How does Salesdamus forecast sales?

It fits proven statistical models to your history — linear trend, Holt-Winters seasonality and more — then backtests them and uses whichever one predicted your recent past best. Bootstrap and Monte Carlo simulation turn that into honest 95% confidence intervals.

Completely. No account, no credit card, no catch. Salesdamus runs entirely in your browser, and your data never touches a server.

How much data do I need?

Two periods to start. Twelve or more unlocks reliable trends and accuracy scores; twenty-four or more lets it spot seasonality and forecast holiday and quarterly cycles.

As accurate as your data allows — and it's transparent about it. Every forecast comes with a backtested error rate and a confidence interval, so you know exactly how much to trust the number.

Can I trust the forecast?

Treat it as a grounded estimate, not a prophecy. The 95% interval shows the realistic range, and it widens further out because the future gets fuzzier the longer you look.

Forecasting glossary

Every term Salesdamus uses, in plain English. No PhD required.

Forecast horizon: How many periods into the future you project. Accuracy and confidence degrade the further out you go.
OLS regression: Ordinary least squares — fits a line by minimizing the sum of squared residuals. The basis of the linear-trend forecast.
Holt-Winters: An exponential-smoothing method that models level, trend and seasonality together. Best for data with a repeating cycle.
Seasonality: A pattern that repeats over a fixed period (e.g. 12 months). Captured by the seasonal model, ignored by a plain trend.
Bayesian shrinkage: Pulling an estimate toward a prior (here, the long-run mean) to stabilize forecasts when data is scarce.
Confidence interval (95%): A range that should contain the true value about 95% of the time. Salesdamus shows it for every forecast.
Prediction interval: The plausible range for a future observation. It widens with the horizon because uncertainty compounds.
Bootstrap: Resampling the model's residuals many times to estimate how much a forecast could vary — no distribution assumed.
Monte Carlo simulation: Drawing many random samples from a forecast's distribution to map out its range of outcomes.
Backtest: Hiding recent periods, forecasting them, and measuring the error — how Salesdamus picks the best model.
R² (R-squared): The share of variance in the data explained by the model. 1.0 is perfect; near 0 means little explanatory power.
MAPE: Mean absolute percentage error — the average size of the forecast error as a percentage. Lower is better.
p-value: The probability the observed trend could appear by chance. Below 0.05 is the usual bar for 'statistically significant'.
Residual: The gap between an actual value and the model's fitted value. Small, patternless residuals signal a good fit.

→ Read how these methods work