EV (Expected Value) for one spin is the average value you should expect per spin if you repeated the same spin many times under the same probabilities. Practically, it helps you decide whether a spin is "worth it" by converting each possible outcome into a single number: the probability-weighted average payout minus the spin cost.
Core concept: expected value for a single spin
- EV is a long-run average per spin, not a guarantee for the next spin.
- The core formula is: EV = Σ(probability × payout) − cost.
- Include all outcomes, especially "nothing" outcomes with payout = 0.
- Use consistent units (THB, gems, points) and convert rewards to the same unit before computing.
- Positive EV can still feel bad short-term because variance can be high.
What expected value (EV) means in one spin
For a single spin, expected value (EV) is the "fair average return" you get when you multiply each possible reward by its chance and add everything up, then subtract what you paid. If you want to คำนวณ EV quickly, you are essentially compressing a probability distribution into one decision-ready number.
EV answers: "If I could run this exact spin repeatedly, what would my average net gain/loss per spin be?" It does not answer: "Will my next spin pay out?" That's a probability question, not an expectation question.
Keep EV bounded to the same rule set: same banner, same drop table, same costs, same conversion rates. If any of those change, the EV changes too.
| Metric | What it tells you | What it can miss |
|---|---|---|
| EV (net per spin) | Average gain/loss per spin over many spins | Short-run swings (variance) and personal utility |
| Win rate | Chance to get something "good" | How big the wins are |
| ROI-style ratio | Return relative to cost (useful for comparing options) | Absolute loss size and risk profile |
Step-by-step EV calculation with a simple example
Use the สูตรคำนวณ Expected Value below and treat each prize tier as an outcome.
- List outcomes and assign a payout value to each (convert items to THB/gems if you can).
- Assign probabilities to each outcome (they should sum to 1.00).
- Compute expected payout: add up (payout × probability) across outcomes.
- Subtract the spin cost to get net EV per spin.
- Interpret: EV > 0 means favorable on average; EV < 0 means you pay more than you get back (on average).
Example (one spin): Cost = 10 units. Outcomes: 1% chance to get a 500-unit item, 9% chance to get a 50-unit item, 90% chance to get 0. Expected payout = 0.01×500 + 0.09×50 + 0.90×0 = 5 + 4.5 + 0 = 9.5. Net EV = 9.5 − 10 = −0.5 units per spin.
Common payout structures and how they affect EV
EV is the same math, but payout structures change what you must model to get a correct "value per spin." This is especially important when you คำนวณความคุ้มค่า กาชา or any random draw system with tiers and guarantees.
- Tiered rarity drops: Common/rare/legendary tiers. Model each tier as outcomes with their probabilities and converted values.
- Duplicate conversion: Duplicates become shards/coins. Use the duplicate value, not the "new item" value, if you already own it.
- Guaranteed pity after N spins: The EV per spin depends on your current pity counter; treat "state" as part of the model.
- Bundles and discounts: A discounted 10-pack changes the per-spin cost; EV must use the discounted cost, not the headline price.
- Mixed rewards (items + currency): Convert all rewards to a single unit (e.g., THB-equivalent or gem-equivalent) before summing.
Example (duplicates): If a "rare" is worth 100 units when new but only 20 units as a duplicate, and you already own it, your EV should use 20 for that outcome-otherwise you overestimate.
Handling probabilities: outcomes, zero-sum, and house edge
Correct EV depends more on correct probabilities than on fancy math. When people คำนวณความคุ้มค่า การสุ่มไอเทม, the biggest errors come from missing outcomes and misreading rates.
- Outcomes must be exhaustive: Include "nothing" (payout 0) and partial rewards (e.g., consolation currency).
- Probabilities must sum to 1: If the game lists tier rates plus sub-rates, multiply correctly (tier rate × within-tier rate).
- Conditional states matter: If odds change after losses (pity, step-up), EV should be conditional on your current state.
- Zero-sum vs non-zero-sum: In PvP trading markets, "value" may transfer between players; in gacha shops, "value" is defined by the game economy and your personal utility.
- House edge concept: If rewards are priced so that average return is below cost, EV will be negative by design. The "edge" is the gap between expected payout and cost.
- EV is not a fairness certificate: Transparent rates help, but EV still depends on your conversion rate for items (how much they're worth to you).
Example (rate multiplication): If "SSR" is 2% and a specific SSR is 25% within SSRs, then that item's probability is 0.02×0.25 = 0.005 (0.5%).
Practical adjustments: fees, multiple bets, and variance
- Include fees and friction: Currency exchange loss, marketplace tax, crafting fees, or time cost should be subtracted as part of "cost" or reduced from "payout."
- Multiple bets per spin: If one spin yields multiple independent rolls, compute EV for each roll and add them (linearity of expectation).
- Variance can dominate feelings: Two spins can have the same EV but wildly different risk; don't confuse "high chance of small rewards" with "better EV."
- Don't double-count pity: If a pity guarantees a drop, don't also count the same guaranteed value elsewhere; model it once in the correct state.
- Be honest about valuation: If an item is not tradable and you wouldn't pay for it, its value to you may be near zero even if it is "rare."
Example (fees): If an item sells for 100 units but you lose 10% to tax, the realizable payout is 90 units for EV purposes.
Interpreting EV: when a positive EV matters in practice
Positive EV is actionable only when you can repeat the spin enough times (or the value is realized reliably) and your valuation is stable. If you rely on a one-time lucky hit, EV won't protect you from downside.
Mini-case: Suppose you found a limited-time spin where your net EV is +1 unit/spin after converting rewards to your personal value scale. If you can do 100 spins under the same rules, the average net gain trends toward +100 units, but actual outcomes may still vary widely.
expected_payout = sum(p[i] * value[i] for i in outcomes)
net_EV = expected_payout - spin_cost
if net_EV > 0 and you can repeat under same rules:
consider it (then check variance and budget)
else:
treat it as entertainment spendWhen comparing options, an "EV calculator" approach (what many people call a เครื่องคิดเลข EV กาชา) is simply applying the same net_EV computation to each banner/package with consistent valuations and costs.
Self-check before you commit (practical EV checklist)
- Did I include a 0-value outcome for "no meaningful reward" and ensure probabilities sum to 1?
- Did I convert every reward into the same unit (THB/gems/points) using my own realistic valuations?
- Did I subtract the full cost (price, fees, taxes, exchange loss) to get net EV?
- Did I account for my current state (pity counter, step-up stage, owned duplicates)?
- Did I consider variance and whether I can repeat enough spins for EV to matter?
Quick clarifications and edge cases
Is EV the same as my chance to win?
No. EV is an average value; win chance is a probability of a specific event. You can have a high win chance with low EV if the wins are tiny.
Can EV be positive while I still lose money short-term?
Yes. EV is a long-run expectation; short-run results can be negative due to variance, especially with rare high payouts.
Do I need exact probabilities to compute EV?
You need accurate enough probabilities for the decision. If rates are missing or ambiguous, your EV becomes an estimate and should be treated cautiously.
How do I value non-tradable items in EV?
Use your personal utility value (what you would realistically pay or what it saves you). If you wouldn't spend for it, assigning a low value is more honest.
How does pity/guarantee change EV per spin?
EV becomes state-dependent: it changes based on how close you are to the guarantee. Compute EV conditional on your current pity counter rather than using a single static number.
Is EV useful if I only plan to spin once?
It's still informative as a "price vs average value" check, but it won't predict the outcome of that one spin. For one-off decisions, also consider worst-case acceptable loss.
What's the most common EV mistake in gacha-style systems?
Overvaluing top-tier drops while ignoring duplicates and the high probability of low/zero payouts. Missing the "nothing" outcome is the fastest way to inflate EV.
