This summary assumes that any state with a current poll average margin of less than 5% could go either way, leading to the two best cases where one candidate wins all of these states. The "Expected" case is when each candidate wins every state they currently lead, regardless of how small that lead may be.
This model assumes poll errors in states are independent of each other. State win probabilities are derived from analyzing how Election Graphs poll averages at similar times before the election have historically deviated from actual election results from 2008 to 2020. This model accounts for how things can change in the time remaining.
2024 EC (Uniform Swing Probs)
Accounting For Time Left Before Election
Median:
Trump by 86
1σ (68.27%) range:
Harris by 66 ----- Trump by 86
2σ (95.45%) range:
Harris by 100 ---------- Trump by 142
3σ (99.73%) range:
Harris by 294 --------------- Trump by 156
Odds:
Harris: 36.4% — Tie: 0.0% — Trump: 63.6%
This model assumes poll errors in states will be identical nationwide. State win probabilities are derived from analyzing how Election Graphs poll averages at similar times before the election have historically deviated from actual election results from 2008 to 2020. This model accounts for how things can change in the time remaining.
Probabilistic View
If The Election Was Always Now
2024 EC (Indep States Probs)
If The Election Was Always Now
Median:
Trump by 44
1σ (68.27%) range:
Trump by 4 ----- Trump by 76
2σ (95.45%) range:
Harris by 34 ---------- Trump by 102
3σ (99.73%) range:
Harris by 78 --------------- Trump by 122
Odds:
Harris: 13.5% — Tie: 0.2% — Trump: 86.3%
This model assumes poll errors in states are independent of each other. State win probabilities are derived from analyzing how the final Election Graphs poll averages have historically deviated from actual election results from 2008 to 2020. This model shows how things would look each day if that day was election day.
2024 EC (Uniform Swing Probs)
If The Election Was Always Now
Median:
Trump by 86
1σ (68.27%) range:
Harris by 66 ----- Trump by 86
2σ (95.45%) range:
Harris by 100 ---------- Trump by 142
3σ (99.73%) range:
Harris by 294 --------------- Trump by 156
Odds:
Harris: 36.5% — Tie: 0.0% — Trump: 63.5%
This model assumes poll errors in states will be identical nationwide. State win probabilities are derived from analyzing how the final Election Graphs poll averages have historically deviated from actual election results from 2008 to 2020. This model shows how things would look each day if that day was election day.
Ten most needed polls: Mississippi, Louisiana, Oregon, Kansas, Nebraska (CD1), Connecticut, Kentucky, Alabama, Arkansas, Delaware
Election Graphs tracks state by state poll averages to estimate Electoral College results, and tracks estimates of the primary delegate races once they begin.
If you find this site interesting or useful, please consider visiting the Donation Page.
The poll average generally uses the last 5 polls (by middate).
If there is tie for the middate of the oldest poll to be included, all polls with that middate are included.
If the result is exactly on the border between categories (0%, 5% or 10% margin) older polls are pulled in one by one until the result is clearly within a category.
When there are not enough actual polls for the poll average, results from prior presidential elections are used to fill in the average.
If a pollster releases multiple results based on the same sample they are weighted so collectively they count as "1 poll".
On state detail pages this is noted by an [N], shading in the listing, and a different color data point in the graph.
State win probabilities are calculated based on 2008-2020 data using the methodology in the January 2023 blog post titled "Prepping the Math Stuff for 2024".