How to check a short squeeze setup with an AI agent, instead of trusting the score

All outputs are real tool calls to the Equibles MCP server from 21 October 2026, trimmed.

Short-squeeze lists rank stocks by how crowded the short side looks. The number at the top summarises several factors, and two stocks with the same score can be on the list for very different reasons.

Here's how I get an AI agent to take a score apart and check each piece against the FINRA data underneath, using the board for the 30 September settlement.

Connect it first

In Claude, open Customize → Connectors → Discover, search for Equibles, click Connect to Claude and Authorize, then switch it on in a chat from + → Connectors. In ChatGPT: Plugins, search Equibles, Install plugin. In Claude Code, claude mcp add --transport http equibles https://mcp.equibles.com/mcp and then /mcp to sign in. The free plan covers 100 requests a day; other clients are in the setup docs.

1. Pull the board with liquidity floors

The prompt: "Using Equibles, show the top short-squeeze scores for stocks over $1B in market cap that trade at least $10M a day." The agent calls GetShortSqueezeScores with minMarketCap=1000000000 and minDollarVolume=10000000. The top five of 2,346:

The floors matter: without them the board fills with stocks too small or too thin to trade. And one row already looks odd. Hasbro (HAS) sits among stocks with 15% or more of their shares sold short, while its own figure is 5.9%.

2. Break the score into its factors

"Break down the squeeze scores for CHH and HAS." The same tool with ticker returns each factor and its percentile against 6,991 scored stocks:

Same headline, different stories. CHH's 98 is mostly positioning: a short position that would take about 11 days of average volume to buy back. HAS's 97 leans on catalysts: a price spike, a volume surge and an earnings date, which together hit the +20 cap.

3. Read the FINRA series underneath

"Show CHH and HAS short interest for the last eight settlements." GetShortInterest returns FINRA's twice-monthly positions:

Two things the score doesn't show:

• CHH's days to cover isn't new. It has stayed between 10 and 14 for the whole window. The latest +10.4% is a rebound after the position shrank from June to mid-September.
• HAS's increase came in one report. The +14.2% is the jump from 15 to 30 September, and the position is roughly back where it was in mid-June.

Days to cover is just the short position over average daily volume: 6,633,981 / 598,121 = 11.1 for CHH.

4. Don't misread daily short volume

"Show CHH's daily short volume for the last two weeks." GetShortVolume:

These volumes cover only trades reported to FINRA's off-exchange facilities, not the whole tape, and 40 to 50% short is the normal baseline because market makers sell short to fill buy orders. A 77% day is above that baseline. It is not 77% of all CHH trading, and it is not 77% new short positions.

5. Wait for the next report

FINRA publishes the 15 October settlement on 26 October. Until then, GetShortInterest adds a model estimate, labelled as not FINRA data:

• CHH: likely range 6,142,322 to 7,309,286 shares; the model gives a rise 45%, which it calls close to a coin flip.
• HAS: likely range 7,375,235 to 9,649,603 shares, 54% for a rise.

Both ranges straddle the 30 September figure, so the honest reading is "no call yet".

6. Put the price on top

"Show CHH and HAS closes since 30 September, and their earnings dates."

• CHH: $102.41 on 30 September, $113.46 on 20 October (+10.8%). Next earnings call: 2 November (GetUpcomingInvestorEvents).
• HAS: $87.68 to $109.52 (+24.9%), including +11.2% on 20 October, the day of its third-quarter earnings call (ListInvestorEvents).

The short positions in the score are from 30 September, but the price and catalyst factors are current, so HAS's call-day jump already sits inside its 97. Whether shorts covered into the move is something only the 26 October report can show.

The prompt I reuse

Using Equibles, for the top short-squeeze scores over $1B and $10M a day:
1. Break the top 3 into factors, base score and catalyst boost.
2. Show 8 settlements of FINRA short interest and days to cover.
3. Show 10 days of short volume, and say it is off-exchange only.
4. Give the model estimate for the next settlement as a range only.
5. Add closes since the settlement date and the next earnings date.

What a squeeze score can't tell you

• Whether the shorts are wrong. A crowded short can be right for months.
• Timing. Positions are reported twice a month with a lag, so they are about three weeks old by the time the score uses them.
• Cause and effect. Price spikes and earnings raise the score, and they are also what short sellers react to.

The full ranking is on the squeeze risk page, and each stock has its own series, like CHH short interest.

Nothing here is investment advice.