📉 The Part of the Spreadsheet I Can't Get for Free September 4, 2026 · https://tavi-blog.github.io/the-part-of-the-spreadsheet-i-cant-get-for-free/

I pulled up the Broadcom print the morning after it hit and read the numbers twice because they didn't square with the stock chart sitting next to them. Revenue up 86 percent year over year. AI chip revenue up 221 percent. Both lines beat what analysts were modeling, not by a rounding error, by a real margin. And the stock fell anyway, because the guidance for next quarter came in a hair under what the Street wanted. Not below the prior quarter. Not below this quarter. Below an expectation that hadn't even been published anywhere I could have read it in advance.

That gap is the whole problem with trying to read these moves from the outside. I've spent the better part of this year telling myself that if I just built the right dashboard, I'd stop finding out about the good trades a month after they happened. Free historical prices, free earnings data, free filings, all of that is one API call away now. What isn't free, and what actually explains why a company can beat every number that gets printed in a headline and still get sold off, is the number that was never public in the first place: the whisper estimate, the buy-side consensus that sits somewhere between the published analyst average and what the biggest desks were quietly modeling. That number lives on a terminal I don't have access to. The public consensus figure I can pull from a free API is already a stale, rounded proxy for the thing that actually moves the stock in the first ninety minutes of after-hours trading.

I want to be fair to the other side of this, because it's the more interesting argument. You could say the information isn't really hidden so much as it's synthesized, and that synthesis is available to anyone through the price action itself. Options open interest, implied volatility going into the print, the way the stock had run into earnings, all of that is public and all of it is a decent proxy for what the market already expected. A trader who reads volume and skew instead of waiting on an earnings call transcript isn't missing the signal, they're reading a different, arguably better version of it. That's a real point. The professional edge in this business was never just "having the number nobody else has," it's having the discipline to sit with the position while retail investors like me are still refreshing a headline trying to figure out why a beat wasn't a beat.

But that argument assumes I'd know what to do with options flow if I had it, and I wouldn't. I scoped this tool back when semiconductors were running and I was annoyed at how small my position was. The plan was straightforward on paper: pull price history, pull filings, pull macro series, layer in some historical context, and get a dashboard that flags when a sector's setup looks like the last time it moved. What actually stalled the project wasn't the code. It was realizing, somewhere in the scoping, that the inputs I could get for free were the ones that explain a move after it happens, and the inputs that would have told me anything before it happened were locked behind subscriptions built for people managing other people's money, not a few hundred dollars of TFSA room.

So the honest version of what I'm building, if I ever get back to it, isn't a signal generator. It's a tool that tells me when I'm not equipped to have an opinion on the next ninety minutes and should stop pretending otherwise. That's a less exciting pitch than "predictive dashboard," but it's the accurate one. Broadcom's quarter didn't teach me anything about semiconductors I didn't already believe. It taught me that the version of this hobby available to someone reading free filings at night is a slower, dumber game than the one being played by whoever set that guidance number as the bar to clear, and that the useful move is to stay in the position for the multi-year thesis and stop treating the day-after reaction as information I was ever going to have in time to use.