Nunchi-tradeNunchi-trade/agent-cli/skills/reflect/SKILL.md

reflect-performance-review

Skill521 starsChanged 7 months agoMITView on GitHub

Reflect, Evaluate, Fine-tune, Learn, Evolve, Correct, Transform — nightly automated performance review

In this repository6 other files it gives its agents.
- yex-traderSkill
- apex-strategySkill
- guard-trailing-stopSkill
- agent-onboardingSkill
- pulseSkill
- opportunity-radarSkill

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Load: Read trade log from data/cli/trades.jsonl
line 23
Distinctive:
Run REFLECT every night after trading stops — no exceptions
line 57
Distinctive:
ALWAYS read the full report before the next trading session
line 58
Distinctive:
NEVER ignore FDR warnings — fees silently kill profitability
line 62

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

SKILL.md

REFLECT — Reflect, Evaluate, Fine-tune, Learn, Evolve, Correct, Transform

Nightly automated performance review engine. APEX runs all day — REFLECT reviews every trade at night, computing metrics, detecting patterns, and producing data-driven improvement recommendations.

Architecture

trades.jsonl → ReflectEngine.compute() → ReflectMetrics → ReflectReporter.generate() → report.md

Load: Read trade log from data/cli/trades.jsonl
Pair: FIFO round-trip matching (buys to sells per instrument)
Compute: Win rate, PF, FDR, holding periods, direction analysis, monster dependency
Recommend: Rule-based recommendations from metrics
Report: Full markdown report saved to data/reflect/YYYY-MM-DD.md
Distill: 3-5 line summary for agent memory

Key Metrics

Metric	Formula	Healthy Range
Win Rate	winning_trades / total_trades	> 50%
Profit Factor (Gross)	gross_wins / gross_losses	> 1.5
Profit Factor (Net)	(gross_wins - fees) / gross_losses	> 1.2
FDR (Fee Drag Ratio)	total_fees / gross_wins * 100	< 20%
Monster Dependency	best_trade_pnl / net_pnl * 100	< 50%
Max Consecutive Losses	longest loss streak	< 5

Usage

hl reflect run                      # Review since last report
hl reflect run --since 2026-03-01   # Review from specific date
hl reflect report                   # View latest report
hl reflect report --date 2026-03-03 # View specific date
hl reflect history                  # Show metric trend over time
hl reflect history -n 30            # Last 30 reports

Agent Mandate

You are the REFLECT reviewer. Your job is to analyze every trade from the past session, compute performance metrics, identify weaknesses, and produce actionable recommendations. You run nightly — the APEX runs by day, you review at night.

RULES:

Run REFLECT every night after trading stops — no exceptions
ALWAYS read the full report before the next trading session
Act on CRITICAL recommendations immediately (FDR > 30%, win rate < 35%)
Track recommendations across reports — if the same issue appears 3+ times, escalate
Save the distilled summary to agent memory for next-session context
NEVER ignore FDR warnings — fees silently kill profitability

Decision Rules

Metric State	Severity	Action
FDR > 30%	CRITICAL	Reduce trade frequency or widen entry criteria immediately
FDR 20-30%	WARNING	Monitor — consider reducing size or frequency
FDR < 20%	OK	Fees are manageable
Win rate < 35%	CRITICAL	Tighten entry criteria — Radar threshold to 200+
Win rate 35-45%	WARNING	Review losing trades for pattern
Win rate > 50%	OK	Entries are working
Monster dep > 60%	WARNING	One trade carrying the session — diversify alpha
Monster dep > 80%	CRITICAL	Fragile — entire PnL depends on one lucky trade
Consec losses > 5	WARNING	Add loss streak circuit breaker to APEX
Long PnL < 0, Short PnL > 0	WARNING	Long entries are leaking — reduce long bias
Holding < 5 min dominates	WARNING	Over-trading — increase min hold time

Anti-Patterns

Ignoring REFLECT reports: Running APEX without reviewing REFLECT is flying blind. The same mistakes repeat.
Acting on single-day anomalies: One bad day doesn't mean the strategy is broken. Look at 5+ day trends via hl reflect history.
Optimizing for win rate alone: High win rate with low profit factor means you're taking small wins and large losses. Focus on PF.
Not tracking FDR: Fees are invisible during trading but compound devastatingly. FDR is the single most important "hidden" metric.
Changing strategy after one REFLECT report: REFLECT recommendations need 3+ consistent appearances before strategy changes.

Error Recovery

Error	Cause	Fix
`No trades found`	No trading activity in period	Normal — nothing to review
`Cannot pair round trips`	Unmatched buys/sells	Open positions — REFLECT pairs only closed trades
`trades.jsonl not found`	First run or wrong data dir	Run at least one trade first
`Report generation failed`	Disk full or permissions	Check `data/reflect/` directory permissions

Composition

REFLECT is the learning layer of the APEX system. Run REFLECT nightly after APEX stops. Feed REFLECT insights back into APEX configuration (Radar thresholds, DSL presets, position sizing). Over time, REFLECT recommendations should converge as the system improves.

Cron Template

# Nightly REFLECT review at 11:55 PM
55 23 * * * cd ~/agent-cli && source .venv/bin/activate && hl reflect run >> logs/reflect.log 2>&1

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free