The Math Behind Critical Hits, Drop Rates, and Betting Odds

You press the button. No crit. Your friend lands one and cheers. In another tab, you see odds for a match. These feel like three worlds. They are not. One set of rules runs all three.

This guide shows you those rules in plain words. You will learn what your chance means in real play. You will learn how long it can take to see a rare drop. You will learn to read betting odds, spot the hidden margin, and check the value of a bet. There is a clear table you can use again and again.

Two tracks, one idea

Think of every try as a small test. It can pass or fail. In games, a crit is a pass. A rare drop is a pass. In betting, a win is a pass. Same idea, new skin. That idea is a Bernoulli trial. If the chance of a pass is p, then the chance of a fail is 1 − p. In long play, the share of passes gets close to p. See a simple walk‑through here: Bernoulli trials explained.

Why crits feel streaky even when they are fair

A crit chance of 20% means one in five on average. It does not mean “one crit every five hits.” You can get three crits in a row. You can also miss ten in a row. Both are rare, not impossible. Your brain looks for patterns. It hates gaps. That is why “fair” can feel “rigged.”

If you land no crits for a while, you may think, “I am due.” That is the gambler’s fallacy. Each hit still has the same p. The past does not push the next hit to “make up” for it. Unless the game uses a pity system (more on that soon), chance does not have memory.

Drop rates and the wait for the first rare

Now think about a rare drop with p = 1%. How many tries until the first one? This is a geometric story: each try is a fail until the first pass. The expected number of tries is 1/p, so 100. But “expected” is not a promise. To have a 95% chance to see at least one drop, you need close to 300 tries. That long tail is normal. A short intro is here: geometric distribution intuition.

Pity timers: cutting the tail

Many games add a pity timer. It is a cap or a ramp. A hard cap might say: “You get the item by the 100th try at worst.” This change trims the long tail. For p = 1% and a hard cap at 100, the expected tries drop to around 63. The 95th percentile also drops from about 300 to about 100. Your wait becomes less wild, and the end feels fairer. Some stores even ask devs to tell players the odds of drops or loot boxes. See policy context here: loot box probabilities and disclosure policies.

“Random” in games is not true random

Games use a pseudo random number generator (PRNG). It is a math box. It makes numbers that look random, but they come from a seed and a rule. Good PRNGs pass tests and feel fair. Some games also use a “pseudo shuffle” that smooths streaks. That can be good for user feel. Learn why some PRNGs are better: PCG family of pseudorandom generators.

Myth break: three quick fixes for bad hunches

• “After ten fails, my chance is higher.” False unless pity raises p. If p is fixed, each try is the same p.
• “Uniform random is always best.” Not for user feel. Some games damp streaks on purpose to feel fair.
• “RTP means I will get X back soon.” No. Return to Player is long run math, not a short run promise. Read a plain guide: Return to Player (RTP) explained.

Odds look many ways. Convert them to one

Sportsbooks show decimal (1.91), fractional (10/11), or moneyline (−110). Convert all to an “implied probability.” For decimal odds d, implied p = 1/d. For 1.91, p ≈ 52.36%. For moneyline, there is a small rule set you can learn. A step by step guide is here: converting odds to implied probability.

Expected value: the one number you can trust

Expected Value (EV) is the average gain per bet if you could repeat it many times. In words: EV = (chance you win × net win) − (chance you lose × stake). If you bet $100 at decimal 1.91 on a true 50/50 event, your net win is $91 on a win. EV = 0.5 × 91 − 0.5 × 100 = −$4.50. That is −4.5%. Over time, this loss is the price you pay to play.

An intro to the math is here: expected value basics.

House edge, RTP, and the small “tax” on each price

Bookmakers add a margin. In a fair coin toss, each side should be even money (2.00). You often see 1.91 vs 1.91. The implied p adds to 104.7% (1/1.91 + 1/1.91). That extra 4.7% is the overround. It bakes in a negative EV for you. In slots and table games, this is the house edge (100% − RTP). Rules and tech tests try to keep games safe and fair. See a standards hub: Nevada Gaming Control Board – Technical Standards.

Combos, parlays, and hidden links

Parlays chain legs. The chance to win all legs falls fast. If each leg has a 55% true chance, three legs together have 0.55³ ≈ 16.6%. Payouts rise, but swings rise more. Also, some legs are linked in real life (correlated). Linked legs make risk larger than the odds suggest. A plain walk‑through of terms is here: American Gaming Association resources.

One frame to rule them all

Here is the link between worlds:

• Crits: Bernoulli per hit. If p is fixed, the long run rate is p. Streaks are real but rare.
• Rare drops: geometric wait time. E[tries] = 1/p. The 90–95% line can be high when p is small.
• Bets: the same pass/fail frame, but value matters. EV tells you the true price. The margin makes the sum of implied p over 100%.

Need a one page view of common distributions? Try this: Probability cheat sheet.

Small lab: do the math on a napkin

Task 1: 20% crit. How long until you likely see one? A 95% target means 1 − (1 − p)^n ≥ 0.95. Solve for n. You get n ≈ log(0.05)/log(0.8) ≈ 14. So do not tilt if you miss ten in a row. It can happen.

Task 2: 1% rare drop. E[tries] = 100. 95% line is about 300. Pity at 100 cuts that hard tail. If you hate long droughts, pick games with a clear pity rule.

Task 3: Even match at 1.91 odds. EV on $100 = 0.5 × 91 − 0.5 × 100 = −$4.50. If you do not have an edge, your bankroll trends down with time. That is math, not mood.

Want to play with randomness? Try this simple page that shows why true random is “lumpier” than you think: true randomness vs PRNG.

One Math, Three Worlds: a quick table

Use this table as your pocket guide. Read the “Takeaway” line when you need fast advice.

Plain talk takeaways

• If the chance is low, waits are long. Plan for the 90–95% line, not just the mean.
• If the price is worse than fair, EV is negative. Over time, that adds up.
• Pity timers are your friend if you hate long tails.
• Parlays boost swings. Keep stakes tiny if you choose them.

Show your work (light math)

We want 1 − (1 − p)^n ≥ 0.95. Solve: (1 − p)^n ≤ 0.05 → n ≥ log(0.05) / log(1 − p). For p = 0.20, n ≈ 14. For p = 0.01, n ≈ 299.

Implied p = 1/d. Sum for both sides. At 1.91 and 1.91, sum ≈ 1/1.91 + 1/1.91 ≈ 1.047. The 4.7% is the margin spread across both prices.

Net win on a hit = (d − 1) × S. EV = q × (d − 1) × S − (1 − q) × S. A fair price has d = 1/q. Any d < 1/q makes EV negative.

Tools, choices, and playing with care

Check the rules, the margin, and the payout paths before you risk money. Live dealer games have their own pace, edge, and table limits. If you want a clean place to compare live dealer rules, margins, and speed in Norway, see our notes here: live casino Norge. This link may be affiliate. We may earn a fee if you sign up. That never changes our view.

If gambling no longer feels fun, stop. Set limits. Do not chase losses. If you need help, talk to a pro. You can find free help and a self‑check here: National Council on Problem Gambling.

Quick FAQ

Q: My crit chance is 25%. Why did I miss 12 in a row?
A: It is rare, but it can happen. Each hit has the same 25% chance if there is no pity. Long gaps do not mean the system is broken.

Q: Is a higher RTP a promise I will win more tonight?
A: No. RTP is a long run rate. Short runs can swing a lot. Think months, not minutes.

Q: How do I spot a bad price fast?
A: Convert to implied p and add both sides. If the sum is far over 100%, the margin is big. Also, compare odds across books if you can.

Q: Are parlays bad?
A: They raise swings. They can be fine for fun with small stakes. They are not good for steady returns unless you have a real edge on each leg.

Q: Do pity timers make games “less random”?
A: They shape the wait time. Your single‑try chance may still be random, but the cap trims long droughts. That can feel more fair.

Q: Can I beat the margin?
A: It is hard. You need better info or a slow book. If you cannot find an edge, the safe move is to treat it as paid fun and keep stakes small.

Methods in short

• Crits and drops use Bernoulli tests. Wait times use geometric math.
• We used E[tries] = 1/p for no‑pity drops, and the 95% line from 1 − (1 − p)^n.
• For a hard pity at N, E[min(T, N)] = (1 − (1 − p)^N)/p.
• Odds to implied p: decimal p = 1/d. Overround = sum of implied p across all sides − 100%.
• EV = (win chance × net win) − (loss chance × stake). We rounded where it helps reading.

Where to learn more

• Bernoulli trials explained — friendly intro.
• Geometric distribution intuition — wait time logic.
• Loot box probability and disclosure — platform policy context.
• PRNG basics: PCG — why PRNG choice matters.
• RTP explained by UKGC — what RTP does and does not say.
• Implied probability from odds — fast conversions.
• MIT OCW on probability — deeper EV background.
• Nevada GCB standards — technical guardrails.
• AGA resources — glossary and responsible play.
• Probability cheat sheet — quick reference.
• True randomness vs PRNG — see “clumps” in action.

Author and update

Author: Alex R., game economy analyst and math nerd. Worked on drop tables and live ops for two RPGs. Built EV tools for sports data hobby work. Writes about risk with clear math and plain words.

Editorial note: No promises of profit. We explain risk so you can make informed choices. We list methods and sources. We disclose affiliate links. We do not take money for a positive view.

Last updated: June 27, 2026

Disclaimer: This article is for education. Gambling is for adults by local law. Play within your limits. If you have risk of harm, seek help.