Illinois Pick 3 & Pick 4 Odds (2026): We Audited 54,029 Draws
Every day, thousands of Illinois players chase "hot," "cold," and "due" numbers in Pick 3 and Pick 4. So we did what nobody selling a system ever does: we pulled every draw since 1982 — 54,029 of them — and ran 962 statistical tests to see whether any pattern exists that a player could actually use. The answer is a clean, well-documented no. Here are the real odds, the real payouts, and the data that ends the "hot numbers" myth for good. More in our Personal Finance Statistics Hub →
The odds & payouts, straight from the rulebook
Pick 3 asks you to match a 3-digit number (000–999); Pick 4 a 4-digit number (0000–9999). A straight bet must match the exact order; a box bet wins if your digits arrive in any order — better odds, smaller prize. Here is what a $1 wager gets you:
| Bet type | Odds | Pays on $1 | Return per $1 |
|---|---|---|---|
| Pick 3 — Straight | 1 in 1,000 | $500 | ~$0.50 |
| Pick 3 — Box (6-way) | 1 in 167 | $80 | ~$0.48 |
| Pick 3 — Box (3-way) | 1 in 333 | $160 | ~$0.48 |
| Pick 4 — Straight | 1 in 10,000 | $5,000 | ~$0.50 |
| Pick 4 — Box (24-way) | 1 in 417 | $200 | ~$0.48 |
| Pick 4 — Box (12-way) | 1 in 833 | $400 | ~$0.48 |
Odds and prizes per the Illinois Lottery Draw Game Rules. Fireball is an optional add-on that creates extra ways to win by substituting a drawn "Fireball" digit, for a matching extra wager; it does not change the fundamental house edge.
Where your dollar really goes
Every one of those bets returns about 50 cents on the dollar over time. That is the single most important number in this entire article: whatever you bet, the state keeps roughly half. It is one of the worst house edges in legal gambling — for comparison, slot machines typically return 88–95 cents on the dollar.
The other ~50 cents is the house edge. No number, system, or app changes this.
Are the numbers actually random? Yes.
If any digit came up more often than others, you could tilt the odds. It doesn't. Across all 27,347 Pick 3 draws, every digit 0–9 lands almost exactly 10% of the time in every position — the tiny wiggles below are exactly the size you get from chance, and they sit comfortably inside the 95% confidence band.
The 1-in-1,000 club, exactly as random theory predicts
Here is the more powerful test. If the game is fair, each of the 1,000 Pick 3 combinations should be drawn about 27 times over 44 years, and the number of combos drawn 10, 20, 30, 40 times should follow a smooth bell-shaped Poisson curve. The green bars are what actually happened; the orange line is the mathematical prediction for a perfectly fair game. They match.
Every one of the 1,000 Pick 3 combinations has been drawn at least once. In Pick 4, 719 of the 10,000 combinations have never been drawn — but fair randomness predicts about 694 would still be missing after this many draws, so that too is normal.
"Hot" and "cold" numbers — the myth in one table
Yes, some numbers have come up more than others. That is exactly what randomness looks like — flip a coin 1,000 times and you won't get 500 exact. The most- and least-drawn Pick 3 numbers are below. The key question isn't "which is hottest," it's "is any of this outside chance?" It isn't.
| 🔥 Hottest | Times |
|---|---|
| 394 | 49 |
| 631 | 44 |
| 494 | 43 |
| 130 | 43 |
| 960 | 43 |
| ❄️ Coldest | Times |
|---|---|
| 685 | 8 |
| 392 | 11 |
| 199 | 13 |
| 648 | 14 |
| 484 | 14 |
Why this doesn't help you
In a fair 44-year history, the single most-drawn combo typically lands around 43–50 times and the least around 12 — so "394 at 49" and "685 at 8" are within, or a hair beyond, the normal range. More importantly, past draws have zero influence on the next one: 685 is exactly as likely as 394 tonight, both at 1 in 1,000. "Due" numbers are the gambler's fallacy in a lottery costume.
The three "patterns" people send us — and why none pay
When you analyze 54,029 draws, a few coincidences always surface. Here are the three most convincing-looking ones we found, and why each collapses on inspection.
1. "Pick 4 evening numbers ending in 9"
In the computer-RNG era, 11.9% of evening Pick 4 draws ended in 9, versus 10% expected (z = 4.0, one-in-ten-thousand on its own). Tempting. But it is the largest of 140 position-and-digit cells we checked, and the largest of 140 reaches that size about 1% of the time by chance. The tell: it vanishes everywhere else — midday draws ending in 9 run 9.5%, and ball-era evenings ran 9.5%. And even if it were real, a $1 straight bet on a combo ending in 9 would return 59 cents — still a 41% loss.
2. "Pick 3 evening numbers repeat"
Since the 2015 switch to a computer RNG, the exact evening Pick 3 number repeated the next evening 11 times versus about 4 expected (p = 0.003). Every one is official-verified — it really happened. But repeats at every other lag, in midday draws, and in Pick 4 are all at or below expectation, and over the full 44-year history the "always bet last night's number" strategy hits at exactly the chance rate. It is one flagged coincidence among hundreds of tests, not a signal.
3. "The number 685 is ice cold"
685 has been drawn just 8 times in 27,347 draws (expected: 27) and never in the RNG era. Spooky — until you remember we're looking at the minimum of 1,000 combinations. A minimum this low shows up in about 2% of fair lottery histories, and a "cold" number is not "due." Its odds tonight are 1 in 1,000, identical to every other combination.
We tried to beat it. Everything failed.
This is where most "analysis" stops. We kept going and actively tried to build a winning system:
- 962 statistical tests. 61 came back "significant" at p<0.05 — but about 48 are expected by chance alone, and zero survived a false-discovery-rate correction. The smallest p-value (0.0011) is right where the best of 962 coin-flips should land.
- 100 simulated fair lotteries. We generated 100 fake histories with random digits and ran the identical battery. Nine of them threw up more "significant" results than the real Illinois data, and four produced a result that survived correction — something the real data never did. The real lottery looks more random than many fakes.
- Every popular strategy, back-tested. Hot, cold, due, repeat-last, and Markov systems — 15 rules per game, walk-forward — all landed at chance.
- Machine learning. A LightGBM model and a neural network scored within a few thousandths of the no-skill baseline; logistic regression did worse than guessing (it overfit noise). No model predicted a digit better than 10% or a straight combo better than 1 in 1,000.
📊 The bottom line
Across 54,029 draws and 962 tests, the Illinois Pick 3 and Pick 4 games contain no detectable — let alone exploitable — pattern. The switch from numbered balls (through 2015) to a computer RNG (since October 2015) changed nothing (G-test p = 0.33). The draws are random and the payouts are fixed, which means there is exactly one honest strategy: treat it as entertainment you can afford to lose, never as an investment. If you want your money to actually grow, that is a different — and winnable — game.
So what is the "best" way to play?
Since no bet beats the ~50% house edge and no number is ever due, the math is simple:
- Box bets win more often; straight bets pay more. Neither has a better expected value — pick whichever you find more fun.
- Your own numbers are as good as any system. Birthdays, a quick pick, or a $49 "prediction app" all have identical odds. Save the money you'd spend on the app.
- Set a fixed entertainment budget and never chase losses. The house edge is relentless over time.
- If your goal is to build wealth, even $25/month invested does what no lottery ticket can — see the best way to invest $25 a month or run the numbers in The Lottery Lie.
See what those tickets could become instead
Our free tool shows what the money spent on lottery tickets would grow to if you invested it — the results are eye-opening.
Try The Lottery Lie → Best Investing AppsFrequently Asked Questions
A straight bet (exact order) is 1 in 1,000 and pays $500 on $1. A 6-way box is about 1 in 167 ($80); a 3-way box is about 1 in 333 ($160). Every option returns roughly 50 cents per $1 over time.
A straight bet is 1 in 10,000 and pays $5,000 on $1. A 24-way box is about 1 in 417 ($200). Like Pick 3, the expected return is about 50 cents per dollar wagered.
None you can use. Across 54,029 draws, the hottest Pick 3 combo (394, 49 times) and coldest (685, 8 times) both fall within chance. Every digit hits about 10% and no number is more likely to come up next.
No. We back-tested hot/cold/due/repeat/Markov systems and machine-learning models (LightGBM, a neural net, logistic regression) with walk-forward testing. None beat random guessing or predicted digits better than 10%.
The data is consistent with fair randomness. Numbered balls (through Sept 2015) and the computer RNG (since Oct 2015) show no difference (G-test p = 0.33). Nothing is tilted for or against players — it is random, with a fixed house edge in the payouts.
No bet has positive expected value; all return about 50 cents per dollar. Box wins more often, straight pays more. Pick whatever is fun, set a budget you can lose, and skip paid "prediction" systems — they can't beat 1-in-1,000 odds.
Sources & methodology
Official results: Illinois Lottery Pick 3 / Pick 4 (illinoislottery.com, Jan 2014 onward, verified by SHA-256). Odds & prizes: Illinois Lottery Draw Game Rules. Drawing method: Illinois Lottery daily drawings; NBC Chicago, "No More Numbered Balls as Illinois Lottery Drawings Go Digital" (Sept 29, 2015). Historical archives: lotterycorner.com and lottostrategies.com (1982–2013, reconciled draw-by-draw; 29 conflicting records excluded). Game history: Illinois Lottery / Wikipedia. Full test battery, back-tests, and reproducible code available on request. Figures current through September 14, 2026.