Florida Fantasy 5, measured
Fantasy 5 has fixed odds and a pool that changes every draw, which makes it harder to judge than a scratch-off. Two things are worth knowing before buying one, and only one of them is actionable.
Returned per dollar spent, over the game's lifetime
0.503
¤5,015,470,896 in prizes against ¤9,971,366,363 in sales, as published by Florida to December 2022.
That is worse than every scratch-off in the state. Instant tickets in this dataset return 60 to 80 cents on the dollar; Fantasy 5 returns about 50 cents. If the question is "which lottery product gives the most back", the draw game is not the answer — and a $1 ticket costing less than a $10 scratch-off does not change the rate.
Two draws a day, identical odds, very different prizes
Both draws pick 5 numbers from 36, which is 376,992 combinations, and both cost a dollar. The chance of winning is exactly the same. What differs is what happens to the money.
| Draw | Draws | Rolled down | Median top prize |
|---|---|---|---|
| MIDDAY | 1,273 | 60.7 % | ¤54,512 |
| EVENING | 1,273 | 34.8 % | ¤99,710 |
Every draw time is compared from 20 Mar 2023, the first day all of them ran. Comparing each over its own whole history would confound the time of day with the era: the pools have roughly halved since 2012. Compared over 20 Mar 2023 to 12 Sep 2026.
So one draw's typical top prize is 1.8× the other's, from the same odds and the same ticket price.
What a rolldown is
When no play matches all five numbers, the top prize money does not stay in the pot for next time. Florida rolls it down into the lower prize tiers of the same draw, so a draw with no jackpot winner pays more to everybody who matched four or three. That rule is published; it is simply not something most players have read.
It is also why the rolldown column above is the interesting one. A draw that rolls down six times in ten is a different product from one that rolls down three times in ten, even though the printed odds on the slip are identical.
What this page will not tell you
There are no hot numbers here and no overdue numbers. Each draw is independent of every draw before it, so any pattern in the history of drawn numbers is noise, and presenting it as a signal would be the gambler's fallacy with our name on it. We hold 12,402 draws going back to January 1995, and we use them to measure pools and payouts, never to predict a number.
The one thing the numbers you choose do change
Not whether you win — every combination is equally likely, and no arrangement of five numbers is closer to being drawn than another. What they change is how much a win is worth, because this prize is pari-mutuel: it is divided among everyone holding the winning combination. People play birthdays, so the numbers 1 to 31 are crowded, and a prize that lands there is split more ways.
Measured over the 4,503 draws since August 2001 that had a top-prize winner: a combination inside 1–31 averaged 2.36 winners and has once been split 98 ways, while a combination with two or more numbers above 31 averaged 1.92 and has never been split more than 8. The medians are two either way, so the difference is entirely in how badly it can go.
That is the only edge a lottery permits, and it is the exact opposite of chasing frequent numbers: the numbers most people play are the ones that divide a prize most. The app shows the whole table, with the number of draws each row rests on, and will pick five of the least-divided shape if you ask — saying plainly, next to them, that they are no likelier to come out.
Where to look next
Every figure here is read from our own database when the page is served rather than typed in, so it moves as draws are recorded. Latest draw held: 12 Sep 2026.