← Back to Dashboard
	FORECAST COMPARISON & MODEL SELECTION
	Side-by-side 12-month forecast comparison across all methods with accuracy ranking
	12-MONTH FORWARD FORECAST (Jan – Dec 2026)
	Period	Month	Trend Fcst	Seasonal Idx	Trend+Seasonal	SMA Ext.	DES Ext.	HW Forecast	Consensus Avg
	37	Jan	6,198	0.823	5,099	5,947	6,600	5,112	5,690
	38	Feb	6,228	0.845	5,261	5,977	6,666	5,285	5,797
	39	Mar	6,258	0.924	5,780	6,007	6,732	5,824	6,086
	40	Apr	6,288	1.032	6,491	6,037	6,798	6,550	6,469
	41	May	6,318	1.109	7,007	6,068	6,864	7,053	6,748
	42	Jun	6,348	1.158	7,354	6,098	6,930	7,412	6,948
	43	Jul	6,379	1.037	6,613	6,128	6,996	6,675	6,603
	44	Aug	6,409	0.977	6,264	6,158	7,063	6,311	6,449
	45	Sep	6,439	0.918	5,914	6,188	7,129	5,957	6,297
	46	Oct	6,469	0.950	6,145	6,218	7,195	6,186	6,436
	47	Nov	6,499	1.034	6,722	6,249	7,261	6,781	6,753
	48	Dec	6,530	1.193	7,787	6,279	7,327	7,878	7,318
	TOTAL		76,363		76,436	73,353	83,561	77,026	77,594
	Min Month								5,690
	Max Month								7,318
	Range								1,628
	HISTORICAL ACCURACY COMPARISON (In-Sample)
	Metric	SMA	WMA	Naive	SES	DES (Holt)	Holt-Winters	Best Method
	MAE	733	673	543	591	626	17	Holt-Winters
	MAPE	12.7%	11.7%	9.5%	10.3%	11.1%	0.3%	Holt-Winters
	RMSE	832	773	694	722	729	21	Holt-Winters
	Bias	135	117	91	215	63	2	Holt-Winters
	Wins (of 4)	0	0	0	0	0	4	Holt-Winters
	MODEL RECOMMENDATION
	Recommended Method			Holt-Winters
	Best MAPE			0.3%
	MAPE Threshold			15.0%
	Threshold Status			PASS ✓
	Best MAE			17
	Best RMSE			21
	TRACKING SIGNAL ANALYSIS (Holt-Winters)
	Period	Month	Actual	HW Fcst	Error	Cumul. Error	Running MAD	Track. Signal	Status
	13	Jan	4,500	4,487	13	13	13	1.00	In Control
	14	Feb	4,650	4,652	-2	11	8	1.51	In Control
	15	Mar	5,100	5,131	-31	-20	16	-1.29	In Control
	16	Apr	5,750	5,755	-5	-25	13	-1.94	In Control
	17	May	6,200	6,175	25	0	15	0.01	In Control
	18	Jun	6,500	6,502	-2	-2	13	-0.18	In Control
	19	Jul	5,850	5,860	-10	-12	13	-0.94	In Control
	20	Aug	5,550	5,533	17	5	13	0.35	In Control
	21	Sep	5,250	5,219	31	36	15	2.35	In Control
	22	Oct	5,450	5,446	4	40	14	2.83	In Control
	23	Nov	5,950	5,983	-33	7	16	0.43	In Control
	24	Dec	6,900	6,928	-28	-21	17	-1.26	In Control
	25	Jan	4,800	4,774	26	4	17	0.25	In Control
	26	Feb	4,950	4,950	0	4	16	0.25	In Control
	27	Mar	5,450	5,454	-4	0	15	-0.01	In Control
	28	Apr	6,100	6,137	-37	-37	17	-2.21	In Control
	29	May	6,600	6,575	25	-12	17	-0.70	In Control
	30	Jun	6,950	6,916	34	22	18	1.23	In Control
	31	Jul	6,250	6,245	5	28	18	1.58	In Control
	32	Aug	5,900	5,910	-10	18	17	1.04	In Control
	33	Sep	5,600	5,566	34	52	18	2.88	In Control
	34	Oct	5,800	5,803	-3	49	17	2.82	In Control
	35	Nov	6,350	6,364	-14	35	17	2.02	In Control
	36	Dec	7,400	7,382	18	53	17	3.08	In Control
	Tracking Signal Limit			4.0		% In Control		100.0%
	METHODOLOGY NOTES
	1. Trend+Seasonal = Linear trend × Seasonal index (multiplicative re-seasonalization).
	2. Consensus Average = simple mean of all 4 forecast methods. Reduces individual method bias.
	3. Tracking Signal: values beyond ± limit indicate systematic bias requiring model recalibration.
	4. MAPE threshold and tracking signal limit are configurable in the Assumptions sheet.