Alphanume (@alphanume_data) header
Alphanume (@alphanume_data) profile picture

Alphanume

@alphanume_data

quantitative trading data for people who actually trade. your agent can pull it all: one MCP server. essays at https://nitter.cf/t.co/fVRc51y3eo

alphanume.com/
Joined February 2025
  • Tweets956
  • Following72
  • Followers125
798Media
Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)Photo by Alphanume (@alphanume_data)
Monetizing The Golden Age of Short Selling

Monetizing The Golden Age of Short Selling

Article · @alphanume_data
463
the SPX 0DTE expected-move archive opens at 4715 to 4765. the latest row: 7545 to 7665. 708 sessions in between, one row each, four columns, zero weekend rows. every edge sits on the 5-point grid. no close, no outcome. this is PUBLIC
Dark terminal card showing a curl call to the SPX 0DTE strike band endpoint and the first row on file, dated 2024-01-02, with lower strike 4715 and upper strike 4765, noting 708 sessions on file and a latest band of 7545 to 7665.

ALT Dark terminal card showing a curl call to the SPX 0DTE strike band endpoint and the first row on file, dated 2024-01-02, with lower strike 4715 and upper strike 4765, noting 708 sessions on file and a latest band of 7545 to 7665.

4
september's 4,726 ex-div events: 3,577 recovered, 956 not, 10 still counting. the recovery flags are counted sessions, not an estimate. -> the recovered rows watched a median of 2 sessions -> the ones that never got back watched all 20 39% were even in one.
Dark data card reading 2 vs 20 sessions: the median bars watched by September 2026 ex-dividend rows that recovered against rows that never did, out of 4,726 events with 10 still counting.

ALT Dark data card reading 2 vs 20 sessions: the median bars watched by September 2026 ex-dividend rows that recovered against rows that never did, out of 4,726 events with 10 still counting.

14
SPWR filed seven S-1 registrations in 2026. market cap at the first: $174M. at the seventh: $85M. 386 S-1s across 272 tickers this year. 23 names came back three times or more. machine-read, sized in shares, stamped with the cap at filing. it's PUBLIC
10
268 of 469 optionable names closed october 30 with implied vol above their own one-year median. realized vol cleared its own median in 218. -> both: 144 -> neither: 127 median iv rank 43.5. median hv rank 34. every name scored against its own year, not the index's.
Dark data card reading 268 of 469: the optionable names whose implied vol sat above their own 52-week median on 2026-10-30, with realized vol above its own median in 218, both in 144 and neither in 127.

ALT Dark data card reading 268 of 469: the optionable names whose implied vol sat above their own 52-week median on 2026-10-30, with realized vol above its own median in 218, both in 144 and neither in 127.

2
309 three-sigma wikipedia attention spikes in october. 193 names behind them. 80 of those tickers tripped it more than once. PINS five times, four sessions running. EA printed 132.27 sigma on october 21, on 18,657 reads. one parameter: zscore_30d_gte=3. it's PUBLIC
10
ask your agent "which is bigger, AMD or ORCL, and when did that change" -> get_historical_market_cap(ticker="AMD") -> daily market cap and shares outstanding, one row per session asked oct 30: AMD $824B, ORCL $395B. they traded places 3 times. the last one was april 24.
Composed agent exchange: the ask is which is bigger, AMD or ORCL, and when that changed, answered by get_historical_market_cap with AMD going from 364 to 824 billion dollars and ORCL from 562 to 395 billion over 2026.

ALT Composed agent exchange: the ask is which is bigger, AMD or ORCL, and when that changed, answered by get_historical_market_cap with AMD going from 364 to 824 billion dollars and ORCL from 562 to 395 billion over 2026.

1
1
44
ask your agent "what share of march 2026 was risk-off" -> get_sp500_risk_regime() -> one row per session since 2011-03-30, 2 columns asked oct 30: 22 of 22 sessions. the whole month. october 2026: 0 of 22. same flag, same call.
Composed agent exchange: the ask is what share of march 2026 was risk-off, answered by get_sp500_risk_regime with march 2026 at 22 of 22 sessions, july 2026 at 7 of 22 and october 2026 at 0 of 22.

ALT Composed agent exchange: the ask is what share of march 2026 was risk-off, answered by get_sp500_risk_regime with march 2026 at 22 of 22 sessions, july 2026 at 7 of 22 and october 2026 at 0 of 22.

14
291 of the 386 tickers in the corporate-default feed have no sector. the map has no date column. it is what is listed now, not what was listed then. the de-SPAC feed's prior tickers do the same: 464 of 621 missing. both sit at 25% coverage. the graveyard left the join key.
Dark data card reading 291 of 386: corporate-default-feed tickers with no sector in the ticker-classification map, alongside 464 of 621 missing de-SPAC prior tickers.

ALT Dark data card reading 291 of 386: corporate-default-feed tickers with no sector in the ticker-classification map, alongside 464 of 621 missing de-SPAC prior tickers.

14
49 of the 212 names ever held in the 10-name momentum basket lasted exactly one rebalance. january 2017 through october 2026, 118 rebalances: -> median tenure 4 rebalances -> 77 names held twice or fewer -> NVDA held 47 the constituents ship with the rank.
Dark data card reading 49 of 212: names that lasted exactly one rebalance in the momentum index across 118 rebalances, against a median tenure of 4 and NVDA at 47.

ALT Dark data card reading 49 of 212: names that lasted exactly one rebalance in the momentum index across 118 rebalances, against a median tenure of 4 and NVDA at 47.

4
13 names sat in the top decile of both implied and realized vol instability on 2026-10-30. independence would predict 5.3. -> 529 names in the cross-section -> 53 in each top decile -> CZR led both: 0.53 and 0.83 against medians of 0.09 and 0.13 unstable vol clusters.
Dark data card reading 13 names: the count sitting in the top decile of both implied and realized vol-of-vol on 2026-10-30, against 5.3 predicted by independence, with CZR leading both at 0.53 and 0.83.

ALT Dark data card reading 13 names: the count sitting in the top decile of both implied and realized vol-of-vol on 2026-10-30, against 5.3 predicted by independence, with CZR leading both at 0.53 and 0.83.

8
a model publishes 5 names every session, then writes down what they actually did. the picks are public. the features are not. 4 columns. date, ticker, return, absolute_move. full-year 2025: 250 sessions, 1,250 picks, 0 left unscored.
Dark terminal card: an alphanume API call to /v1/next-day-movers for 2026-10-30 returning one record with date, ticker AAOI, return null and absolute_move null, plus CIFR, FSLY, INOD and IOVA, noting 5 picks a session across 4 columns scored after the close.

ALT Dark terminal card: an alphanume API call to /v1/next-day-movers for 2026-10-30 returning one record with date, ticker AAOI, return null and absolute_move null, plus CIFR, FSLY, INOD and IOVA, noting 5 picks a session across 4 columns scored after the close.

11
the largest acceleration in the default feed: $3,300,000,000. SIVB, item 2.04, march 2023. the sentence that did it ships as a column: "The filing of the Chapter 11 Case constituted an 'event of default' under the indenture" 559 records. go pull it.
22
two spac mergers closed with $0 left in trust. 100% redeemed. the entire trust gone at closing. median across the 129 closed deals that report both columns: 80.9% redeemed. this is PUBLIC. go pull it.
10
ask your agent "which tickers have abnormal wikipedia attention, not just lots of it" -> get_wikipedia_views(date="2026-10-29") -> views, the 30-day baseline, and the z-score asked oct 29: AMZN led the universe at 9,518 views, 0.73 sigma. CART, read 5,499 times, was 31.21.
Composed agent exchange: the ask is which tickers have abnormal wikipedia attention rather than just a lot of it, answered by get_wikipedia_views with CART at 31.21 sigma, NOW at 22.75 and AMZN, the most-read page, at 0.73.

ALT Composed agent exchange: the ask is which tickers have abnormal wikipedia attention rather than just a lot of it, answered by get_wikipedia_views with CART at 31.21 sigma, NOW at 22.75 and AMZN, the most-read page, at 0.73.

14
ask your agent "who goes ex-dividend in the next 30 days, and did their last one recover" -> get_dividend_capture(upcoming=true, future_days=30) -> 776 events across 704 tickers asked oct 30: EVR goes ex nov 27. aug 28 drop: 2.01 on a 0.89 dividend, breakeven in 5 sessions.
Composed agent exchange: the ask is who goes ex-dividend in the next 30 days and did their last one recover, answered by get_dividend_capture with EVR, HII and FBP rows showing next ex-date, dividend, last ex-day drop and breakeven.

ALT Composed agent exchange: the ask is who goes ex-dividend in the next 30 days and did their last one recover, answered by get_dividend_capture with EVR, HII and FBP rows showing next ex-date, dividend, last ex-day drop and breakeven.

10
189 tickers entered the market-cap panel in september. -> the next busiest month of 2026 managed 98 -> 858 new tickers across the year so far -> 9,437 in the coverage map total one endpoint returns every ticker and the first date it has data for.
Dark data card reading 189 in september: new tickers entering the alphanume market-cap panel, against 98 in the next busiest month of the year and 9,437 tickers in the coverage map total.

ALT Dark data card reading 189 in september: new tickers entering the alphanume market-cap panel, against 98 in the next busiest month of the year and 9,437 tickers in the coverage map total.

10
the straddle overpriced the move 65% of the time. full-year 2025, 1,467 scored earnings events. -> 67.6% when the stock fell (765 events) -> 62.2% when it rose (696 events) the premium was fatter on the way down. the running track record ships in the row.
Dark data card reading 65% overpriced: across 1,467 scored full-year 2025 earnings events, the straddle overpriced the move 67.6% of the time when the stock fell and 62.2% when it rose.

ALT Dark data card reading 65% overpriced: across 1,467 scored full-year 2025 earnings events, the straddle overpriced the move 67.6% of the time when the stock fell and 62.2% when it rose.

12
49,625,000 shares offered against a 7,428,326 market cap in february. 50,000,000 shares against a 299,868 cap in october. same ticker. the row stamps the cap at filing, so the denominator is the one that was true that day. share count alone tells you nothing.
Dark data card reading 24.8x apart: the same ticker filed two S-1s in 2026, stamped at a 7,428,326 market cap in february and a 299,868 market cap in october.

ALT Dark data card reading 24.8x apart: the same ticker filed two S-1s in 2026, stamped at a 7,428,326 market cap in february and a 299,868 market cap in october.

8
UPXI options priced vol at 5.5x what the stock actually delivered. XP priced it at 0.17x. same session, same cross-section. 539 names in one call, median name at 0.95. nobody is looking at this.
27
one company filed 281 sec documents in a single day. morgan stanley, august 3 2026. the dataset counts filings per ticker per day, zero days included. 3,370 ticker-days in 2026 ran 12 or more. go pull it.
9
Load more