CLI
Evaluate a ticker directly from the environment where your agent is already working.
$leroy evaluate ABNB
Read the CLI guide Don't let your agent trade blind.
Propose a trade and see how the exact same setup performed in the past. Backed by evaluated trades.
leroy evaluate ABNBBUY evidence
Evidence Strength
ABNB@ $191.13
Signal Pattern
50signal-pattern observations
Strategy mix
41 evaluatedBuy 18Neutral 14Sell 9
Observed Outcomes
15 Min+0.84%62% win rate · lift +8.40
30 Min+2.46%64% win rate · lift +12.50
60 Min+3.18%68% win rate · lift +18.10
Connect to the trading stack you already use.
Bring Leroy into the same agent workspace you use to research and trade. It adds current context and historical evidence before execution.
The same evidence, wherever your agent runs.
Use the CLI from a terminal or connect through MCP. Both interfaces reach the same Market Memory and return structured evidence your agent can reason over.
MCP
Give an agent a direct research tool with explicit status, context, and historical outcomes.
Set a wake rule through MCP, and Leroy will wake your agent with signed evidence when a fresh setup matches.
connect leroy / research
See example requests From request to response.
Scroll through the five passes behind a CLI or MCP answer. Every step adds context before Leroy returns the evidence.
Inputs
The agent sends the setup it is considering. Leroy does not need a candle dump or a giant prompt.
Command
Leroy CLI
$ leroy evaluate ABNBAssess Market
The current market frame adds context around the setup: volume, benchmarks, breadth, and volatility.
Volume
1.38× avg
SPY
+0.42%
VTI
+0.31%
Breadth
61% advancing
Volatility
Normal
Read Strategy Signals
Each supported strategy reports a signal so the agent can see agreement, disagreement, and neutral signals.
Opening Range 30
+0.82
Donchian 20
+0.67
Relative Strength
+0.41
VWAP
0.00
RSI 14
+0.04
Bollinger
−0.08
Find Historical Matches
Leroy searches the verified record for comparable setups and measures what followed them.
AAPL · Buy · 15 minutes
184 comparable historical setups
Average Return
+3.84%
Win Rate
63%
Profit Factor
1.42
Return Evidence
The agent receives the evidence, assumptions, and boundary in one deterministic response.
Possible Return
EvaluatedAAPL · buy
15-minute historical average
+3.84%
Samples
184
Win Rate
63%
Profit Factor
1.42
The archive behind the answer.
Terabyte-class market memory with more than 105 billion strategy evaluations across 41 registered strategies, indexed across decades of market history since 2010.
Cumulative strategy evaluations in Market Memory.
Since 2010
Historical market sessions prepared for repeatable research.
5/5
Match tiers from exact ticker setup to side baseline.
From market memory to a model.
Market Memory gives an agent current context, strategy signals, and historical outcomes. We're developing a paper model on top of that foundation—testing how those signals combine, exposing its decision threshold, and keeping every result inspectable.
LiveUpdated 1:11:42 PM Eastern
How’s the model doing today?
At this decision threshold, 268 trades would have executed and marked +0.79% after 15 minutes. Slide the threshold to compare all protectiveness levels.
Decision Threshold
Higher settings require stronger evidence, so fewer signals pass.
39%
SPY
−0.35%
VTI
−0.35%
Executed Trades
268
Wins v. Losses
137/114
15m Mark
+0.79%
Model Result
+0.79%
Executed Trades
268
137/114 Wins v. Losses
ModelSPYVTI9:30 AM – 1:10 PM Eastern
How did the model compare to SPY over the last 5 sessions?
The model beat SPY on 4 of 5 verified sessions.
Sep 8, 2026Beat
Model (P44)
+0.69%
SPY
−0.41%
Sep 9, 2026Beat
Model (P52)
+0.26%
SPY
−0.22%
Sep 10, 2026Beat
Model (P52)
+0.13%
SPY
−0.01%
Sep 11, 2026Beat
Model (P50)
+0.11%
SPY
−0.03%
Sep 14, 2026Trailed
Model (P50)
+0.15%
SPY
+0.28%
Give your agent a better question to answer.
Install the CLI, connect through MCP, and start with the evidence your agent needs.