30-year hedge fund veteran. Earnings calls, management promises and trading research put to the test. Behind Earnings Call Intelligence & Finance Paper Mill.
Bay Area
Joined June 2012
- Tweets2K
- Following1.4K
- Followers2.3K
- Likes174
Pinned Tweet
I read earnings calls and test investment research.
What changed? What supports the claim? What should we check next?
Follow for one useful finding each weekday: management promises, business economics and research put to the test.
Two examples below.
Before I score a management promise, I write down:
1. The exact target.
2. The deadline.
3. The metric and accounting basis.
4. The evidence needed to resolve it.
An open promise is not a missed promise. An omitted update is a reason to investigate.
A drawdown number needs a clock.
Our historical SPY simulation, Jan 2007-Jun 2026:
Daily observations: 39.1% maximum drawdown.
Month-end observations: 20.7%.
Same trades. Different measurement dates.
Includes commissions; excludes financing and auction slippage.
Anthropic's $895.4B spending commitment is monumental. About $69.8B is due over the next 12 months; the rest stretches across years, with some arrangements conditional or cancelable.
Give them credit: $131.8B in cash and short-term investments is a substantial cushion.
But Q3 revenue grew 47% sequentially while compute costs grew 77%. Adjusted operating margin fell from 19.3% to 14.5%.
The test is whether cheaper long-term capacity can turn extraordinary demand into durable profits. Cash buys time. Margins determine what the buildout earns.
Full Q3 review: spending commitments, the cash reserve, margin pressure and the financing terms behind the buildout. Free to read.
earningscallintelligence.sub…
From $ABM's Q3 FY26 cash-flow statement:
Quarterly free cash flow: $128.4M vs $150.2M a year earlier.
Nine-month free cash flow: $199.6M vs $42.4M.
One period fell. The other improved sharply.
Before calling cash conversion stronger, specify which period you mean.
Does month-end deserve a special trade?
In our SPY house test, Jan 2016-Sep 2026, the last session plus the next month's first three beat other days by just 1.09 bps per day before costs.
41% of 500 random four-session placements did better.
The comparison matters.
$WDC's Q1 FY27 revenue grew 55% year over year.
The beat versus management's prior guidance midpoint was about 7%.
Growth compares the business with last year. A guidance beat compares delivery with the target.
I want both numbers before judging the quarter.
Earnings Call Intelligence: $WDC's Q1 FY27 revenue grew 55% year over year. Its beat versus the prior guidance midpoint was about 7%.
Both matter. They answer different questions.
earningscallintelligence.com…
Finance Paper Mill: one simulated portfolio, two drawdowns.
Daily: 39.1%.
Month-end: 20.7%.
Same trades, different observation dates. Includes commissions; excludes financing and auction slippage.
financepapermill.com/guides/…
300 stocks can still share the same underlying risk. Chehab and colleagues jointly estimate factor exposures and links between assets across three markets. No FPM replication yet. Our review asks whether this map improves diversification in stressed markets.
Free preview: the diversification idea and original paper. Paid analysis on Substack examines the portfolio comparison, stressed markets and replication constraints. financepapermill.com/researc…
Does a new ranking deserve a new trade? Igor Halperin combines correlation distance with changes in return and volatility ranks in three matrices. No FPM replication yet. Our review examines trading bands and the benchmark needed to judge the idea.
Free preview: the three-matrix representation and original paper. Paid analysis on Substack examines trading bands, familiar exposures, universe selection and significance checks. financepapermill.com/researc…
2,175 parameters versus 39,586. Bongiorno and colleagues compress a neural risk model to target volatility drag. Smaller does not establish safe leverage. No FPM replication yet. Our review examines financing, drawdowns and forced liquidations.
Free preview: the model-size comparison and original paper. Paid analysis on Substack examines leveraged portfolios, financing, forced liquidations and execution assumptions. financepapermill.com/researc…
SpaceX is funding its next buildout ahead of GPU deliveries. Elon says the $40bn financing is primarily callable debt for GPUs arriving in Q1 and Q2. A concrete growth checkpoint: turning those deliveries into live capacity and revenue.
Predicting a return and ranking a stock ask different questions. Our earnings-model comparison changed return levels to within-month ranks while holding the model and evaluation rules fixed. The analysis examines what that single training choice changes.
Free preview: the target choice and original paper. Paid analysis on Substack compares our earnings-model results across broad and liquid universes, separately from the authors' study. financepapermill.com/researc…
Who is missing from your insider-trading screen? Hvide and Nielsen study own-company purchases below the top executive tier using Norwegian registers. No FPM return test was completed. Our review asks what public filings can actually reveal.
Free preview: the study and its data-access limit. Paid analysis on Substack examines the observed population, registers versus public filings, and requirements for a credible return test. financepapermill.com/researc…
