We Counted “Excited” on 174,996 Earnings Calls. It Is a Sell Signal.
Every earnings call has the same moment. A question lands, and the CEO answers it with the word “excited”. We are excited about the pipeline. We are really excited about the second half. Honestly, we could not be more excited. It is the most reliable tic in corporate English, and after a while you stop hearing it.
We did not stop hearing it. We counted it, on every earnings call since 2016, 174,996 of them, and then we did the thing that turns a word count into a finding: we waited a year and looked at the stock.
- Excitement is a sell signal, and it has been since 2021. Among calls from 2021 onward, the median stock whose management never said “excited” trailed the S&P 500 by 11 points over the next year. Say it eleven or more times, and the median trailed by 25 points. The share that beat the index at all fell from 37% to 29%.
- The market does not notice on the day. The morning after the call, 48% of stocks went up whether the CEO was excited or not. It takes a year for the word to cost anything.
- The most excited CEOs in America, per call since 2024: Salesforce (21), Netflix (19), Rivian (15), Walmart (15), Carvana (14), Apple (14), Robinhood (12). The record is Acuity Brands, a lighting company, at 74 in one call.
- The twist: companies whose management had been saying it eight-plus times a call and then dropped to one or none were the best group we found. 42% beat the index the following year, against 34% for the companies that stayed excited.
- Before 2021 it was the other way round. From 2016 to 2020, excited calls did slightly better on average. The word did not change; the market did.
Everyone is excited
The word is everywhere and getting more so. In 2016 the average earnings call contained “excited”, “exciting” or “thrilled” 1.8 times and 58% of calls said it at least once. In 2026 so far the average is 3.0 and 70% of calls say it.1 The chart below is the average count per call, by year:
Notice the step in 2021. Excitement jumped by half in a single year, the year of free money and record IPOs, and it has never come back down. Keep that year in mind. It matters later.
Who says it most? We ranked well-known companies by their average count per call since the start of 2024:
| Company | Per call | Most on one call | Per 1,000 words |
|---|---|---|---|
| Oklo | 21.1 | 48 | 1.39 |
| Salesforce | 21.1 | 34 | 1.83 |
| Netflix | 18.8 | 27 | 2.69 |
| Rivian | 15.2 | 22 | 1.62 |
| Walmart | 14.9 | 25 | 1.42 |
| Carvana | 14.1 | 29 | 1.44 |
| Apple | 14.0 | 21 | 1.70 |
| Target | 12.4 | 24 | 1.18 |
| Robinhood | 11.8 | 23 | 1.33 |
| PayPal | 11.8 | 20 | 1.26 |
Netflix is the densest: one “excited” every 370 words of management speech, on calls that are short to begin with. The all-time record belongs to Acuity Brands, which said it 74 times on its July 2021 call. Walmart managed 46 in February 2021. Snap, 38, in April 2021. You will notice a year recurring.
Excite-nomics
Counting is the easy part. The question is whether the word tells you anything about the stock, and here the honest approach is the one we used for the AI study: sort the calls by the count, follow each stock for twelve months against the S&P 500, and report the median and the hit rate alongside the mean, because the mean lies on its own.2
We have twelve-month outcomes for 132,564 calls. Here is the median stock, against the index, for calls since 2021, sorted by how many times management said the word:
It is monotonic. Every step up in excitement is a step down in the typical outcome, from −11% to −25%. The hit rate walks the same path:
And unlike the AI study, where the mean and the median disagreed, here they agree:
Since 2021, a portfolio of every company whose management said “excited” eleven or more times on the call would have trailed the S&P 500 by 17 points a year on average. The silent ones trailed by 6.
In case you are wondering whether this is just small companies losing to a capitalisation-weighted index: partly, yes, which is why the level of every bar is negative. But the gradient survives the crude fix. Measured in raw returns, with no index at all, the median stock whose management never said the word was flat over the following year and 50% of them went up. The median stock in the most excited group, more than two mentions per thousand words, fell 22%, and only a third went up.
Day one, year one
You might expect the market to punish the brochure on the spot. It does not. The trading day after the call, 48.5% of stocks went up in every excitement bucket, and the average move was within a quarter of a percent of zero across the board.3 The excited calls are more volatile, with more 10% pops and more 10% drops, but no direction.
Three months out, a small gap: since 2021 the median excited stock trailed by 6 points, the silent by 3. A year out, the full 14-point gap. Whatever excitement is telling you, the market takes a year to hear it.
Accidental brochure, again
If this sounds familiar, it should. In the AI study we found that the word changed the speaker more than the stock. Excitement does both, and it changes the speaker the same way:
Promotion rises from 4.5 to 6.3 as the count climbs; candor slips from 7.0 to 6.6; evasion creeps up. A call with eleven “exciteds” in it is, on average, a more promotional, less candid, slightly more evasive call. That is not a shock. What is a shock is that a word this cheap to say carries this much information about the year to come.
When they stop
Here is the cut we did not expect. We compared each call to the same company’s previous call:
The companies that stopped being excited, eight-plus mentions one quarter and one or none the next, were the best group in the entire dataset: a mean of +4.3% against the index and 42% ahead a year later. The companies that stayed excited were among the worst, at −6.7% and 34%. We can think of two readings. The charitable one is that a management that drops the adjectives has something concrete to say instead. The uncharitable one is that the excitement was doing work the numbers could not, and when it stopped, so did the need for it. Either way, the silence is the bullish signal, which is not what the investor-relations handbook says.
It is not all clean
No study of 132,564 stocks is free of trade-offs, so here are ours.
The big one is the regime. From 2016 to 2020 the relationship ran the other way: the most excited calls returned a mean of +18% against the index, the silent ones +5%, with medians about equal. The sell signal is a post-2020 phenomenon, and it has held in each of the five years since, including 2025. Five years is enough to take seriously and not enough to call a law. It may simply be that excited managements cluster in the growth and small-cap names the market fell in love with in 2021 and out of love with afterwards; excitement would then be a symptom, not a cause. We cannot rule that out and we would not bet against it.
The rest is the usual. Our price histories carry survivorship bias, which flatters every bucket about equally. The median trails the index everywhere partly because the index is capitalisation-weighted and the typical listed company is small; the gradient is the finding, not the level. The word count is a literal count of “excited”, “exciting” and “thrilled” in management speech and knows nothing about sarcasm, though we have yet to hear a sarcastic CFO. The tone scores come from a language model reading the same transcripts. And the 2026 cohort, the most excited on record, has not matured.
What we are watching
Seventy percent of this year’s calls said the word and the average call said it three times, both records. Their twelve-month outcomes arrive through 2027 and we will publish them whichever way they go. In the meantime the ranking shows every company’s count, updated after every call, and each stock page shows the number next to the typical range for the market. If your company is above the range, it might be worth asking what the excitement is for.
- The count is a literal match on excited, exciting, excitement and thrilled in management speech only, analysts and the operator excluded. Calls under 2,500 words or without a parsed Q&A are excluded from rankings and from the outcome analysis. ↩
- Outcome window: from the first close after the call to the same date twelve months later, minus the S&P 500 (SPY) over the same window. Prices from the Artul.ai archive. Outcomes exist for 132,564 of the 174,996 calls; the rest are too recent or lack prices. Calls since 2021 with a full year of prices: 61,663. Buckets by count: 0 (n = 17,193), 1–2 (19,030), 3–5 (14,176), 6–10 (8,241), 11+ (3,023). ↩
- Reaction: last close before the call to first close after it. Share of stocks up the next day: 48.5%, 48.6%, 48.3%, 48.7%, 48.5% across the five buckets. Share moving more than 10% either way: 20% for one to two mentions, 28% for eleven or more. ↩