Your odds, title by title
| Title | Forecast release | Your age | P(you're around) |
|---|---|---|---|
| GTA VIout now | 19 Nov 2026 | 21 | played? |
| GTA VIIlikely | Jan 2045 | 39 | 99% |
| GTA VIIIlikely | Mar 2070 | 64 | 92% |
| GTA IXlong shot | Nov 2104 | 99 | 6% |
| GTA X | Sept 2152 | 147 | <0.1% |
How do we calculate this?The full working paper. Contains equations. And regret.
§0Abstract
We fit a log-linear regression to all seven observed intervals between mainline Grand Theft Auto releases (1997–2026) and find that each successive wait is 1.38× longer than the last. Combining this release-interval model with a Gompertz survival model of human mortality, we estimate for any reader the probability of being alive at each future launch. Results are robust to everything except reality.
§1Data
Our sample is the full mainline series, using first worldwide release dates. Spin-offs, expansions and remasters are excluded as non-canonical noise. Decimal years use 365.2425-day years.
| n | Interval | Released | Gap (yrs) |
|---|---|---|---|
| 1 | GTA → GTA 2 | 22 Oct 1999 | 1.90 |
| 2 | GTA 2 → GTA III | 22 Oct 2001 | 2.00 |
| 3 | GTA III → Vice City | 27 Oct 2002 | 1.01 |
| 4 | Vice City → San Andreas | 26 Oct 2004 | 2.00 |
| 5 | San Andreas → GTA IV | 29 Apr 2008 | 3.51 |
| 6 | GTA IV → GTA V | 17 Sept 2013 | 5.39 |
| 7 | GTA V → GTA VI | 19 Nov 2026 | 13.17 |
§2Release-interval model
Let gn be the gap preceding the n-th sequel. We estimate by ordinary least squares:
Naively extrapolating the fitted line puts GTA VIII in 2052, which is frankly optimistic. The live-service era (GTA Online, 2013–) introduced a structural break, so following standard practice in regime-shift forecasting we anchor projections to the most recent observed interval:
| Title | Gap (yrs) | Forecast release |
|---|---|---|
| GTA VII | 18.2 | Jan 2045 |
| GTA VIII | 25.1 | Mar 2070 |
| GTA IX | 34.7 | Nov 2104 |
| GTA X | 47.9 | Sept 2152 |
§3Survival model
Adult mortality is well described by the Gompertz law: the hazard of death rises exponentially with age. Parameterised by the modal age at death M, the probability that a person aged x survives another t years is:
These values reproduce a life expectancy at birth of roughly 83 years, in line with Western European life tables. We use the population average and do not ask for your sex, your diet, or how many hours you put into GTA Online.
Age is derived from birth year assuming a mid-year birthday: x = (forecast date) − (birth year + 0.5). If you were born in January, congratulations, you are slightly worse off than we said.
§4Expected titles
Your expected number of future GTA releases is the sum of your survival probabilities at each forecast launch:
Your "last GTA" is the latest title you have at least a 50% chance of being alive for. If that's not even GTA VII, it's GTA VI, which is out now.
§5Limitations
- The model assumes no breakthroughs in medicine, cryonics or mind uploading. It also assumes no breakthroughs at Rockstar.
- With seven data points we have exactly as much statistical power as a group chat argument.
- Forecasts carry an error of ±1 delay announcement, and possibly ±1 apology on X.
- Bully 2 was treated as an unobservable latent variable and left out of the model.
App. BRejected specifications
"Each gap equals the sum of all previous gaps." Puts GTA VII in 2055 and GTA VIII in 2113. Rejected by the ethics committee for causing distress in readers over 25.
"The gap multiplier itself grows." Puts GTA VII in 2078, i.e. Cyberpunk 2077 plus one delay. Rejected as too on the nose.
Try another year
Bonus forecast
Will you live to see Half-Life 3? Same institute, same rigour.