Size went up when the price went up. That is not conviction, it is a scheduleDušan Šenkypl & Jan Barta · Pale Fire Capital SICAV a.s. · Coursera, Inc. (NYSE: COUR)
| Date | Shares | Wtd avg | Range | Owned after | % of class |
|---|---|---|---|---|---|
| 3 Nov 2026 | 2,526,702 | $3.7209 | $3.7054–3.7738 | 31,333,117 | 11.85% |
| 4 Nov 2026 | 2,698,400 | $3.8537 | $3.7970–3.8795 | 34,031,517 | 12.87% |
| 5 Nov 2026 | 3,130,149 | $3.7220 | $3.6700–3.7675 | 37,161,666 | 14.06% |
| Total | 8,355,251 | ~$3.764 | $3.6700–3.8795 | +29.0% | from 10.9% |
- What they did
- Bought 8,355,251 Coursera shares for about $31.45M across 3, 4 and 5 November 2026, taking the position from 28,806,415 to 37,161,666. The buying began five sessions after Coursera’s 29 October Q3 report, which beat — revenue $376.7M, up 94% year on year and 2.2% ahead of consensus; adjusted EPS $0.21 against $0.18 — but guided Q4 revenue to a $362M midpoint, 2.7% below the $372M expected. The stock fell about 8% to $4.27 on the print and kept sliding to $3.67 by 5 November.
Consensus read
The largest shareholder stepping into a guidance-driven selloff. Coursera beat on every reported line and sold off on a 2.7% revenue guide miss; at $3.72 the company is capitalised at roughly $1.24bn against $376.7M of quarterly revenue. Pale Fire, which already owned 10.9%, put another $31.45M in over three days.
Read as the house view that the market mispriced a guidance quibble, pressed with real size by the holder who knows the asset best.
Contrarian read
The decision was taken before the first trade and the price had little to do with it. This is a block accumulation sized to a target stake and worked mechanically — a participation programme, not a reaction.
What makes it checkable: the daily share count rises monotonically while the price does not. They put 6.8% more shares to work on 4 November at a 3.6% higher average than on 3 November. A buyer reacting to cheapness does the opposite of that.
Evidence that discriminates
- Size is monotonic; price is not. 2,526,702 → 2,698,400 → 3,130,149 shares, against $3.7209 → $3.8537 → $3.7220. The middle day was the most expensive and got more shares than the cheapest first day. If the thesis were “this is cheap, buy more when cheaper,” the ordering would invert at least once. It never does. A steadily ramping daily quantity is the signature of an order being worked, not a view being expressed.
- The September tranche is inverted the same way, in the opposite direction. 25 September: 1,332,742 shares at $5.0639. 29 September: 1,000,000 shares at $4.6062 — 25% fewer shares at a 9.0% lower price. Two separate episodes, six weeks apart, in both of which the quantity ignores the discount. That is two independent observations of the same indifference, which is what lifts this above a three-day coincidence.
- The same balance sheet was buying something else at the same time. PFC SICAV bought 986,557 Dole shares on 2–4 November for about $12.9M, overlapping these sessions exactly. $44.4M into two unrelated businesses — an Irish-domiciled fresh-produce company and a US online-education platform — inside four days. Two simultaneous high-conviction stock calls is a strained explanation. One allocator deploying a determined amount of capital is not.
- They disclaim, and they hold nothing personally. Every row is indirect, held by the SICAV; Šenkypl and Barta appear as control persons and each disclaims beneficial ownership except to the extent of pecuniary interest. The 29 September 13G/A reports sole voting power 0.00 and shared voting power 28,806,415 for every reporting person, with the Item 10 certification that the shares are not held to change or influence control. Neither man is buying Coursera. A fund they control is.
Where I am inferring and where I am short. The participation-programme reading would be much stronger with daily volume — if the three days’ share counts are a near-constant fraction of each day’s traded volume, the case is close to settled. I did not obtain volume data, so the monotonic-size argument stands on its own and is suggestive rather than conclusive. Separately, Investing.com describes Pale Fire Capital SE as “a 10% owner and director” of Coursera; the Form 4 as rendered checks only the 10% Owner box. I could not resolve that and have not relied on a board seat either way.
What would falsify me
The per-price execution detail the filers undertake in footnotes F3–F5 to provide on request. If the fills cluster at each day’s lows rather than spreading across the stated ranges, they were working the price and not a schedule, and this entry is wrong. A Form 4 in the next few weeks showing a much larger block bought on a sharply lower day — size moving inversely to price — would do the same. So would any public statement from Pale Fire framing the purchase as a response to the 29 October guidance.
Pattern vs their own historyPale Fire crossed into Coursera disclosure with a Form 3 at about 26.5 million shares and amended to 10.9% on 29 September. Prior tranches were in the $10–11M range. This one is $31.45M — 2.8× the largest previous — and it lifts them past 14% of the class. The structure is unchanged from the Dole pattern: Czech SICAV holds, investični společnost manages, the two men appear as controllers and disclaim. What changed this week is tempo and simultaneity across two positions.