Papers › Long-Horizon Forecasting of Complete Financial Statements with Forma

Long-Horizon Forecasting of Complete Financial Statements with Forma

11 Aug 2026arXiv:2608.11327added by Syntology

Travis L. Johnson, Jiannan Jiang, Soumyabrata Chaudhuri, Yihao Chen, Lauren Falvey, Donal O'Cofaigh

Title, abstract, authors and date from arXiv's metadata (CC0); this paper is not in the Papers with Code archive (frozen 2025-07-28).

Specialist training beats generalist scale when forecasting financial statements. To our knowledge, no prior work jointly forecasts complete financial statements beyond one year, yet in a discounted-cash-flow valuation most firm value sits past that window. We release ProForma-20Q, a reproducible benchmark for forecasting 78 statement line items 1-20 quarters ahead, for anonymized firms, from past statements and an industry code, scored by change-space R². On it, Forma, a transformer that reads statements as sets of (account, quarter, value) tuples and maximizes a masked-tuple Gaussian likelihood, beats every competitor we field: classical machine learning, chained gradient boosting, a zero-shot time-series foundation model, and frontier large language models. Its lead widens with horizon, where valuation needs accuracy most, and its Gaussian predictive intervals never under-cover. Forma's forecasts nearly satisfy accounting identities; exact coherence is recoverable at no statistically significant accuracy cost. Its tuple interface supports scenario analysis without retraining, and we show that pinning future revenue paths sharpens the rest of the statement.

PaperPDFCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 46 code samples harvested from 2 repositories linked to this paper: 34 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 4 honoured, 1 violated, 6 with no contract checked; 23 where Syntology's instrument failed) · 12 unverified. Where Syntology's instrument failed: 22 where our contract draft was wrong; 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: found in paper text by Syntology: 46 samples from 2 repositories, 34 ran (11 with no instrument failure; 23 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

forma-lab-mccombs/proforma-20q found in paper text by SyntologySyntology: 33 samples, 23 ran (15 instrument-failed) · 10 unverified report
forma-lab-mccombs/forma-release found in paper text by SyntologySyntology: 13 samples, 11 ran (8 instrument-failed) · 2 unverified report

Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

46 samples harvested: 34 ran (of which 0 constructed an object rather than computing a result; 11 with no instrument failure: 4 honoured, 1 violated, 6 with no contract checked; 23 where Syntology's instrument failed) · 12 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

4ran · honoured contract
1ran · violated contract
22ran · our draft was wrong
1ran; the contract check could not run
6ran
12unverified

Licence: 0 of the 46 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

forma-lab-mccombs/proforma-20q found in paper text by Syntology 33 samples, 23 ran (15 instrument-failed) · 10 unverified
canon_firm_ids src/proforma20q/truth_grid.py ran · honoured contract Apache-2.0 (permissive) · 301947e878db8b85 · report fixture: signature s:Series/1/object · returned Series · level recorded 2026-09-24
canon_origin src/proforma20q/truth_grid.py ran · our draft was wrong Apache-2.0 (permissive) · a78efa7957303097 · report fixture: signature s:Series/1/datetime64[us] · returned Series · level recorded 2026-09-24
coerce_declared_dtypes src/proforma20q/download.py ran · our draft was wrong Apache-2.0 (permissive) · da0e01698bd93b38 · report fixture: signature df:DataFrame/0/?, declared:dict/0/? · returned DataFrame · level recorded 2026-09-24
column_hashes src/proforma20q/checksums.py ran · our draft was wrong Apache-2.0 (permissive) · 2a66bf767312634b · report fixture: signature df:DataFrame/0/?, decimals:int/0/int · returned dict · level recorded 2026-09-24
column_stats src/proforma20q/checksums.py ran · our draft was wrong Apache-2.0 (permissive) · bb0cbbaa5075af5c · report fixture: signature df:DataFrame/0/?, decimals:int/0/int · returned dict · level recorded 2026-09-24
crps_per_entry src/proforma20q/evaluate.py ran Apache-2.0 (permissive) · a76eb758aeb383af · report fixture: signature res:ndarray/1/float, sigma:ndarray/1/float, family:str/0/str, df:NoneType/0/? · returned ndarray · level recorded 2026-09-24
dataset_suffix src/proforma20q/build.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 1ff0b04199cec615 · report fixture: signature feature_set:str/0/str, dataset_tag:str/0/str · returned str · level recorded 2026-09-24
de_regularize src/proforma20q/transforms.py ran fingerprinted Apache-2.0 (permissive) · b6b31a46d62faab7 · report fixture: signature x:ndarray/1/float, k:float/0/float, mu:float/0/float, sigma:float/0/float · returned ndarray · level recorded 2026-09-24
discover_targets src/proforma20q/truth_grid.py ran · our draft was wrong Apache-2.0 (permissive) · afcac8a20985153e · report fixture: signature truth_df:DataFrame/0/? · returned tuple · level recorded 2026-09-24
discover_targets src/proforma20q/baselines/common.py ran · our draft was wrong Apache-2.0 (permissive) · 0ca2d63496710dd4 · report fixture: signature df:DataFrame/0/? · returned list · level recorded 2026-09-24
feature_columns src/proforma20q/baselines/common.py ran · our draft was wrong Apache-2.0 (permissive) · bda3b8cd54df0aa7 · report fixture: signature df:MockDataFrame/0/? · returned list · level recorded 2026-09-24
firm_ff48_map src/proforma20q/build.py ran Apache-2.0 (permissive) · 5e4ca8cc3c6957dd · report fixture: returned dict · level recorded 2026-09-24
forecast_warnings src/proforma20q/schema.py ran · our draft was wrong Apache-2.0 (permissive) · 7dfe133fa504f9d1 · report fixture: signature df:DataFrame/0/? · returned list · level recorded 2026-09-24
hash_file in 7 papers src/proforma20q/checksums.py ran · our draft was wrong Apache-2.0 (permissive) · 9ee1a6d2b43a7d73 · report fixture: signature path:str/0/str · returned str · level recorded 2026-09-24
md5sum src/proforma20q/gated.py ran · our draft was wrong Apache-2.0 (permissive) · bbf363b68f36e91f · report fixture: signature path:str/0/str, chunk:int/0/int · returned str · level recorded 2026-09-24
nll_per_entry src/proforma20q/evaluate.py ran Apache-2.0 (permissive) · 95c7b7683a4e2768 · report fixture: signature res:ndarray/1/float, sigma:ndarray/1/float, family:str/0/str, df:NoneType/0/? · returned ndarray · level recorded 2026-09-24
normalize_columns in 5 papers src/proforma20q/schema.py ran · our draft was wrong Apache-2.0 (permissive) · 4e410c00366e3314 · report fixture: signature df:DataFrame/0/? · returned DataFrame · level recorded 2026-09-24
regularize in 6 papers src/proforma20q/transforms.py ran fingerprinted Apache-2.0 (permissive) · e30d565ae03f1693 · report fixture: signature v:ndarray/1/float, k:ndarray/1/float, mu:ndarray/1/float, sigma:ndarray/1/float · returned ndarray · level recorded 2026-09-24
repo_url src/proforma20q/gated.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 9f9f17bb754a92af · report fixture: signature repo_id:str/0/str, repo_type:str/0/str · returned str · level recorded 2026-09-24
seasonal_lag src/proforma20q/baselines/naive.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · d3ef37b65cee0533 · report fixture: signature horizon:int/0/int · returned int · level recorded 2026-09-24
sic_to_ff48 src/proforma20q/build.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 22a49639552b5660 · report fixture: returned Series · level recorded 2026-09-24
target_columns src/proforma20q/baselines/common.py ran · our draft was wrong Apache-2.0 (permissive) · fd688ce01b5d2cde · report fixture: signature df:DataFrame/0/?, targets:NoneType/0/? · returned list · level recorded 2026-09-24
transform in 65 papers
census: 6 implementations from 7 papers, 6 distinct outputs across 5 buckets, one shared input per bucket
src/proforma20q/transforms.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · c132b630ab682403 · report fixture: signature v:float/0/float, k:float/0/float · returned float64 · level recorded 2026-09-24
canonical_id_map_path src/proforma20q/config.py unverified Apache-2.0 (permissive) · f6f6c1c151f75660 · report
canonical_reg_stats_path src/proforma20q/config.py unverified Apache-2.0 (permissive) · cfeb85cf2fdf33ad · report
download in 53 papers src/proforma20q/download.py unverified Apache-2.0 (permissive) · 544c8030719e72aa · report
ensure_canonical_reg_stats src/proforma20q/config.py unverified Apache-2.0 (permissive) · 2cdf157d2a39a5f0 · report
fit_predict_sklearn src/proforma20q/baselines/sklearn_models.py unverified Apache-2.0 (permissive) · 7c0e589892b7b9a6 · report
fundq_projection src/proforma20q/download.py unverified Apache-2.0 (permissive) · 05916d507e4c68d1 · report
hf_download src/proforma20q/gated.py unverified Apache-2.0 (permissive) · 297e695e02aa59bd · report
predict_fade src/proforma20q/baselines/naive.py unverified Apache-2.0 (permissive) · fa8232f82d95fe27 · report
predict_naive src/proforma20q/baselines/naive.py unverified Apache-2.0 (permissive) · e0de3c23ec909c8b · report
validate_forecast src/proforma20q/schema.py unverified Apache-2.0 (permissive) · 66af5735d4ffc9e8 · report
forma-lab-mccombs/forma-release found in paper text by Syntology 13 samples, 11 ran (8 instrument-failed) · 2 unverified
canon_firm_ids src/forma/scoring/eval_cache.py ran · honoured contract Apache-2.0 (permissive) · 9a2e329955bc07b5 · report fixture: signature s:Series/1/str · returned Series · level recorded 2026-09-24
discover_feature_sets src/forma/scoring/evaluate.py ran · our draft was wrong Apache-2.0 (permissive) · c5693488b4bc49d5 · report fixture: signature forecasts_dir:PosixPath/0/?, feature_set_filter:NoneType/0/?, splits:list/1/str · returned list · level recorded 2026-09-24
discover_forecast_files src/forma/scoring/evaluate.py ran · our draft was wrong Apache-2.0 (permissive) · 0d7dad9545ad421f · report fixture: signature forecasts_dir:PosixPath/0/?, feature_set:NoneType/0/?, split:NoneType/0/? · returned list · level recorded 2026-09-24
file_fingerprint src/forma/scoring/eval_cache.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · dfd2a41a71fd5d10 · report fixture: signature path:str/0/str · returned dict · level recorded 2026-09-24
finalize_forecast_frame src/forma/models/forecast_io.py ran · our draft was wrong Apache-2.0 (permissive) · 75c0f05939196a6e · report fixture: signature forecasts_df:DataFrame/0/? · returned DataFrame · level recorded 2026-09-24
level_group_vals src/forma/scoring/evaluate.py ran · our draft was wrong Apache-2.0 (permissive) · d2e26630e38ad032 · report fixture: signature level_name:str/0/str, target:str/0/str, horizon:int/0/int · returned tuple · level recorded 2026-09-24
masked_mse_loss src/forma/models/ffnn.py ran · our draft was wrong Apache-2.0 (permissive) · 253b6e5cb59c7135 · report fixture: signature pred:Tensor/2/float, y:Tensor/2/float, normalization:str/0/str · returned Tensor · level recorded 2026-09-24
masked_nll_loss src/forma/models/ffnn.py ran Apache-2.0 (permissive) · 422231df525f5469 · report fixture: signature mu:Tensor/2/float, log_sigma_sq:Tensor/2/float, y:Tensor/2/float, loss_type:str/0/str, student_t_df:float/0/float, log_sigma_sq_min:float/0/float, log_sigma_sq_max:float/0/float, normalization:str/0/str, beta_nll:float/0/float · returned Tensor · level recorded 2026-09-24
quantile_col src/forma/models/chronos_ts.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 0ea2bc04ab003a55 · report fixture: signature level:float/0/float · returned str · level recorded 2026-09-24
warn_if_missing src/forma/_deps.py ran · violated contract Apache-2.0 (permissive) · 40250448ed22b573 · report fixture: signature feature:str/0/str · returned bool · level recorded 2026-09-24
wide_forecasts_to_tall src/forma/models/baselines.py ran · our draft was wrong Apache-2.0 (permissive) · e29052aecf3560cc · report fixture: signature forecast_wide:DataFrame/0/?, target_columns:list/1/str, sigma_wide:DataFrame/0/? · returned DataFrame · level recorded 2026-09-24
canon_quarters src/forma/scoring/eval_cache.py unverified Apache-2.0 (permissive) · 83b192c3d523d761 · report
require_proforma20q src/forma/_deps.py unverified Apache-2.0 (permissive) · 57403bcfbb636aa9 · report

Results from the paper

The Papers with Code archive ends with its 2025-07-28 snapshot. This paper's arXiv identifier, 2608.11327, was issued in August 2026, after that date, so the archive has no leaderboard rows for it.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections