Every figure this model derived for GOTU, the value it took, and
the element in the company's own XBRL it was read from — so any of it can be checked
against the filing rather than taken on trust.
As at 2025-12-31.
The figures, and where each came from
This is the part no other screener shows. Each figure below names the XBRL element it
was read from, so it can be checked against the filing itself.
Figure
Value
XBRL tag
As at
capex
24,238,000
PaymentsToAcquireProductiveAssets
2025-12-31
cash
472,607,000
CashAndCashEquivalentsAtCarryingValue
2025-12-31
current assets
569,210,000
AssetsCurrent
2025-12-31
current liabilities
607,097,000
LiabilitiesCurrent
2025-12-31
equity
179,284,000
StockholdersEquity
2025-12-31
ocf
59,501,000
NetCashProvidedByUsedInOperatingActivities
2025-12-31
revenue
878,977,000
Revenues
2025-12-31
sbc
5,608,000
ShareBasedCompensation
2025-12-31
The ratios
The figures themselves rather than their scores — what somebody looking up
GOTU on one particular measure came for.
Ratio
Value
cash operating margin
-5.85
current ratio
2.03
current ratio reported
0.94
deferred revenue growth
27.98
dilution pct
-4.84
fcf margin
4.01
growth cv
5.82
operating margin
-8.19
owner earnings margin
3.37
revenue growth
40.9
revenue growth per share
48.07
rule of 40
32.71
sbc pct revenue
0.64
What to be careful about
What this model itself distrusts about GOTU's figures. Written by the
derivation as it ran, not added afterwards.
revenue is annual, not TTM
operating income is annual, not TTM
no market cap; set `market_cap` in overrides.yaml
share count is from 2025-12-31 (242 days old, weighted average diluted); a stock split since then would make the market cap wrong. Set `market_cap` in overrides.yaml to be sure.
no forward EPS estimate; growth rests on filings only. Analyst consensus is not published in XBRL, so it has to be fetched from a data provider or set by hand in overrides.yaml as `eps_growth_2y`
quality rests on 4 of 5 inputs; not tagged: roic
valuation DROPPED from the total: only 0 of 2 inputs could be computed (ev sales, ev owner earnings not tagged), so the score is not meaningful. Its weight is spread over the other dimensions
resilience rests on 2 of 3 inputs; not tagged: interest cover
no interest expense tagged while 5M of debt is on the balance sheet, so interest cover is unmeasured rather than perfect