Forecasters

The outfits that turn polls into win probabilities — FiveThirtyEight, Cook, the Economist, Sabato, Silver Bulletin and more — graded on the skill of their final pre-election call (2016–2024), against our own model. Most publish only a win probability — so the skill boards below are win-call skill. But the two that also publish a margin (538, the Economist) get a head-to-head on VotePredictor Elections' home turf.

Overall Pollsters Forecasters ML models

Difficulty-adjusted skill, with uncertainty

Because forecasters cover different races, raw accuracy isn't comparable — rating only safe states flatters you. A Bayesian two-way model separates race difficulty (from how everyone did on each race) from forecaster skill, so skill is independent of which races they chose. The bar is each forecaster's 90% credible interval — it's wide when they've rated few races — and the board is ranked by a conservative estimate, so a strong record on a handful of races can't top a proven one. Lower (left) is better.

The Governor tab is 538 vs our model only. 538's gubernatorial win probabilities (2018, 2022) were recoverable from its archived forecasts; the other outfits published governor ratings as maps, not numbers, so there's nothing to grade them on. For the full field on Governor — pollsters included — see the combined and pollster boards.

#	Forecaster	Races	Brier
1	FiveThirtyEightinactive	683	0.072
2	JHK Forecasts	391	0.078
3	Cook Political	615	0.080
4	Race to the WH	276	0.079
5	Fox News	174	0.079
6	Inside Elections	390	0.082
7	Sabato's Crystal Ball	615	0.083
8	The Economist	501	0.083
9	Split Ticket	196	0.083
10	Silver Bulletin	40	0.081
11	VotePredictor Electionsour model	683	0.086
12	Princeton Election Consortium	131	0.087
13	DDHQ/Decision Desk	531	0.089
14	CNalysis	276	0.090
15	Elections Daily	254	0.091
16	New York Times	51	0.092
17	RealClearPolitics	214	0.119

Margin accuracy vs the pros

The boards above grade win-calls, where the pros' proprietary modeling leads. But VotePredictor Elections' real edge is the margin — and on the same races, against the only two forecasters who publish one, it shows. We recovered 538's and the Economist's real final predicted margins (538 '18/'20/'22; Economist '20 president) and scored everyone by average margin error. VotePredictor Elections beats the Economist outright and edges 538 overall — winning on well-polled President and Senate races; 538 leads only on the thin-poll House/Governor seats where it leans on fundamentals. Lower is better.

VotePredictor Elections vs FiveThirtyEight

2018, 2020, 2022 · 504 races

Margin error (pts) — lower is better	VotePredictor Elections	FiveThirtyEight	polls
All races · 504	5.26	5.43	5.77
Competitive (<10 pts) · 240	4.82	4.12	4.60

Competitive races, by office	VotePredictor Elections	FiveThirtyEight	winner
President · 16	2.76	4.41	VotePredictor Elections
Senate · 29	3.82	5.91	VotePredictor Elections
Governor · 24	3.56	2.25	FiveThirtyEight
House · 171	5.37	4.06	FiveThirtyEight

VotePredictor Elections vs The Economist

2020 · 51 races

Margin error (pts) — lower is better	VotePredictor Elections	The Economist	polls
All races · 51	3.00	4.38	5.50
Competitive (<10 pts) · 14	2.42	4.45	4.07

House: a dedicated model closes the gap

2022 · vs 538

Margin error (pts) — lower is better	lean only	fundamentals	hybrid	538
All districts · 388	4.93	4.75	4.81	4.71
Competitive (<10 pts) · 72	5.36	4.87	4.51	4.22

538 normally beats us on the House because most districts are barely polled and 538 leans on district fundamentals. A dedicated district model (538-style partisan lean + incumbency) — and a hybrid that uses district polls where they exist and fundamentals elsewhere — flips that: it matches 538 overall (4.81 vs 4.71) and nearly matches on competitive seats (4.51 vs 4.22), versus the old poll-only model's 5.37. Boundary-aligned 2022 backtest (only 2022 & 2024 share district lines); stable on a leave-one-out 2024 check (4.33MAE). Not yet a live 2026 forecast — that requires this year's redistricted maps.

Win-call skill vs a prior-result baseline

Skill is how much each forecaster's win probability beats a naive prior-result baseline on the same races — so being right about obvious races earns nothing. The Competitive tab (races under 10 points) is the real test. Click a column to sort, a name for the by-office and by-cycle breakdown.

How good the win-probability call was (Brier skill vs a prior-result baseline). All races, 2016–2024.

#	Name	Races	Win skill	Brier
1	Silver Bulletin	40	64%	0.045
2	FiveThirtyEightinactive	683	57%	0.072
3	Cook Political	615	51%	0.081
4	VotePredictor Elections	804	50%	0.084
5	Sabato's Crystal Ball	615	49%	0.084
6	The Economist	501	49%	0.088
7	Race to the WH	276	49%	0.072
8	Inside Elections	390	48%	0.073
9	Fox News	174	45%	0.088
10	JHK Forecasts	391	45%	0.083
11	DDHQ/Decision Desk	531	44%	0.095
12	Poll average (naive)	804	43%	0.096
13	Princeton Election Consortium	131	43%	0.060
14	Split Ticket	196	43%	0.089
15	VotePredictor Elections House segmented	332	40%	0.120
16	CNalysis	276	39%	0.086
17	Elections Daily	254	38%	0.088
18	VotePredictor Elections House fundamentals	321	34%	0.130
19	New York Times	51	27%	0.081
20	RealClearPolitics	214	13%	0.127

Compare on common races

The fairest test when coverage differs: score everyone on the exact same races — the ones they all forecast. Drag the threshold to drop low-coverage forecasters; the fewer you require, the more races the survivors share. It defaults to the fair board — the six forecasters who covered ≥500 races, on the 510 they all called. Lower the bar and watch the paradox: admit one that rated only a handful and the common set collapses toward zero. Click any name to fine-tune the set.

Two columns, two questions — click either header to sort. Brier rewards confident, well-calibrated probabilities; Called right is the blunt “did they name the winner” rate. They can disagree: a forecaster can have a better Brier yet name fewer winners (its misses are mostly near-tossups it called at ~49%), so it's worth checking both.

Minimum coverage≥ 500 races · 6 forecasters · 501 common

Competitive only(final margin < 10 pts — where the hard calls are)

Everyone scored on the 501 races all 6 selected cover — identical races, exactly apples-to-apples.

#	Name	Brier ▾	Called right
1	FiveThirtyEight	0.075	91%
2	Cook Political	0.084	98%
3	The Economist	0.088	87%
4	Sabato's Crystal Ball	0.088	88%
5	DDHQ/Decision Desk	0.095	88%
6	VotePredictor Elections	0.096	87%

Raw Brier (for reference)

The same forecasters on the raw Brier score of their final win probability (lower is better) — not difficulty-adjusted, so safe-race coverage flatters it. The skill board above is the fairer ranking.

392 races where the final margin was within 15 points. Lower Brier is better. Only forecasters that rated every race in the set are ranked; the rest are listed unranked, with VotePredictor's Brier on exactly the races they rated.

#	Forecaster	Races	Brier	Called right	VP, same races
1	FiveThirtyEight	392	0.113	86%	0.137
2	Cook Political	392	0.123	95%	0.137
3	Sabato's Crystal Ball	392	0.129	83%	0.137
4	VotePredictor Elections	392	0.137	82%
—	Race to the WH	172	0.116	81%	0.127
—	Inside Elections	222	0.122	89%	0.124
—	JHK Forecasts	255	0.123	81%	0.134
—	Fox News	123	0.123	93%	0.146
—	Split Ticket	137	0.127	82%	0.139
—	The Economist	342	0.128	80%	0.140
—	CNalysis	172	0.136	79%	0.127
—	DDHQ/Decision Desk	357	0.137	83%	0.137
—	Elections Daily	158	0.141	80%	0.131
—	RealClearPolitics	133	0.199	79%	0.151