Epam Systems, Inc.: possible revenue paths

Model computed 2026-11-04 · Data cutoff 2026-06-30

Latest verified revenue: $1.415B for the quarter ended 2026-06-30. Reported source.

Forecasts are conditional on the listed data cutoff. Sources were refreshed for this model run on 2026-11-04; the cutoff above identifies the latest verified revenue observation included in the fit. Some forecast quarters may already have ended by the computation date. Read each target date before treating an estimate as a future-quarter outlook.

PyMC seasonal AR(1) quarterly log-revenue growth with Student-t shocks; posterior predictive paths include parameter uncertainty and new shocks at every horizon.

Each line is one possible future revenue path. The 80% and 95% intervals are pointwise model probabilities, not simultaneous coverage for the whole path. The median is not an expected dollar mean. Open full-size chart.

Future period-end dates are approximate: month-end issuers retain month-end dates; weekly issuers advance by 13 weeks. Future 53-week adjustments are not known.

Historical forecast checks

Each historical cutoff refits the model using only earlier observations. Across the available outcomes, the 80% intervals covered 10/16 and the 95% intervals covered 13/16. Small samples and overlapping outcomes do not establish nominal coverage; covering every outcome can indicate overly wide intervals.

All point-forecast errors use the same historical cases. A simple baseline may have lower error than the Bayesian median; this rollout does not establish that the Bayesian model is superior for every company.

Up to four historical cutoffs spaced four quarters apart, with at least 16 training quarters; full four-quarter outcomes required. Fixed specification, no winner selection.

These are retrospective checks using latest-restated data, not forecasts archived before their outcomes. Results from DDOG or another company are not transferred to this company.

Management guidance versus reported revenue

Reviewed the 8 most recent located earnings filings. This incomplete earliest-located sample can omit earlier announcements, revisions and call-only guidance. Only consolidated quarterly reported-USD revenue is compared; annual and constant-currency growth outlooks are excluded. Q4 2024 guidance is excluded because it omitted the pending First Derivative acquisition while actual revenue included it after closing. That scope difference is not counted as management forecast bias.

Management ranges are stated targets, not probability intervals. The revenue model is fitted independently; verified guidance is superimposed for comparison. Historical bias uses the earliest located guidance for each quarter: (guidance midpoint − actual revenue) / actual revenue. Positive values mean guidance above actual results; negative values mean guidance below actual results. This describes outcomes, not management intent.

6 matched quarters · mean guidance bias -1.18% · mean absolute error 1.18%. Actual revenue fell inside 16.7% of 6 published ranges.

This is a descriptive sample, not evidence of a permanent management bias or a calibrated probability range. Call-only guidance and earlier revisions may be missing; guidance and results can reflect different business assumptions.

Preserved Bayesian forecast vintage

Computed 2026-11-04 using revenue through 2026-06-30. These original numbers are retained unchanged to allow subsequent scoring; they were not refitted with the new results. Forecasts computed after a target quarter ended are retrospective estimates, not pre-release predictions.