rafhq · omen · predictions
The record.
Every OMEN forecast is committed here before election day. Pre-registration is timestamped (minimum 72 hours of lead), the commit hash is logged, and the file doesn't change after write. Losses count. Demo backtests are shown separately below the live record.
Live pre-registrations
No live predictions yet. First target coming with the 2026 primary cycle.
Calibration
Reliability: if OMEN says "80% chance X wins" over many predictions, X should actually win about 80% of the time. Below, resolved predictions are binned by stated confidence; the bar shows observed hit rate. A calibrated forecaster sits on the diagonal. Not enough live predictions yet; showing demo backtests. Once we accumulate 20+ live resolutions this will switch to the live record.
9
Resolved (n)
78%
Winners correct
0.234
Mean Brier (winner)
StatedObservednhit rate
50–60%978%
60–70%0—
70–80%0—
80–90%0—
90–100%0—
Retroactive backtests (demo)
9 historical races replayed through the full hybrid forecaster. Winners: 7/9. Mean share err: 1.79pp.
2024 US Presidential (Popular Vote)2024-11-05demo
US-national · general · pre-reg 2026-04-21
Lean R · 63%trumpmargin 0.4pp
Calibrated share (p50 · 80% interval · p(win))
harris
49.8%
[49.7–50.2%] · p(win) 37%
trump
50.2%
[49.8–50.3%] · p(win) 63%
margin p50 0.2pp · 80% interval [0.0, 0.6]pp · 100 trials
Engine contribution
polling
52%
fundamentals
19%
synthetic
18%
event
10%
shrinkage 15% · pop: segment mix· 5,000 voters
Forecast trajectory
T−90
0p · 3e
T−60
2p · 4e
T−30
5p · 7e
T−14
5p · 8e
T−7
6p · 9e
T−0
9p · 11e
harristrumpactual
Actual outcome · resolved 2026-04-21
harris 48.3%trump 49.8%turnout 64%
winner correctshare err 0.95ppvs pregame ↓ 1.75pp
2024 NY-16 Democratic Primary2024-06-25demo
NY-16 · primary · pre-reg 2026-04-21
Solid D · 100%latimermargin 6.3pp
Calibrated share (p50 · 80% interval · p(win))
latimer
53.2%
[52.6–53.5%] · p(win) 100%
bowman
46.8%
[46.5–47.4%] · p(win) 0%
margin p50 6.1pp · 80% interval [5.2, 7.0]pp · 100 trials
Engine contribution
polling
28%
fundamentals
23%
synthetic
37%
event
13%
shrinkage 15% · pop: ACS PUMS (ny16)· 5,000 voters
Forecast trajectory
T−90
0p · 0e
T−60
0p · 1e
T−30
1p · 3e
T−14
2p · 4e
T−7
3p · 5e
T−0
4p · 7e
latimerbowmanactual
Actual outcome · resolved 2026-04-21
latimer 58.7%bowman 41.3%turnout 22%
winner correctshare err 5.53ppvs pregame ↓ 13.70pp
2022 Georgia Senate Runoff2022-12-06demo
GA-statewide · runoff · pre-reg 2026-04-21
Solid D · 100%warnockmargin 6.6pp
Calibrated share (p50 · 80% interval · p(win))
warnock
53.3%
[53.3–53.9%] · p(win) 100%
walker
46.7%
[46.1–46.7%] · p(win) 0%
margin p50 7.1pp · 80% interval [6.6, 7.7]pp · 100 trials
Engine contribution
polling
35%
fundamentals
26%
synthetic
25%
event
13%
shrinkage 15% · pop: ACS PUMS (ga-statewide)· 5,000 voters
Forecast trajectory
T−90
0p · 0e
T−60
0p · 0e
T−30
0p · 0e
T−14
1p · 3e
T−7
2p · 4e
T−0
5p · 7e
warnockwalkeractual
Actual outcome · resolved 2026-04-21
warnock 51.4%walker 48.6%turnout 56%
winner correctshare err 1.91ppvs pregame ↑ 0.40pp
2022 Arizona Senate2022-11-08demo
AZ-statewide · general · pre-reg 2026-04-21
Solid D · 100%kellymargin 6.9pp
Calibrated share (p50 · 80% interval · p(win))
kelly
53.4%
[53.3–53.7%] · p(win) 100%
masters
46.6%
[46.3–46.7%] · p(win) 0%
margin p50 7.0pp · 80% interval [6.5, 7.4]pp · 100 trials
Engine contribution
polling
42%
fundamentals
23%
synthetic
25%
event
10%
shrinkage 15% · pop: ACS PUMS (az-statewide)· 5,000 voters
Forecast trajectory
T−90
0p · 1e
T−60
0p · 1e
T−30
1p · 3e
T−14
3p · 5e
T−7
4p · 6e
T−0
6p · 7e
kellymastersactual
Actual outcome · resolved 2026-04-21
kelly 51.4%masters 46.5%turnout 64%
winner correctshare err 1.05ppvs pregame ↓ 1.05pp
2022 Pennsylvania Senate General2022-11-08demo
PA-statewide · general · pre-reg 2026-04-21
Solid D · 100%fettermanmargin 2.6pp
Calibrated share (p50 · 80% interval · p(win))
fetterman
51.3%
[51.2–51.6%] · p(win) 100%
oz
48.7%
[48.4–48.8%] · p(win) 0%
margin p50 2.7pp · 80% interval [2.4, 3.1]pp · 100 trials
Engine contribution
polling
46%
fundamentals
22%
synthetic
21%
event
11%
shrinkage 15% · pop: ACS PUMS (pa-statewide)· 5,000 voters
Forecast trajectory
T−90
0p · 2e
T−60
0p · 4e
T−30
0p · 5e
T−14
2p · 8e
T−7
4p · 9e
T−0
6p · 10e
fettermanozactual
Actual outcome · resolved 2026-04-21
fetterman 51.2%oz 46.5%turnout 55%
winner correctshare err 1.15ppvs pregame ↓ 1.15pp
2021 Virginia Governor2021-11-02demo
VA-statewide · general · pre-reg 2026-04-21
Lean R · 55%mcauliffemargin 0.0pp
Calibrated share (p50 · 80% interval · p(win))
youngkin
50.0%
[49.4–50.6%] · p(win) 55%
mcauliffe
50.0%
[49.4–50.6%] · p(win) 45%
margin p50 0.6pp · 80% interval [0.2, 1.5]pp · 100 trials
Engine contribution
polling
47%
fundamentals
22%
synthetic
17%
event
14%
shrinkage 15% · pop: ACS PUMS (va-statewide)· 5,000 voters
Forecast trajectory
T−90
0p · 0e
T−60
0p · 1e
T−30
0p · 2e
T−14
0p · 4e
T−7
3p · 6e
T−0
6p · 8e
youngkinmcauliffeactual
Actual outcome · resolved 2026-04-21
youngkin 50.6%mcauliffe 48.6%turnout 55%
winner wrongshare err 1.02ppvs pregame ↓ 3.00pp
2020 US Presidential (Popular Vote)2020-11-03demo
US-national · general · pre-reg 2026-04-21
Solid D · 100%bidenmargin 4.3pp
Calibrated share (p50 · 80% interval · p(win))
biden
52.1%
[51.9–52.4%] · p(win) 100%
trump
47.9%
[47.6–48.1%] · p(win) 0%
margin p50 4.2pp · 80% interval [3.8, 4.8]pp · 100 trials
Engine contribution
polling
44%
fundamentals
19%
synthetic
22%
event
15%
shrinkage 15% · pop: segment mix· 5,000 voters
Forecast trajectory
T−90
0p · 0e
T−60
1p · 2e
T−30
2p · 4e
T−14
3p · 4e
T−7
4p · 6e
T−0
6p · 8e
bidentrumpactual
Actual outcome · resolved 2026-04-21
biden 51.3%trump 46.9%turnout 67%
winner correctshare err 0.90ppvs pregame ↓ 0.90pp
2018 Texas Senate2018-11-06demo
TX-statewide · general · pre-reg 2026-04-21
Solid R · 100%cruzmargin 4.5pp
Calibrated share (p50 · 80% interval · p(win))
cruz
52.2%
[52.1–53.0%] · p(win) 100%
orourke
47.8%
[47.0–47.9%] · p(win) 0%
margin p50 4.8pp · 80% interval [4.2, 5.9]pp · 100 trials
Engine contribution
polling
39%
fundamentals
31%
synthetic
23%
event
7%
shrinkage 15% · pop: ACS PUMS (tx-statewide)· 5,000 voters
Forecast trajectory
T−90
0p · 0e
T−60
1p · 1e
T−30
2p · 3e
T−14
3p · 5e
T−7
4p · 6e
T−0
5p · 7e
cruzorourkeactual
Actual outcome · resolved 2026-04-21
cruz 50.9%orourke 48.3%turnout 53%
winner correctshare err 0.95ppvs pregame ↓ 1.70pp
2016 MI Presidential2016-11-08demo
MI-statewide · general · pre-reg 2026-04-21
Solid D · 100%clintonmargin 1.4pp
Calibrated share (p50 · 80% interval · p(win))
trump
49.3%
[49.0–49.4%] · p(win) 0%
clinton
50.7%
[50.6–51.0%] · p(win) 100%
margin p50 1.7pp · 80% interval [1.2, 1.9]pp · 100 trials
Engine contribution
polling
39%
fundamentals
22%
synthetic
21%
event
18%
shrinkage 15% · pop: ACS PUMS (mi-statewide)· 5,000 voters
Forecast trajectory
T−90
0p · 0e
T−60
0p · 0e
T−30
1p · 2e
T−14
2p · 3e
T−7
3p · 6e
T−0
5p · 8e
trumpclintonactual
Actual outcome · resolved 2026-04-21
trump 47.5%clinton 47.3%turnout 64%
winner wrongshare err 2.60ppvs pregame ↓ 3.10pp