Skip to content

feat: Execution Replay & Notes (Tasks 5+6) #5

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
digital-thinking wants to merge 1 commit into
base: master
Choose a base branch
from
Open

feat: Execution Replay & Notes (Tasks 5+6) #5

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
284 changes: 283 additions & 1 deletion src/fin_trade/pages/portfolio_detail.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,6 +8,7 @@

from fin_trade.models import AssetClass, PortfolioConfig, PortfolioState, TradeRecommendation
from fin_trade.services import PortfolioService, AgentService, SecurityService
from fin_trade.services.execution_log import ExecutionLogService
from fin_trade.agents.service import (
DebateAgentService,
LangGraphAgentService,
Expand Down Expand Up @@ -68,7 +69,9 @@ def render_portfolio_detail_page(

st.divider()

tab1, tab2, tab3, tab4 = st.tabs(["Holdings", "Performance", "Execute Agent", "Trade History"])
tab1, tab2, tab3, tab4, tab5 = st.tabs(
["Holdings", "Performance", "Execute Agent", "Trade Log", "Execution History"]
)

with tab1:
_render_holdings(config, state, security_service)
Expand All @@ -84,6 +87,10 @@ def render_portfolio_detail_page(
with tab4:
render_trade_history(state.trades, security_service, config.asset_class)

with tab5:
_render_execution_history(
portfolio_name,
)

def _render_summary(
config: PortfolioConfig,
Expand Down Expand Up @@ -397,6 +404,10 @@ def _render_performance_chart(
except Exception:
pass # Silently skip benchmark if unavailable

# Load notes for chart annotations
log_service = ExecutionLogService()
notes = log_service.get_notes(config.name)

# Build the interactive chart
fig = _build_performance_figure(
filtered_timestamps,
Expand All @@ -407,6 +418,7 @@ def _render_performance_chart(
initial_investment,
asset_class=config.asset_class,
benchmark_data=benchmark_data,
notes=_map_notes_to_points(notes, filtered_timestamps, filtered_values),
)

st.plotly_chart(fig, use_container_width=True, config={"displayModeBar": True})
Expand Down Expand Up @@ -636,6 +648,7 @@ def _build_performance_figure(
initial_amount: float,
asset_class: AssetClass = AssetClass.STOCKS,
benchmark_data: dict | None = None,
notes: list[dict] | None = None,
) -> go.Figure:
"""Build the interactive stacked area Plotly figure."""
fig = go.Figure()
Expand Down Expand Up @@ -757,6 +770,34 @@ def _build_performance_figure(
)
)

# Note markers
if notes:
note_hover = []
for note in notes:
tags = ", ".join(note.get("tags", [])) if note.get("tags") else "No tags"
preview = note.get("note_text", "")
if len(preview) > 140:
preview = preview[:140] + "..."
note_hover.append(
f"<b>Note</b><br>{preview}<br><i>{tags}</i>"
)
fig.add_trace(
go.Scatter(
x=[note["timestamp"] for note in notes],
y=[note["value"] for note in notes],
mode="markers",
name="Notes",
marker=dict(
symbol="circle",
size=10,
color="#9C27B0",
line=dict(color="#4A148C", width=1),
),
hovertemplate="%{customdata}<extra></extra>",
customdata=note_hover,
)
)

# Initial investment line
fig.add_hline(
y=initial_amount,
Expand Down Expand Up @@ -814,6 +855,247 @@ def _build_performance_figure(
return fig


def _map_notes_to_points(
notes: list[dict],
timestamps: list,
values: list[float],
) -> list[dict]:
"""Map notes to chart points based on closest prior timestamp."""
if not notes or not timestamps:
return []

note_points = []
ts_list = list(timestamps)

for note in notes:
note_date = note.get("note_date")
if not note_date:
continue

idx = None
for i in range(len(ts_list) - 1, -1, -1):
if ts_list[i].date() <= note_date:
idx = i
break

if idx is None:
continue

note_points.append({
"timestamp": ts_list[idx],
"value": values[idx],
"note_text": note.get("note_text", ""),
"tags": note.get("tags", []),
"id": note.get("id"),
})

return note_points


def _render_execution_history(
portfolio_name: str,
) -> None:
"""Render execution history with full context and notes."""
import json
import pandas as pd

st.subheader("Execution History")

log_service = ExecutionLogService()
logs = log_service.get_logs(portfolio_name=portfolio_name, limit=50)

if not logs:
st.info("No execution logs yet. Run the agent to generate logs.")
return

for log in logs:
recommendations = []
if log.recommendations_json:
recommendations = json.loads(log.recommendations_json)

executed = set()
if log.executed_trades_json:
executed = set(json.loads(log.executed_trades_json))

rejected = set()
if log.rejected_trades_json:
rejected = set(json.loads(log.rejected_trades_json))

applied_count = len(executed)
rejected_count = len(rejected)
pending_count = len(recommendations) - applied_count - rejected_count

outcomes = log_service.get_recommendation_outcomes(log.id)
outcome_score = None
if outcomes:
outcome_values = [o["hypothetical_pl"] for o in outcomes if o["hypothetical_pl"] is not None]
if outcome_values:
outcome_score = sum(outcome_values)

outcome_label = "Outcome: N/A"
if outcome_score is not None:
outcome_label = f"Outcome: {'+' if outcome_score >= 0 else ''}{outcome_score:,.2f}"

expander_title = (
f"{log.timestamp.strftime('%Y-%m-%d %H:%M')} — {log.model} — "
f"{len(recommendations)} recs | {applied_count} applied / {rejected_count} rejected / "
f"{pending_count} pending — {outcome_label}"
)

with st.expander(expander_title, expanded=False):
col1, col2, col3 = st.columns(3)
with col1:
st.metric("Duration", f"{log.duration_ms}ms")
with col2:
st.metric("Total Tokens", f"{log.total_tokens:,}")
with col3:
st.metric("Agent Mode", log.agent_mode)

if log.error_message:
st.error(log.error_message)

# Recommendations with outcomes
st.markdown("### Recommendations & Outcomes")
if not outcomes:
st.caption("No recommendations available.")
else:
rows = []
for outcome in outcomes:
rows.append({
"Status": outcome["status"].capitalize(),
"Ticker": outcome["ticker"],
"Action": outcome["action"],
"Qty": outcome["recommended_quantity"],
"Rec Price": outcome["recommended_price"],
"Current/Exit": outcome["exit_price"] or outcome["current_price"],
"Hypo P/L": outcome["hypothetical_pl"],
"Actual P/L": outcome["actual_pl"],
})

df = pd.DataFrame(rows)
st.dataframe(
df,
column_config={
"Rec Price": st.column_config.NumberColumn("Rec Price", format="$%.2f"),
"Current/Exit": st.column_config.NumberColumn("Current/Exit", format="$%.2f"),
"Hypo P/L": st.column_config.NumberColumn("Hypo P/L", format="$%.2f"),
"Actual P/L": st.column_config.NumberColumn("Actual P/L", format="$%.2f"),
},
hide_index=True,
use_container_width=True,
)

# Full context from markdown logs
context = log_service.get_execution_with_context(log.id).get("log_context", {})

if context.get("analysis"):
st.markdown("### Full Agent Reasoning")
st.markdown(context["analysis"])
elif context.get("overall_reasoning"):
st.markdown("### Full Agent Reasoning")
st.markdown(context["overall_reasoning"])

if context.get("research"):
with st.expander("Research", expanded=False):
st.markdown(context["research"])

if context.get("debate") or context.get("bull_case") or context.get("bear_case"):
with st.expander("Debate Transcript", expanded=False):
if context.get("bull_case"):
st.markdown("**Bull Case**")
st.markdown(context["bull_case"])
if context.get("bear_case"):
st.markdown("**Bear Case**")
st.markdown(context["bear_case"])
if context.get("neutral_analysis"):
st.markdown("**Neutral Analysis**")
st.markdown(context["neutral_analysis"])
if context.get("debate"):
st.markdown("**Debate Rounds**")
st.markdown(context["debate"])
if context.get("moderator_verdict"):
st.markdown("**Moderator Verdict**")
st.markdown(context["moderator_verdict"])

if context.get("prompt"):
with st.expander("Full Prompt", expanded=False):
st.markdown(context["prompt"])

# Add Note UI
st.markdown("### Add Note")
note_key = f"note_{log.id}"
note_text = st.text_area(
"Note",
key=f"{note_key}_text",
placeholder="Add your observation about this execution...",
height=120,
)
common_tags = ["Earnings", "Fed Decision", "Market Correction", "Strategy Tweak"]
selected_common = st.multiselect(
"Quick Tags",
options=common_tags,
key=f"{note_key}_common",
)
tags_input = st.text_input(
"Tags (comma-separated)",
key=f"{note_key}_tags",
)
if st.button("Add Note", key=f"{note_key}_add"):
tags = []
if tags_input:
tags.extend([t.strip() for t in tags_input.split(",") if t.strip()])
tags.extend(selected_common)
tags = list(dict.fromkeys(tags))
try:
log_service.add_note(
portfolio_name=portfolio_name,
note_text=note_text,
execution_id=log.id,
tags=tags,
)
st.success("Note added.")
except Exception as e:
st.error(f"Failed to add note: {e}")

st.divider()
_render_notes_panel(portfolio_name, log_service)


def _render_notes_panel(
portfolio_name: str,
log_service: ExecutionLogService,
) -> None:
"""Render notes panel with filtering and search."""
st.subheader("Notes")

notes = log_service.get_notes(portfolio_name)
if not notes:
st.info("No notes yet. Add notes from the execution history.")
return

all_tags = sorted({tag for note in notes for tag in note.get("tags", [])})
tag_filter = st.selectbox(
"Filter by tag",
options=["All"] + all_tags,
index=0,
)
search_query = st.text_input("Search notes")

filtered = []
for note in notes:
if tag_filter != "All" and tag_filter not in note.get("tags", []):
continue
if search_query and search_query.lower() not in note["note_text"].lower():
continue
filtered.append(note)

for note in filtered:
tags_label = ", ".join(note.get("tags", [])) if note.get("tags") else "No tags"
header = f"{note['note_date'].strftime('%Y-%m-%d')} — {tags_label}"
with st.expander(header, expanded=False):
st.markdown(note["note_text"])


def _render_agent_execution(
config: PortfolioConfig,
state: PortfolioState,
Expand Down
Loading