@alphanume_data

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

alphanume.com/
Joined February 2025
  • Tweets909
  • Following69
  • Followers122
757Media

Monetizing The Golden Age of Short Selling

Article · @alphanume_data
448
every completed SPAC merger since january 2021 read out of the super 8-K and the proxy trail into one structured row redemptions, trust remaining, PIPE, sponsor. the numbers nobody else bothers to structure the graveyard has a schema this is PUBLIC
Terminal card: curl api.alphanume.com/v1/de-spac-events returns one completed de-SPAC record for FAC with 23,051,313 shares redeemed, 11,200,000 of trust remaining and a 100,900,000 PIPE.

ALT Terminal card: curl api.alphanume.com/v1/de-spac-events returns one completed de-SPAC record for FAC with 23,051,313 shares redeemed, 11,200,000 of trust remaining and a 100,900,000 PIPE.

3
ask your agent "is AAPL's implied vol expensive next to its own year" -> get_iv_rank -> the 52-week high, low and median ride the row with the reading asked september 3 2026: AAPL iv rank 43, sitting at 20% inside a 52-week range of 14% to 28%.
composed exchange: asking whether AAPL's implied vol is expensive next to its own year returns iv rank 43 with the 52-week range 0.14 to 0.28 around a 0.20 reading

ALT composed exchange: asking whether AAPL's implied vol is expensive next to its own year returns iv rank 43 with the 52-week range 0.14 to 0.28 around a 0.20 reading

4
86 Item 2.04 corporate default records in full-year 2025, machine-read off EDGAR into 15 columns. 76 of them carry the verbatim sentence they were read from. all 86 carry the filing link and a confidence score. the extraction shows its work. that is the whole methodology.
data card: 76 of the 86 Item 2.04 corporate default records from full-year 2025 carry the verbatim sentence they were read from, and all 86 carry the filing link

ALT data card: 76 of the 86 Item 2.04 corporate default records from full-year 2025 carry the verbatim sentence they were read from, and all 86 carry the filing link

5
the sector map is a closed vocabulary. 11 sectors, 25 industries, both validated server-side: send a value that isn't on the list and the endpoint hands you the whole list back. no date column, so nothing to clamp. it's FREE.
Dark terminal card: a GET /v1/ticker-classification call with an invalid sector returning INVALID_SECTOR and the whole valid_values list, over a footer reading sectors 11, industries 25, date columns 0.

ALT Dark terminal card: a GET /v1/ticker-classification call with an invalid sector returning INVALID_SECTOR and the whole valid_values list, over a footer reading sectors 11, industries 25, date columns 0.

17
550 S-1 stock offerings hit EDGAR in 2025. 462 went effective median time from filing to effectiveness: 13 days that is the median runway between the paperwork and the share flood. calendar 2025, every filing in the dataset
Data card: the median 2025 S-1 went from filing to effectiveness in 13 days; 550 filings, 462 effective.

ALT Data card: the median 2025 S-1 went from filing to effectiveness in 13 days; 550 filings, 462 effective.

5
every US ex-dividend event, resolved. declaration through pay date. the cum-day close, the ex-day drop, the ratio between them. then whether the stock got back to breakeven in 1, 3, 5, 10 or 20 sessions. 32 columns. the forward calendar ships in the same file. this is PUBLIC
Terminal card: the dividend-capture endpoint called with upcoming=true and future_days=120, listing the columns date, cum_close, drop_ratio_close and recovered_within_5d with plain-English descriptions.

ALT Terminal card: the dividend-capture endpoint called with upcoming=true and future_days=120, listing the columns date, cum_close, drop_ratio_close and recovered_within_5d with plain-English descriptions.

6
Alphanume retweeted
Ethan Kho
@ethanrkho22h
$480M Hedge Fund Manager David Orr (@orrdavid) on why big firms underallocate to their best PMs: "In a single year I'm getting more reps than a lot of hedge fund managers get in a decade." "What you're looking for in a portfolio manager, there's just a guy who consistently wins. And these people are out there. You just have to go find them." "But what the bigger firms seem to do is just still give them only a small allocation." "One of the guys, he's short only, and he's won 40% a year short only for years now. I mean, the guy's a killer." "Somehow he gets a tiny allocation at the fund he's at. If you do that at my fund, man, we're gonna give you a lot of money." "The thing is, I heard it doesn't scale super well, so maybe that's why they don't upsize him."
Ethan Kho
@ethanrkhoOct 23
Hedge Fund Manager David Orr (@orrdavid) on why he’ll back a delivery guy over a Harvard grad, as long as the track record is real. David Orr runs Militia Capital, a ~$480M hedge fund. ~50% annualized, Sharpe 2. “If you didn’t go to Harvard, they’re not going to give you money. They’re gonna lead you on for years.” We cover: - Edge is just reps. He's placed a few hundred independent bets every year since 2019, more in one year than a lot of hedge fund managers get in a decade. - The two-hour rule: if he can't decide on a stock in an hour or two, he skips it. When a trade goes against him, he quits. "The market said I'm wrong, I'm wrong." - 500 positions in his partition and zero notes. Some of his best shorts are down 40% and he's forgotten what the company does. - The Gogo lesson: he spent 100 hours on a short and was right on the fundamentals (it's now $2), but it ripped to $20 first. All that research made him stubborn on the way up. - The AI risk system: it groups the book into custom baskets like "Banks Done Right" and "yield traps," recalculated daily. A 25% bet can quietly become 45% in two weeks. - How Militia recruits PMs: 95% of track records are junk, and one cherry-picker ran 30 accounts and sent the best one. The PMs who make it include a game show winner, a Spanish government worker and a programmer, with 0.15 cross-correlation. - The best short seller in the world, quietly: 40% a year short only for years, and he still gets a tiny allocation at his fund. - Why multi-managers' real business is marketing. Collect $2B a year in fees on $100B and it's "mission accomplished," whatever the returns. - The long/short tax trick: harvest losses on losing shorts, and never cover one that's down 90%. No tax is due until it's delisted, which can take 10 years. - $150M raised from one email asking LPs for $100M. He says he could raise a billion but won't, and admits they "must have gotten at least somewhat lucky." - His AI stock map: hyperscalers locked customers into take-or-pay deals at ~35% return on capital. If compute prices crash 80%, the end customers get wiped out, not the hyperscalers. - Is investing 99% talent? For big alpha, almost entirely. Pod shop alpha is teachable, maybe for 1 in 50 people rather than 1 in 10,000. Highlights: 00:00 Intro 00:44 How AI custom factor baskets track dynamic correlation in levered portfolios 03:24 Sourcing non-consensus PM talent and verifying live Sharpe track records 07:24 Structuring carry and tax deferral to attract elite proprietary traders 12:50 Why institutional multi-manager platforms fail to back pure alpha 20:22 The two-hour research limit: Avoiding sunk-cost bias in short selling 27:58 Momentum screening vs deep value: Why top longs hit new highs 29:40 Why traditional finance books fail and what operator biographies teach 37:16 Why rapid technological adaptation beats traditional economic moats 39:26 Analyzing the AI trade: Why hyperscalers offer superior downside protection 44:57 Internalizing stock borrow desks to maximize net-of-fee fund compounding 55:25 Avoiding confirmation bias and why shorting revenue growth fails 01:01:42 Pod shop leverage factories vs raw discretionary alpha generation 01:07:12 Unplugging from market narrative feeds to preserve independent conviction
1
3
57
10,439
your agent can answer "which stocks are getting unusual wikipedia attention" in one call -> get_wikipedia_views, one z-score filter -> the 2026-08-11 pull: 18 names above z=3, the loudest 8.6 sigma over its own average the question is the whole interface
Composed agent exchange: asking which stocks are getting unusual wikipedia attention runs get_wikipedia_views and returns ONON, PFE and HSY with views, 30-day averages and z-scores from 2026-08-11.

ALT Composed agent exchange: asking which stocks are getting unusual wikipedia attention runs get_wikipedia_views and returns ONON, PFE and HSY with views, 30-day averages and z-scores from 2026-08-11.

8
most expected-move models want a vol surface, a term structure, and a prayer the SPX 0-DTE band wants one number the market already priced the day. the band just reads it back out
Data card: the SPX 0-DTE expected-move band needs one input; no vol surface, no term structure.

ALT Data card: the SPX 0-DTE expected-move band needs one input; no vol surface, no term structure.

17
every us ticker's market cap, stamped on the session that made it. four columns: date, ticker, market_cap, shares_outstanding. the panel opens 2015-09-30, 3,513 names on day one. ask for one ticker, get its whole history. this is PUBLIC
Dark terminal card: a curl to /v1/historical-market-cap for AAPL returning one row with date, ticker, market_cap and shares_outstanding. Footer: four columns, one row per ticker per session, coverage opens 2015-09-30.

ALT Dark terminal card: a curl to /v1/historical-market-cap for AAPL returning one row with date, ticker, market_cap and shares_outstanding. Footer: four columns, one row per ticker per session, coverage opens 2015-09-30.

11
ask your agent "did this name have listed options in august 2026, and how often did they expire" -> get_optionable_tickers -> a stamped snapshot, not the live universe backfilled september 1 2026: 3,532 optionable names, 554 on weeklies. NVDA expiring every 3 days, KO every 7.
composed exchange: asking whether NVDA had listed options in august 2026 returns the september 1 2026 snapshot, NVDA and AAPL expiring every 3 days, KO and SNDK every 7, all on weeklies

ALT composed exchange: asking whether NVDA had listed options in august 2026 returns the september 1 2026 snapshot, NVDA and AAPL expiring every 3 days, KO and SNDK every 7, all on weeklies

15
every row's earnings track record is stamped lookahead-free the hit rate on an event counts only the events before it, never itself, never the ones after so a backtest that reads the column as served is not peeking AAPL, july 30 2026 row: 10 prior events, 90%
Data card: 90% in huge type. The straddle-overpriced hit rate stamped on AAPL's july 30 2026 earnings row, computed from the 10 events before it and none after, showing the running track record is lookahead-free.

ALT Data card: 90% in huge type. The straddle-overpriced hit rate stamped on AAPL's july 30 2026 earnings row, computed from the 10 events before it and none after, showing the running track record is lookahead-free.

18
implied vol against realized vol for every optionable name. one row each, 23 columns wide. the intraday provisional row ships too, flagged is_final. the settled one is stamped 16:30 ET. each row carries the strike, the spot and the days to expiry. this is PUBLIC
Terminal card: GET /v1/iv-hv-premium returns priced vol against delivered vol per optionable name with the spread, the ratio, cross-sectional ranks and z-scores, the strike and spot behind the row, and an is_final flag; 23 columns per row, settled rows stamped 16:30 ET, coverage since 2024-01-02.

ALT Terminal card: GET /v1/iv-hv-premium returns priced vol against delivered vol per optionable name with the spread, the ratio, cross-sectional ranks and z-scores, the strike and spot behind the row, and an is_final flag; 23 columns per row, settled rows stamped 16:30 ET, coverage since 2024-01-02.

13
two columns in the s&p risk-regime feed: the date, and the call. no confidence score. no neutral state. not one session where the model declines to answer, going back to march 30 2011. a regime signal that is allowed to abstain is easy to make look good.
Data card: the s&p risk-regime endpoint serves two columns, the date and a binary call, with no confidence score, no neutral state and not one null in the flag since 2011-03-30.

ALT Data card: the s&p risk-regime endpoint serves two columns, the date and a binary call, with no confidence score, no neutral state and not one null in the flag since 2011-03-30.

15
10 names, rebalanced monthly, and the rank ships inside the row. coverage back to january 2017. three columns: date, ticker, rank. every basket it has ever published is still queryable. you can audit it row by row. this is PUBLIC
dark terminal card: curl api.alphanume.com/v1/quant-galore-momentum-index returning the row schema, date, ticker and a rank of 1 to 10, footer noting 10 names per rebalance, monthly, back to 2017-01-03

ALT dark terminal card: curl api.alphanume.com/v1/quant-galore-momentum-index returning the row schema, date, ticker and a rank of 1 to 10, footer noting 10 names per rebalance, monthly, back to 2017-01-03

12
your agent can now ask "which names have the least stable volatility" -> get_vol_of_vol(min_iv_vov_rank=0.95) -> every optionable name scored on how much its own vol moves asked 2026-10-08: 27 of 529 names cleared it. NLY 0.52, INO 0.41, UWMC 0.34. alphanume.fastmcp.app/mcp
Composed exchange card: the ask 'which names have the least stable volatility', the call get_vol_of_vol with min_iv_vov_rank=0.95, and three returned rows, NLY at iv_vov 0.52 and cross-sectional rank 1.0, INO at 0.41, UWMC at 0.34. Asked 2026-10-08: 27 of 529 names cleared the cutoff.

ALT Composed exchange card: the ask 'which names have the least stable volatility', the call get_vol_of_vol with min_iv_vov_rank=0.95, and three returned rows, NLY at iv_vov 0.52 and cross-sectional rank 1.0, INO at 0.41, UWMC at 0.34. Asked 2026-10-08: 27 of 529 names cleared the cutoff.

18
full-year 2025: the next-day mover model published 1,250 picks, 5 a session across 250 sessions. average realized absolute move: 6.9% median: 4.7% 48% of them moved more than 5% every pick's outcome is backfilled into the same rows. audit it yourself.
Data card: 6.9% in huge type. The average realized absolute move of the next-day mover model's 1,250 picks in full-year 2025, 5 picks a session across 250 sessions, with 48% of them moving more than 5%.

ALT Data card: 6.9% in huge type. The average realized absolute move of the next-day mover model's 1,250 picks in full-year 2025, 5 picks a session across 250 sessions, with 48% of them moving more than 5%.

8
the market-cap coverage map has 2 columns and 0 date columns. no date column means no trailing-window clamp. a free key sees all of it. every ticker carries its own first_available_date. the panel floor is 2015-09-30. check coverage before you query. it is PUBLIC.
Dark terminal card for GET /v1/historical-market-cap/tickers showing a two-column coverage map with 0 date columns and no free-tier clamp: ticker holds the symbol, first_available_date holds that name's own start, and the earliest date in the panel is 2015-09-30.

ALT Dark terminal card for GET /v1/historical-market-cap/tickers showing a two-column coverage map with 0 date columns and no free-tier clamp: ticker holds the symbol, first_available_date holds that name's own start, and the earliest date in the panel is 2015-09-30.

21
ask your agent "who told the SEC their debt just accelerated" -> get_corporate_default_events, the proving 8-K sentence quoted verbatim -> the 2026-08-12 pull surfaced a filer that elected not to pay $183 million in interest the feed reads EDGAR so you don't have to
Composed agent exchange: asking who told the SEC their debt just accelerated runs get_corporate_default_events and returns the SATS missed-interest event with its verbatim evidence quote.

ALT Composed agent exchange: asking who told the SEC their debt just accelerated runs get_corporate_default_events and returns the SATS missed-interest event with its verbatim evidence quote.

10
"expensive" is two questions. the iv-rank row answers both -> iv_rank: where the name's vol sits in its own 52-week range -> iv_rank_cs_ranked: where that sits against every other optionable name in the session one is history. the other is the tape
Data card: 2 in huge type, the number of ranks in every iv-rank row. iv_rank places a name inside its own 52-week range; iv_rank_cs_ranked places that same reading against every other optionable name in the session.

ALT Data card: 2 in huge type, the number of ranks in every iv-rank row. iv_rank places a name inside its own 52-week range; iv_rank_cs_ranked places that same reading against every other optionable name in the session.

11
Load more