State	Rating	Flip Status	Median Margin	Dem % chance of winning	GOP % chance of winning
0	AL	Safe R	False	-31.367226	0.00000	100.00000
1	AR	Safe R	False	-28.504572	0.00000	100.00000
2	AZ	Lean R	False	-1.573731	42.13134	57.86866
3	CA	Safe D	False	25.783446	100.00000	0.00000
4	CO	Safe D	False	20.183131	100.00000	0.00000
5	CT	Safe D	False	18.993116	100.00000	0.00000
6	FL	Safe R	False	-14.502005	0.00000	100.00000
7	GA	Likely R	False	-6.610166	16.94917	83.05083
8	HI	Safe D	False	21.287061	100.00000	0.00000
9	IA	Safe R	False	-21.355824	0.00000	100.00000
10	ID	Safe R	False	-33.652297	0.00000	100.00000
11	IL	Safe D	False	19.611540	100.00000	0.00000
12	KS	Lean D	False	3.353948	66.76974	33.23026
13	MA	Safe D	True	26.443659	100.00000	0.00000
14	MD	Safe D	True	29.225509	100.00000	0.00000
15	ME	Likely D	False	5.809016	79.04508	20.95492
16	MI	Likely D	False	5.060947	75.30474	24.69526
17	MN	Safe D	False	9.430038	97.15019	2.84981
18	NE	Safe R	False	-19.891467	0.00000	100.00000
19	NH	Safe R	False	-24.093405	0.00000	100.00000
20	NM	Likely D	False	5.936734	79.68367	20.31633
21	NV	Tossup	True	-0.766418	46.16791	53.83209
22	NY	Safe D	False	11.237171	100.00000	0.00000
23	OH	Safe R	False	-24.701163	0.00000	100.00000
24	OK	Likely R	False	-5.570485	22.14758	77.85242
25	OR	Lean D	False	1.349297	56.74648	43.25352
26	PA	Safe D	False	10.594860	100.00000	0.00000
27	RI	Safe D	False	18.208487	100.00000	0.00000
28	SC	Safe R	False	-17.792756	0.00000	100.00000
29	SD	Safe R	False	-18.491980	0.00000	100.00000
30	TN	Safe R	False	-31.164392	0.00000	100.00000
31	TX	Safe R	False	-12.380219	0.00000	100.00000
32	VT	Safe R	False	-39.905781	0.00000	100.00000
33	WI	Tossup	True	-0.852010	45.73995	54.26005
34	WY	Safe R	False	-52.213766	0.00000	100.00000

Safe Republican - A <=95% chance of voting Republican

Likely Republican - A <=75% chance of voting Republican

Leans Republican - A <=55% chance of voting Republican

Tossup - A >=55% chance of voting for either party

Leans Democratic - A <=55% chance of voting Democratic

Likely Democratic - A <=75% chance of voting Democratic

Safe Democratic - A <=95% chance of voting Democratic

This model is made up of two components, two which everyone that's a politics junkie knows well - the fundamentals and the polls.

The model fundamentals are made up of three components in of themselves - a presidential composite using

• 5% 2012 presidential election results
• 15% 2016 presidential election results
• and 80% 2020 presidential election results.

On top of that, the presidential composite is adjusted for district trends over the last 2 years, which altogether makes a partisan lean index. But there are 3 states where there's been top-two primary results from 2022 that can suitably replace a partisan lean index - Alaska, California, and Washington, so AK/CA/WA adjusted (for primary-general swing) primary results MOSTLY replace the partisan lean index in those states. However, a partisan lean index is on its own not a good way to forecast elections, so that's why I've also used;

• gubernatorial approval ratings from Morning Consult (MA and MD are excluded)
• a candidate index (using 2018/2020 results for current incumbents and candidate experience (i.e. whether a candidate is a mayor, state representative, etc) for non-incumbents)
• state-by-state ratings from the Cook Political Report with Amy Walter , Inside Elections from Nathan Gonzales , Sabato's Crystal Ball from the UVA CFP , and Split Ticket .
• and a demographic index based off AP votecast (although demographics has very little weight.)

As for the polls, it's a little more simple; polling data simply uses a weighted average of every poll based off of recency. Recency weight is also adjusted for situations where there's less than 10 polls. Poll margins are also edited based off if they're a Democratic or Republican internal polls, since internal polls are usually 5 points more biased toward the surveying party.
After the fundamentals and poll margins are calculated, they're weighted together using the amount of polls that have been produced, which produces an exact margin. Once an exact margin is created, a randomizer of +2 to -2 points is ran 250,000 times, and the median margin is used as the neutral year forecast margin. The neutral year forecast margin is then added to the state level environment (which is calculated from the national environment), which produces the official acctuallydavid.com forecast margin for the state.