Can you predict a hangover? How the Nomicho hangover-forecast app works
What if, while you were still drinking, you had a rough sense of how rough the morning after would be? Honestly, first: why a hangover happens is still not fully understood. There's no magic that nails it. But the factors that drive hangover severity are reasonably well known, and from those you can estimate "about how bad this is likely to be." Here's how that works — and where it stops.
Why a hangover happens at all
The alcohol (ethanol) you drink is broken down mostly in the liver, in two steps: first an enzyme turns it into acetaldehyde, then acetaldehyde is broken down to acetate, and finally to water and carbon dioxide. The acetaldehyde in the middle is toxic — it's behind the flushing, racing heart, and nausea some people feel.
But "leftover acetaldehyde" doesn't fully explain a hangover. Acetaldehyde is rarely even detected in the blood during a hangover; the leading view is that several factors combine — mild withdrawal, dehydration, poorer sleep, and congeners (trace compounds in drinks).[1][4] There's no single cause — which is exactly why no single number can name it.
What makes a hangover worse
Even without a complete explanation, research repeatedly points to the same "conditions that make it worse."[1]
| Factor | Effect on severity |
|---|---|
| Pure alcohol drunk (per kg) | More is worse; a smaller body feels the same dose more |
| Peak blood alcohol concentration | The higher it climbs, the more likely it bites |
| Sleep | Short or poor sleep tends to make it worse |
| Constitution (flushing) | Weaker acetaldehyde breakdown tends to mean a rougher time |
| Type of drink | Congeners — higher in darker drinks — are reported to play a part |
| How drunk you felt | The more intoxicated it felt, the heavier the hangover tends to be |
The amount drunk (pure alcohol) and how intoxicated it felt both correlate with severity across several studies.[1][2] Short or poor sleep adds to severity independently of the dose.[3] People who flush easily (weaker acetaldehyde breakdown) also tend to have it worse.[5] But these are all "on average" tendencies — at the same conditions, how it lands varies a lot from person to person.
So it's a 1–5, not a single number
With many factors and large individual variation, the honest way to estimate a hangover is a rough tier, not a precise figure. Nomicho estimates severity on a 1–5 scale.
Nomicho's hangover forecast
Nomicho is an iPhone app for logging what you drink, and its hangover forecast predicts the morning after before it arrives. It runs when a drinking session ends, pulling together the factors above — pure alcohol drunk, peak blood alcohol concentration, sleep, flush response, type of drink, and how drunk it felt — and estimates severity from 1 to 5. When there's no alcohol (zero pure-alcohol grams), it shows no forecast at all: it won't manufacture a number it has no basis for.
It gets closer to you the more you use it
This is the big difference from a lookup table. After it forecasts, once the alcohol has cleared, Nomicho gently asks: "How was the hangover?" When you answer with the actual 1–5, it learns from the gap between forecast and actual and nudges the next forecast toward you.
It starts from a population average. Once a handful of your real answers accumulate (roughly five), your own pattern starts to show and the forecast drifts toward you. The figure below sketches that "drawing closer," conceptually.
Forecast → ask the actual → record it → use it next time. That small loop is what turns the hangover forecast into an estimate that's actually yours.
"Just drink water" — does that work?
Drink extra water, take a turmeric supplement, have a coffee — the usual moves. But there's no known way to reliably prevent or cure a hangover. Dehydration is one of the factors, so hydrating isn't pointless — but it won't make a hangover disappear on its own. Congeners (higher in darker drinks) do show up as a difference by type, yet what matters most is still the total amount drunk and your constitution.[1]
What the forecast can and can't tell you
The forecast is an estimate, not a measurement. Hydration, food, and the day's condition — factors it can't fully capture — also move a hangover. It isn't a medical diagnosis. What it can do is hand you a clue to your own pattern: "when I drink about this much, the morning after tends to land about here." Not a score of good or bad — just a way to watch, out of curiosity, how alcohol lands on your own body.
Note too that "the alcohol clearing" and "no hangover" are separate things: your level can fall to near zero and a rough morning can still remain. We get into why in the article on how long alcohol takes to clear. For how the in-the-moment "how drunk am I right now" gets estimated, see the real-time prediction article. How flush response relates to intoxication and hangovers is covered in the flush-response article.
Sources
- [1] Mackus et al. Proposal of a comprehensive model of alcohol hangover. J Clin Med. 2020;9(2):539. PMC7019760 (the factors that drive hangover severity).
- [2] Penning et al. The pathology of alcohol hangover. Psychopharmacology. 2012. doi:10.1007/s00213-012-2866-y (amount drunk vs. severity).
- [3] Devenney et al. Sleep after heavy alcohol consumption and physical activity levels during alcohol hangover. J Clin Med. 2021. PMC8658514 (sleep and hangover).
- [4] MHLW e-healthnet, "Mechanism of hangover" (Japanese)
- [5] MHLW e-healthnet, "Flushing reaction" (Japanese; on the flush-prone constitution)
For exactly which equations and primary sources Nomicho uses, see the methodology.
This article is general information, not medical advice. The forecast is an estimate and does not guarantee any specific person's values. The figures are conceptual illustrations.