@tradejs/cli 3.1.29-beta.261 → 3.1.29-beta.263

package/README.md

@@ -102,8 +102,10 @@ yarn exec tradejs backtest -c TrendFollow:base -d 30 --cacheOnly --jev --jevReco
 Without `--jev`, ordinary backtests do not evaluate Jev, even if a configuration
 contains `JEV`. Use `--ai --jev` to save completed trades with Jev features for
 AI training. `--jev` alone records Jev assessments without enabling AI export.
-Jev can ask eight independent questions: trend, swing, current participation,
-setup participation, entry extension, confirmation, setup strength, and geometry.
+Jev uses one current set of nine questions: directional agreement, correction
+containment, current volume expansion, setup volume contraction, entry proximity,
+entry candle conviction, impulse persistence, boundary regularity, and overall
+signal strength.
 It asks each question only when the corresponding signal-time facts exist and
 marks missing facts explicitly. The shared projection
 selects a few signal-time trend, swing, volume, delta, entry-distance and
@@ -114,6 +116,38 @@ which is available to the strategy's AI gate and AI export. Jev scores do not
 approve or reject entries. Backtests do not apply the AI gate; `ai-train`
 evaluates it on exported outcomes, while runtime follows `AI_MODE`.
 
+`--jev` and runtime `JEV` always use the current compact nine-question set.
+There is no question-set selector or fallback to older rubrics. The recorded
+`micro-v3` marker and question hash identify the current meanings; old or
+unversioned recordings and local models are rejected. Use a new recording
+directory for the new backtest and retrain local models from current answers.
+
+The first eight questions describe separate signal-time facts. The ninth asks
+how strong and actionable the signal is: do the available facts support opening
+a trade now? It weighs direction, setup, confirmation, placement and the
+stop/target plan together. Essential missing facts and unresolved conflicts
+reduce support. All eligible questions share the same compact state in one
+request; outcomes and other Jev answers are excluded.
+
+Its ten rubric levels appear as
+`additionalIndicators.jev.scores.signalStrength`, from **1 to 10** (decimal
+expected scores are retained). Confidence remains 0–1. The API returns a
+zero-based 0–9 score; recordings and local teacher models normalize it to 0–1,
+and AI features convert it to 1–10. This is an entry assessment, not a calibrated
+probability of profit, and never automatically approves or blocks a trade.
+
+```bash
+yarn exec tradejs backtest -c <Strategy>:<config> -d 365 --cacheOnly --ai --jev \
+  --jevRecordsDir data/ai/jev/<new-study>
+```
+
+Use `--ai` to retain completed outcomes for comparing the score with realized
+PnL. Freeze score thresholds on the first 60% with three internal stability
+blocks before checking the last 40%. Compare with the current gate and ordinary
+features on the same rows, include independent event support and costs, and do
+not choose thresholds on the outer test. A fresh run does not make previously
+inspected historical outcomes untouched.
+
 Any strategy can optionally attach `jevEvidence` through its `StrategyAPI.entry`
 `additionalIndicators` with a version, `knownAt` timestamp, up to 16 scalar
 setup facts and 16 scalar geometry facts:
@@ -147,9 +181,12 @@ Provider responses and decisions are saved under `data/ai/jev`, or the directory
 selected with `--jevRecordsDir`. Input, selected question version and provider
 identity determine each signal record. Equivalent facts reuse one provider
 response even when signal timestamps or symbols differ. Input v4 records
-explicitly store missing facts and their sources. The changed question hash
-prevents older four-question recordings and local models from being replayed as
-answers to the new questions.
+explicitly store missing facts and their sources. Paid state groups fact paths
+by namespace and retains exact values, units and missing reasons; signal identity,
+internal versions and source metadata remain local. The current questions use
+short rubrics with explicit thresholds and unknown-data rules. Recordings and
+local models must match the current question marker and hash. Older versions
+are not migrated or used as substitutes for missing current answers.
 `--jevRecorded` requires the same signal-time input facts, question set and
 provider as the recording run. A shorter backtest may warm up indicators from
 a different history or reach the same date with a different trade state, so its
@@ -179,9 +216,11 @@ path for each run.
 
 Training fits shallow trees to Jev scores, without trade outcomes as features or
 labels. It requires at least 30 samples across 10 distinct signal times, selects
-one strategy and one teacher/provider lineage, and separates train/validation/test
-chronologically as 60/20/20. Three additional expanding-window folds check later
-periods with independently fitted trees. The report measures teacher agreement, score error
+one strategy and one teacher/provider lineage, and splits timestamp groups
+chronologically into 60% development and 40% test. Three expanding-window folds
+fit earlier development data and check later development data, entirely inside
+the first 60%. The final teacher model uses the full development partition.
+The outer test never fits or selects the model. The report measures teacher agreement, score error
 and matched-outcome economics. These summaries are not sequential portfolio
 backtests and do not establish profitability. Validate a frozen gate on a later
 period with `--jevModelFile` before runtime use. Studies are bounded to 20,000