3/8/24: Current plan in case of multiple parties: If multiple parties get shares, I'll resolve in proportion to the stake of all non-Chinese parties who get more than 5% (i.e. a reportable ownership share), unless someone buys a controlling stake, in which case it'll resolve to them alone.
Update 2026-02-02 (PST) (AI summary of creator comment): Resolution methodology for the 35.1% held by 8 smaller investors:
The creator will use a Monte Carlo simulation with a symmetric Dirichlet(1,1,1,1,1,1,1,1) prior to model uncertainty about how the 35.1% is distributed among the 8 smaller investors (Susquehanna, General Atlantic, Milner, Dragoneer, Dell Family Office, Alpha Wave Partners, Revolution, and NJJ Capital).
Expected resolution percentages:
Oracle: ~21.85%
Silver Lake: ~21.85%
MGX: ~21.85%
Each of the 8 smaller investors: ~4.30%
This approach accounts for the probability that some of the 8 smaller investors exceed the 5% threshold (expected ~2.9 investors qualify on average), while treating all 8 symmetrically given the lack of specific ownership information.
Update 2026-02-02 (PST) (AI summary of creator comment): The creator will resolve the market based on current information but may update the resolution if confirmed ownership information comes to light after the fact.
1,000https://newsroom.tiktok.com/announcement-from-the-new-tiktok-usds-joint-venture-llc?lang=en
TikTok USDS Joint Venture has three managing investors, Silver Lake, Oracle and MGX, each holding 15%. Completing the full consortium of investors are: Dell Family Office, the investment firm of Michael Dell, Founder, Chairman and CEO of Dell Technologies; Vastmere Strategic Investments, LLC, an affiliate of Susquehanna International Group, LLP; Alpha Wave Partners; Revolution; Merritt Way, LLC controlled and managed by partners of Dragoneer; Via Nova, an affiliate of General Atlantic; Virgo LI, Inc., investment arm of a foundation established by Yuri and Julia Milner in support of science; and NJJ Capital, the family office of Xavier Niel, a French entrepreneur and pioneer in telecommunications. ByteDance retains 19.9% of the Joint Venture.
By my maths:
19.9% - Bytedance
15% - Silver Lake
15% - Oracle
15% - MGX
35.1% - Remainder, split among 8 other parties = 4.3875% each on average
I'll resolve in proportion to the stake of all non-Chinese parties who get more than 5%
I parse this as 1/3 Oracle/MGX/Silver Lake, unless there's evidence that any of the other parties have uneven ownership (which seems pretty plausible)
@draaglom I’d imagine Susquehanna will hold more than 5%, but haven’t seen an exact figure yet.
@ChadCotty I agree it's premature to resolve this given that the investors are likely not equal (in particular Susquehanna is likely to clear the threshold)
@TotalVerb also note the original reporting here reserved 30.1% for current affiliates of ByteDance investors: https://www.yicaiglobal.com/news/bytedance-to-retain-largest-stake-in-new-us-data-security-jv-set-up-to-avoid-tiktok-ban
When I ask Gemini to identify which of the 8 new investors were linked to Bytedance investors, it suggests only 4 of them are — which is again pretty strong evidence that e.g. SIG has higher than 5% ownership.
To my knowledge, we're not actively expecting any further reporting -- TikTok USDS Joint Venture LLC is, I believe, a private Delaware corporation so probably doesn't have to declare anything publicly about its ownership. So I'd be in favour of resolving rather than waiting.
One somewhat principled way of doing this would be to attribute 30.1 evenly between prior investors that are mentioned in the USDS venture.
https://www.thewirechina.com/wp-content/uploads/2020/08/Who-Really-Owns-ByteDance.pdf
I make this Susquehanna, Milner, General Atlantic, and Dragoneer
Rounding the rest of the ownership to 0, I think this makes:
20% Oracle (15/(30.1+45))
20% Silverlake
20% MGX
10% Susquehanna (1/4)*(30.1)/(30.1+45)
10% General Atlantic
20% Other (2/4)*(30.1)/(30.1+45)
Is there a timeline where investors would have to disclose ownership shares above a certain percent? Or does this not apply for private companies? Agree makes sense to resolve if no concrete further info to expect.
@CalebW I don't really know much about US financial disclosure law but ChatGPT says they don't have to disclose anything
https://chatgpt.com/share/6980da93-b07c-800c-8d28-b1a20b990977
@draaglom I think they would need at least a 5% stake to be included in the distribution under the resolution criteria, but I’d be in favor of including the investors that almost assuredly have at least 5%.
@ChadCotty Right, the argument I'm making is that if 30.1 went to the existing investors, and we know of 4 existing investors in the list, then they got 7.5% each roughly.
Do we have any other principled ways of assessing if an investor has 5+%?
(I've closed the market given the discussion seems to mainly be about resolution at this point. Let me know if anyone feels hard done by by this. I'll also plan to forfeit any profits of mine that might be attributable to resolution discretion.)
Also appreciate that in hindsight the 5% thing was maybe a bit foolish - was trying to account for the possibility of like, an IPO where much of the ownership was distributed.
My current thinking:
- reasonable to want to, but ultimately speculation which is hard to do objectively, to estimate which smaller investors are likely to have larger shares
- want to account for this uncertainty in a principled way
From some chatbot-ing, proposed approach below (I am taking it at its word on the math):
Summary: Expected Resolution Using Uninformative Prior
The Problem
We know Oracle, Silver Lake, and MGX each hold 15% of TikTok US, and ByteDance retains 19.9%. The remaining 35.1% is split among 8 other investors (Susquehanna, General Atlantic, Milner, Dragoneer, Dell Family Office, Alpha Wave Partners, Revolution, and NJJ Capital), but we don't know the exact allocation. The market resolves proportionally to all non-Chinese parties holding more than 5%, so we need to determine which of these 8 investors likely exceed the threshold and what the expected resolution should be.
The Approach
Without specific information privileging any of the 8 investors, we used a symmetric Dirichlet(1,1,1,1,1,1,1,1) prior to model our uncertainty about how the 35.1% is distributed. This is the standard uninformative prior for an unknown distribution over multiple parties—it treats every possible allocation as equally likely a priori. For each investor to qualify for resolution, their share of the 35.1% pool must exceed 14.25% (since 0.351 × 0.1425 = 5%).
Under this prior, each of the 8 investors has approximately a 36% probability of individually exceeding the 5% threshold. The expected number qualifying is around 2.9 investors. However, because the qualification threshold creates a nonlinear filter and we must normalize within each possible world, we cannot simply use ratios of expectations. Instead, the correct approach is Monte Carlo simulation: for each draw from the Dirichlet, we determine which investors qualify (>5%), compute their pro-rata resolution shares in that world, then average across all simulated worlds.
The Results
Running this Monte Carlo simulation yields the expected resolution:
Oracle: 21.85%
Silver Lake: 21.85%
MGX: 21.85%
Each of the 8 other investors: 4.30%
By symmetry, all 8 unknown investors have identical expected resolutions. The three named managing investors (Oracle, Silver Lake, MGX) each drop from the naive 33.3% (if no others qualified) to about 22%, reflecting that on average 2-3 of the other investors will exceed 5% and claim their share. This represents the expected resolution accounting for our uncertainty about the underlying ownership structure, given only that 8 parties split 35.1% and applying the market's >5% qualification rule.
This gives a final market resolution of Oracle/Silver Lake/MGX at ~22% each, Susquehanna/General Atlantic at ~4.3% each, and Other at ~26%.
@CalebW If indeed these publicly listed companies own the shares (rather than their billionaire owners, which I think equally likely, and I think would mean the market should not count things answers like Oracle) then they have to disclose if it is significant:
- 8-K if the deal was significant. I don't know Oracles threshold (using Oracle as an example, I know there is already resolution here)
- Annual and Quarterly reports from Oracle might list it.
Here's a similar market but it's just a yes/no on each company or person owning a stake of the US version: https://manifold.markets/IsaacLiu/as-of-april-2025-will-the-following