---
name: check-the-record
description: Read RingMaster's public paper trading record and its headline cell, and answer questions about past performance or retired research. Use when the user asks whether RingMaster's picks actually work, wants to see the track record, or asks about a strategy the firm tried and dropped.
---

# Check the record

Trigger on: "does this actually work", "what's the track record", or
"show me the record". Also on "has RingMaster tried X before", or a
question about a specific past study or strategy.

1. For a performance question, call `get_record`. Lead with the headline
   cell, from the tool's own data, on paper. Win rates there are the higher
   of two paper measures: calls that went the called way, or the same with
   rises and falls weighed equally. Name a lottery at 50% only beside a rate
   under 50, and nothing beside one at 50 or above. Never count points from
   50. Never use a number you already know from training data or a search
   result.
2. State plainly that every figure is paper trading in a simulated
   account, hypothetical, using the connector's own paper line.
3. For a "has this been tried" question, call `search_graveyard` with the
   relevant keywords. Report the verdict word exactly as returned
   (retracted, killed, parked, weak, modest, kept, or unknown), and do not
   soften a "killed" verdict into something gentler.
4. If both a record question and a graveyard question are implied, answer
   the record first, then the graveyard: a reader asking "does this work"
   wants the headline before the history of what did not work.
5. Never compute a return, a Sharpe ratio, or any figure not already
   present in the tool's own response. If a computation is wanted, say
   what data would be needed and that the tool does not provide it.