_WEAK_POS
| Metric | Group B |
|---|---|
| Trades | 3 |
| Winning trades | 3 |
| Losing trades | 0 |
| Win rate | 100.00% |
| Gross +R | +2.8310R |
| Gross -R | 0.0000R |
| Net R | +2.8310R |
| Average R per trade | +0.9437R |
| TP exits | 2 |
| SL exits | 0 |
| MAX_HOLD exits | 1 |
Group B was also very small, but it showed useful SELL material inside PARALLEL_ASCENDING, ASIA_00_05 and PCT_70_90 conditions.
Selected Group C
SELL + LONDON_06_10 + PCT_70_90
| Metric | Group C |
|---|---|
| Trades | 6 |
| Winning trades | 5 |
| Losing trades | 1 |
| Win rate | 83.33% |
| Gross +R | +5.9269R |
| Gross -R | -1.0375R |
| Net R | +4.8894R |
| Average R per trade | +0.8149R |
| TP exits | 5 |
| SL exits | 1 |
| MAX_HOLD exits | 0 |
Group C was the largest of the three selected groups.
It contained one losing trade during Week 28 but remained strongly positive across the available sample and contributed important trade count to the combined model.
Combined A+B+C Candidate
The final retrospective candidate for EURJPY V1.29 combined Group A, Group B and Group C.
| Metric | A+B+C Allowlist |
|---|---|
| Trades | 13 |
| Winning trades | 12 |
| Losing trades | 1 |
| Zero-result trades | 0 |
| Win rate | 92.31% |
| Gross +R | +13.7487R |
| Gross -R | -1.0375R |
| Net R | +12.7112R |
| Average R per trade | +0.9778R |
| TP exits | 11 |
| SL exits | 1 |
| MAX_HOLD exits | 1 |
The combined candidate reduced the admitted trade count from 104 to 13.
Its retrospective result improved from -2.2487R for the broad baseline to +12.7112R for the selected groups.
The displayed total may differ by 0.0001R from the sum of the rounded group figures because the combined result was calculated from unrounded trade-level data.
Combined Candidate by Week
| Week | Trades | Winning trades | Losing trades | Gross +R | Gross -R | Net R |
|---|---|---|---|---|---|---|
| Week 27 | 7 | 7 | 0 | +7.8436R | 0.0000R | +7.8436R |
| Week 28 | 6 | 5 | 1 | +5.9051R | -1.0375R | +4.8676R |
The selected historical profile was materially stronger than the broad baseline.
Week 27 exceeded the internal +5R target. Week 28 remained positive but finished slightly below the target after one SL exit.
This result remains preliminary because the review covers only two weeks and only 13 selected trades.
Analysis of the Excluded Trades
If the A+B+C allowlist is applied retrospectively, 91 of the 104 baseline trades are excluded.
| Metric | Excluded Trades |
|---|---|
| Trades | 91 |
| Winning trades | 33 |
| Losing trades | 57 |
| Zero-result trades | 1 |
| Win rate | 36.26% |
| Gross +R | +38.3679R |
| Gross -R | -53.3278R |
| Net R | -14.9599R |
| Average R per trade | -0.1644R |
| TP exits | 21 |
| SL exits | 46 |
| MAX_HOLD exits | 24 |
The excluded population still contained 33 profitable trades and more than +38R in Gross +R.
However, it also contained 57 losing trades and more than -53R in Gross -R, producing a combined result of -14.9599R.
This does not prove that every excluded trade is structurally bad. It shows that the broad excluded population did not justify continued admission under the new research model.
Version Decision
Decision: REPLACE EURJPY V1.28 AS THE ACTIVE V1 RESEARCH VERSION
Previous active version: EURJPY V1.28 M15 Directional Half Filter
New active research version: EURJPY V1.29 Selective Groups
Priority: Tested Logic
Testing mode: PAPER_SIM_ONLY
EURJPY V1.28 will remain available as an audit reference, but its broad signal structure will not continue as the active EURJPY V1 research model.
EURJPY V1.29 represents a structural transition:
- from broad signal admission to direct allowlist selection;
- from one general M15 filter to three specific signal groups;
- from broad directional activity to selected BUY and SELL conditions;
- and from incremental V1.28 filtering to a new V1.29 research hypothesis.
EURJPY V1.29 Logic
EURJPY V1.29 is a mixed-direction selective paper research model.
Allowed Group A
- Direction: BUY
- Session: ASIA_00_05
- Position percentage: PCT_10_30
Allowed Group B
- Direction: SELL
- Corridor type: PARALLEL_ASCENDING
- Session: ASIA_00_05
- Position percentage: PCT_70_90
- Entry body: BODY_WEAK_POS
Allowed Group C
- Direction: SELL
- Session: LONDON_06_10
- Position percentage: PCT_70_90
A candidate signal is admitted only when it matches Group A, Group B or Group C.
All other candidate signals are blocked with:
SIGNAL_BLOCKED_NOT_IN_EURJPY_V1_29_SELECTIVE_GROUP
EURJPY V1.29 was derived independently from the EURGBP V4 and EURGBP V5 branches. Its selected conditions and mixed-direction structure are specific to the available EURJPY research data.
Why V1.29 Is Not a Demo Candidate
The selected retrospective result is materially stronger than the broad EURJPY V1 baseline, but it is not sufficient for Demo Candidate status.
The selected conditions were identified from the same two-week dataset used to calculate their historical results. This introduces selection bias and creates a high risk that the observed performance is specific to the reviewed sample.
Current limitations include:
- only 13 selected trades;
- only two observed weeks;
- only 4 Group A trades;
- only 3 Group B trades;
- only 6 Group C trades;
- no independent forward observation period;
- and no evidence that the selected structure will remain stable under new market conditions.
EURJPY V1.29 therefore remains in the Tested Logic category.
Week 29 Research Objective
Week 29 will be the first fresh forward paper observation period for EURJPY V1.29.
The main questions are:
- Will any of the three selected groups produce new qualifying signals?
- Will the mixed BUY and SELL structure remain positive on unseen data?
- Will the selected groups generate sufficient trade frequency?
- Will Gross -R remain controlled?
- Will BODY_ZERO and PARALLEL_HORIZONTAL exclusions reduce the negative side?
- Will the exit distribution remain dominated by TP rather than SL?
- Will the task and output files remain operationally stable?
No priority upgrade will be made solely because the retrospective A+B+C result is positive.
Operational Status
EURJPY V1.29 was implemented as a separate paper-only research task.
- Mode: PAPER_SIM_ONLY
- orders_sent: NO
- Live orders: No live orders are sent
- Order execution: Disabled
- Output files: Separate V1.29 files
- VBS wrapper: Separate V1.29 wrapper
- Scheduled task: Separate V1.29 task
- Previous task: V1.28 disabled after V1.29 verification
The active EURJPY V1 research branch is now EURJPY V1.29 Selective Groups.
Research Conclusion
EURJPY V1.28 completed Week 28 with a negative result of -1.9448R.
The M15 Directional Half Filter did not sufficiently reduce the negative side of the broad EURJPY V1 model. The full deduplicated history confirmed the problem: 104 trades produced a combined result of -2.2487R.
The strongest observed structure was concentrated inside three smaller EURJPY-specific groups:
- BUY during ASIA_00_05 from PCT_10_30;
- SELL during ASIA_00_05 from PARALLEL_ASCENDING and PCT_70_90 with BODY_WEAK_POS;
- SELL during LONDON_06_10 from PCT_70_90.
The combined historical selection produced +12.7112R from 13 trades. However, this result was derived from the same two-week dataset used to select the conditions and is not independent confirmation.
The Week 28 decision is:
- stop V1.28 as the active broad EURJPY V1 research version;
- retain V1.28 as an audit reference;
- activate EURJPY V1.29 as Tested Logic;
- run V1.29 in PAPER_SIM_ONLY mode;
- and evaluate it using fresh Week 29 forward paper data.
Risk Note
This report documents an internal trading robot research process.
EURJPY V1.28 and EURJPY V1.29 operate only in PAPER_SIM_ONLY mode. orders_sent remains NO, and no real orders are sent.
The V1.29 conditions were selected from a small historical paper sample covering only two weeks. Their reported result may be materially affected by selection bias and does not demonstrate how the version will perform under future market conditions.
Historical analysis, backtesting, live paper observation and weekly reports do not guarantee future performance.
Forex and CFD trading involve significant risk. This report is provided for research and educational purposes only and does not constitute financial advice, investment advice, a recommendation to trade, or a promise of future returns.