What does a 65% probability on a prediction market actually mean?

It is a price. A side quoted at 65% costs about 65 sats for every 100 sats it returns, so it pays roughly 1.54 times your stake if it wins. Read it as a break even hit rate: that side has to come in more than 65% of the time for the position to be profitable over many bets.

How do I convert a percentage into a payout multiple?

Divide 100 by the percentage. A side at 25% returns 4x your stake, a side at 50% returns 2x, and a side at 80% returns 1.25x. To go the other way, divide 100 by the multiple: a market offering 3x is quoting 33.3%.

Why does the probability change after I place a bet on some markets but not others?

Lightning Faucet runs two pricing engines. Pool based markets derive the percentage from where the sats sit, so it moves with every position and your projected payout floats until the market closes. Fixed odds sports markets post a price up front and lock it at the moment you bet, so the displayed line holds steady while the pools grow beneath it.

Why does a pool based market show YES as the other side's pool share?

Because that is what a YES position is actually worth. If YES wins, YES holders split the NO pool, so the value of backing YES depends on how much money is sitting on NO. Heavy money on NO makes YES cheap and valuable, and heavy money on YES makes it crowded and thin.

Should I trust the percentage on a brand new market?

No. A pool based market with no money in it displays 50/50 as a placeholder, not a forecast, and a market carrying only a few thousand sats can have its price set by one participant. Check the pool total and unique bettor count before you give the number any weight.

Are prediction markets provably fair like the casino games?

They work differently. Casino games such as dice use a hashed server seed, your client seed, and an incrementing nonce, so you can recompute any result yourself once the seed is revealed. Prediction markets resolve against real world data instead: a price feed, a difficulty epoch, a weather station, or a final score. The thing to verify is the resolution criterion on the market card, which should be precise enough to leave no argument.

What happens to my position if the event is cancelled?

Cancelled markets and pushes refund every position in full, with no settlement fee taken. That is why a market that looks like a certainty sometimes just returns your sats instead of paying out.

What is the smallest prediction I can make?

Most markets take positions from 10 sats, and each market publishes its own minimum and maximum on the card. The small minimum is the point: taking many tiny positions and then checking how often the ones you accepted at 60 to 70 percent actually landed is the fastest way to learn whether you are reading prices well.

How to read prediction market probabilities

Read prediction market probabilities like a price: convert percent to payout, spot pool based vs fixed odds pricing, and know when the number is noise.

By Lightning Faucet team August 6, 2026 10 min read

A prediction market probability is a price wearing a percentage costume. When a market shows YES 65%, it is telling you that a winning YES position costs roughly 65 sats for every 100 sats it returns. That is the whole definition. Everything else is detail about where the number came from and how much you should trust it.

The detail matters more than most explainers admit, and here is the part that is specific to Lightning Faucet: the same percentage bar on our market cards is produced by two completely different engines, and the number means something different in each. On a pool based market, the YES percentage you see is literally the NO side's share of the pool. On a fixed odds sports market, it is the posted price for that side, normalized so the two sides sum to 100. Identical visual, different arithmetic, different implications for what your bet pays. If you read every card the same way you will misprice half of them.

The one line definition, then the important asterisk

Implied probability is the percentage a market assigns to an outcome. Convert it to a price and it becomes actionable: a side quoted at 65% costs 65 to win 100, so it returns about 1.54 times your stake if it lands. A side quoted at 20% costs 20 to win 100, so it returns about 5 times your stake.

The asterisk: implied probability is not the market's honest belief about the world. It is the number that clears at the current price, after fees, after whatever structure the venue uses. Learning to read it means learning what the venue does to it before it reaches your screen.

On Lightning Faucet's prediction markets everything is denominated in sats, positions start as small as 10 sats on most markets, and settlement lands in your balance the moment the market resolves. That small minimum matters for learning: you can take twenty positions across twenty markets for the price of a coffee and actually watch how your read of a percentage performs.

The only math you need

Percentage to payout multiple

Divide 100 by the percentage.

80% implied, 100 / 80 = 1.25x your stake back if it wins
50% implied, 100 / 50 = 2.0x
25% implied, 100 / 25 = 4.0x
10% implied, 100 / 10 = 10.0x

A 1,000 sat position at 25% returns roughly 4,000 sats gross if it lands, of which 3,000 is profit.

Payout multiple back to percentage

Divide 100 by the multiple. A market offering 3x is quoting 33.3%. A market offering 1.1x is quoting about 91%.

Break even, which is the number that actually decides bets

The implied percentage is your break even hit rate. If you take a side quoted at 65%, you need that side to come in more than 65% of the time over many bets to come out ahead. This is the single most useful habit you can build: read the percentage as "how often does this have to be right", not "how likely is this".

A position quoted at 90% is not safe. It is a position that must land nine times out of ten just to break even, and one miss costs you nine wins.

Two engines, one percentage bar

This is the section a generic crypto blog cannot write, because it is about how our markets are actually built.

Most of what we run is parimutuel. Everyone backing YES pays into one pool, everyone backing NO pays into another, and at resolution the winning side splits the combined pool.

Here is the counterintuitive bit, and it is worth reading twice: the YES probability we display is the NO pool's share of the total. Not the YES pool's share. The reason is mechanical. If YES wins, YES holders split the NO pool, so what a YES position is worth depends entirely on how much money is sitting on the other side. Heavy money on NO makes YES cheap and valuable. Heavy money on YES makes YES crowded and thin.

Three consequences you should internalize:

The percentage moves as money arrives. It is not a forecast that gets refined, it is a live snapshot of where sats sit. A market can swing 20 points because one large position landed, with nothing whatsoever having changed in the real world.
Your payout estimate floats until the market closes. When you place a pool based position we show a projected return computed from the pools as they stand at that moment. Money that arrives after you does change your final payout. Late money on your own side dilutes you. Late money on the other side pays you.
A settlement fee comes out of the pool at resolution, not off your stake up front. So the projected return you see already accounts for it, and the raw pool ratio math will always be slightly more optimistic than what you actually receive. Trust the projection on the bet slip, not your own back of envelope pool arithmetic.

The strategic upside of pool based pricing is real: on a thinly traded market you are being paid by the crowd's imbalance, not by a professionally set price. If everyone piles onto the obvious side, the unpopular side becomes genuinely cheap.

Fixed odds sports markets: the percentage is the posted line, normalized

Some of our sports markets run as a fixed odds book instead. The mechanics are different in every way that matters to you.

The price is posted before the money arrives, derived from live sports odds rather than from whoever bet first.
Your price locks the instant you place the bet. Later money does not touch it. If you take a side at 40% you are paid at 40% at settlement, no matter what happens to the line afterwards.
The displayed percentages are normalized. A bookmaker's two quoted prices always sum to slightly more than 100, because that gap is how a book stays in business. We normalize the pair back to 100 for the display bar, so it reads as a probability. Your payout is calculated from the raw quoted price, so normalization changes what you see, never what you get.

We built the display that way after a specific failure. Early on we rendered fixed odds markets using the pool ratio, exactly like pool based markets. On a thin, lopsided book the pool ratio is meaningless: a few hundred sats on a heavy underdog made that underdog render as the favorite. The card was telling people the opposite of the truth. Now a fixed odds market shows the book line, always, and it does not drift when someone bets.

How to tell which engine you are looking at: watch the bar after money moves. Pool based percentages shift with every position. Fixed odds percentages hold steady while the pools grow beneath them. The bet slip is the other tell: a fixed odds slip quotes you a fixed return, a pool based slip quotes a projection that can move.

When the percentage is noise, and you should ignore it

A brand new market

Before any money has arrived, a pool based market has nothing to derive a price from, so it displays 50/50. That is a placeholder, not a forecast. The first meaningful reading arrives with the first real positions.

A thin pool

A pool based market carrying a few thousand sats is a market whose price one person can set. That is an opportunity if you have a view and a trap if you assumed the number encoded a crowd's wisdom. Check the pool total and the unique bettor count on the card before you weight the percentage at all.

A market with visible house liquidity

On fixed odds markets we post two sided liquidity behind our own line so the card is not sitting at an empty zero. That inventory is returned at settlement rather than paid out as a winner, so it never distorts the economics, but it does mean pool size on a book market is not a crowd signal. On those markets, read the line and disregard the pool depth entirely.

A suspended line

Sports odds get pulled on team news and near lock. When our hourly refresh finds no live price for an event, betting pauses on that market rather than continuing to deal a stale line. If you get "odds are being updated", that is the guard working. The last displayed number was already out of date.

What the percentage silently leaves out

Time. Every market has a close time that comes before the event resolves. A high percentage on a market closing in five minutes and the same percentage on one closing in three days are not the same bet, because the second one has time for the world to change while your sats are committed.

Resolution source. This is where prediction markets differ fundamentally from our casino games, and it is worth being direct about. Our dice game and the rest of the casino floor are provably fair: a hashed server seed, your client seed, and an incrementing nonce combine into every roll, and after the seed is revealed you can recompute any result yourself and confirm nothing moved. Prediction markets cannot work that way, because the outcome lives in the real world, not in a hash. A market resolves against an external data source: the price feed, the mining difficulty epoch, the weather station, the final score. So the question you ask of a prediction market is not "was the roll fair" but "is the resolution criterion stated precisely enough that there is no argument". Read the resolution terms on the card before you read the percentage.

Position limits. Each market publishes a minimum and a maximum, and your total exposure across multiple positions on one market is capped at that maximum. If you intended a large position, size it in one go rather than discovering the cap halfway in.

Cancels and pushes. If an event is cancelled or lands exactly on a push condition, every position is refunded in full with no fee taken. A percentage never shows you the chance of a void, but it exists, and it is why a "sure thing" market sometimes just hands your sats back.

Reading probabilities across the market types we run

Different subject matter demands a different level of trust in the number.

Bitcoin price markets

Fast moving and genuinely uncertain. Crowd pricing works reasonably here because participants share the same public information, and it updates continuously. Percentages near the money on a short horizon deserve the most respect of anything we list.

Mining difficulty markets

These resolve on the difficulty adjustment at the end of an epoch, and the outcome becomes progressively more knowable as the epoch fills in. Late in an epoch, the estimated adjustment is close to arithmetic rather than opinion. A percentage that lags what the blocks already imply is the clearest mispricing available on the whole board, and it is available to anyone willing to look up the current epoch progress.

Mempool fee markets

Fee levels are mean reverting and seasonal in a weekly pattern. The market's percentage often reflects the current fee level more than the level at resolution.

Weather markets

Resolved against a stated weather station for a stated city and day. Public forecasts are strong here, so the honest question is whether the market's percentage has caught up to the forecast yet.

Sports markets

Where you are most likely to meet a fixed odds book rather than a pool. Reread the fixed odds section above. The number is a posted price, locked at placement, and it will not move because you bet.

Grade your own reading with calibration

The single best exercise for getting good at this costs almost nothing, because our minimum position is 10 sats.

Take small positions across many markets. Log the percentage you accepted each time. Then bucket your results: of everything you took at 60 to 70 percent, how often did it actually resolve your way? If the answer is around 65%, the market was calibrated and you were paying a fair price. If those positions landed 80% of the time, the market was systematically underpricing that category and you found an edge worth pressing. If they landed 40% of the time, you were paying for stories.

Calibration is the only feedback loop that distinguishes a real read from a lucky one, and it needs volume, not size. Twenty 10 sat positions teach you far more than one 200 sat position.

Place your first prediction

Get sats in the balance. Claim the free faucet, work through the earn surfaces, or deposit over Lightning. Deposits and withdrawals both run through LNURL: scan the QR with any Lightning wallet, your wallet handles the invoice, and it settles in seconds. Everything is denominated in sats, never in fiat.
Open the markets board and pick a category you genuinely follow. Difficulty and weather are the friendliest starting points because the resolution criteria are unambiguous.
Read the resolution terms first, the percentage second. Confirm you know exactly what makes it settle YES.
Identify the engine. Does the bar move when the pool moves? Pool based, and your payout floats. Does it hold steady? Fixed odds, and your price locks.
Convert the percentage to a break even hit rate and ask honestly whether you would take that side that often.
Size small and spread wide. Ten positions at 10 sats will teach you the mechanics faster than any article, including this one.

Once you are comfortable reading a price, the same skill transfers directly to multiplayer poker, where pot odds are the identical calculation with different labels: the pot quotes you a price, and you decide whether your hand comes in often enough to beat it. Prediction markets are simply that question asked about the world instead of about a deck of cards.

Frequently asked questions

What does a 65% probability on a prediction market actually mean?: It is a price. A side quoted at 65% costs about 65 sats for every 100 sats it returns, so it pays roughly 1.54 times your stake if it wins. Read it as a break even hit rate: that side has to come in more than 65% of the time for the position to be profitable over many bets.
How do I convert a percentage into a payout multiple?: Divide 100 by the percentage. A side at 25% returns 4x your stake, a side at 50% returns 2x, and a side at 80% returns 1.25x. To go the other way, divide 100 by the multiple: a market offering 3x is quoting 33.3%.
Why does the probability change after I place a bet on some markets but not others?: Lightning Faucet runs two pricing engines. Pool based markets derive the percentage from where the sats sit, so it moves with every position and your projected payout floats until the market closes. Fixed odds sports markets post a price up front and lock it at the moment you bet, so the displayed line holds steady while the pools grow beneath it.
Why does a pool based market show YES as the other side's pool share?: Because that is what a YES position is actually worth. If YES wins, YES holders split the NO pool, so the value of backing YES depends on how much money is sitting on NO. Heavy money on NO makes YES cheap and valuable, and heavy money on YES makes it crowded and thin.
Should I trust the percentage on a brand new market?: No. A pool based market with no money in it displays 50/50 as a placeholder, not a forecast, and a market carrying only a few thousand sats can have its price set by one participant. Check the pool total and unique bettor count before you give the number any weight.
Are prediction markets provably fair like the casino games?: They work differently. Casino games such as dice use a hashed server seed, your client seed, and an incrementing nonce, so you can recompute any result yourself once the seed is revealed. Prediction markets resolve against real world data instead: a price feed, a difficulty epoch, a weather station, or a final score. The thing to verify is the resolution criterion on the market card, which should be precise enough to leave no argument.
What happens to my position if the event is cancelled?: Cancelled markets and pushes refund every position in full, with no settlement fee taken. That is why a market that looks like a certainty sometimes just returns your sats instead of paying out.
What is the smallest prediction I can make?: Most markets take positions from 10 sats, and each market publishes its own minimum and maximum on the card. The small minimum is the point: taking many tiny positions and then checking how often the ones you accepted at 60 to 70 percent actually landed is the fastest way to learn whether you are reading prices well.