Expected value (EV) is the single number that separates a reasoned decision from a random one in prediction markets. Yet most of what is written about it explains the idea and never gives you the arithmetic — and worse, it borrows the fixed-odds formula ("buy at 0.40, collect 1.00") that does not apply at all to PolySouq, because PolySouq runs on a parimutuel pool model: your return is not a fixed price but a moving ratio between the size of the losing side and the size of the winning side. In this guide we derive the expected value formula directly from the real settlement equation payout = stake × (1 + S_lose/S_win × 0.9), give you the five steps to calculate EV with numbers, and three fully worked examples — one of which actually lost money despite a positive expected value — along with ready-made tables for breakeven probability, an EV grid across a range of estimates, and the effect of your position size on the maths. Event trading involves real money and losing the full amount you commit is entirely possible.
Expected value is the average financial outcome of your decision if the same situation repeated a large number of times under the same conditions. It is not a prediction of what will happen in the next event, and it is not a promise of profit — it is a probability-weighted average: multiply every possible outcome by its probability, then add them up. If the result is positive, the decision is profitable on average over the long run; if it is negative, you are paying more than the forecast is worth, however convincing it feels. This is the core of EV in trading, the same concept you find in options pricing and portfolio management, and we covered its conceptual background in expected value in prediction markets.
The problem is that most sources explain the expected value formula using the fixed-odds model: you buy a contract at $0.40 and collect
The practical difference is enormous. In the fixed model your return is known the moment you enter and never changes. In a parimutuel pool your return depends on a ratio between two numbers that keep moving until the market closes: the total staked on the losing outcomes, and the total staked on your own side. The more crowded your side becomes, the lower your return; the more crowded the opposite side, the higher it goes. Any EV calculation that does not start from that ratio is a misleading calculation, and it can push you into a position you believe is profitable when it is in fact negative. That is where this guide begins: from the real settlement equation, not a borrowed one.
Before we derive the formula, make sure the probability basics are clear — how to read a percentage, the difference between probability and frequency, and why a 70% probability does not mean the outcome is "nearly certain". Review probability basics for prediction markets if you need a quick foundation, then come back here for the arithmetic.
We start from the operational reality. When a market settles on PolySouq, what returns to each participant on the correct side is calculated with the following formula, where S_win is the total staked on the winning outcome, S_lose is the total staked on all other outcomes, and c is the commission rate of 10%:
payout(i) = stake(i) × ( 1 + S_lose / S_win × (1 − c) )
Note three essential properties. First, your original stake comes back to you in full on top of your share of the profits — the "1" inside the bracket is your own money. Second, commission is taken from the losing pool only, which is why the factor (1 − c) = 0.9 sits attached to S_lose and not to your stake. Third, your profit is not a fixed number but your stake multiplied by the pool ratio. To simplify everything that follows, we define the profit factor:
g = 0.9 × ( S_lose / S_win )
With that definition, your net profit if you win is s × g, where s is your position size, and what lands in your account if you win is s × (1 + g). If you lose, you forfeit s entirely. Now apply the definition of expected value: multiply every outcome by its probability and add. If p is your own estimate of the probability that the outcome you backed occurs, then:
EV = p × (s × g) + (1 − p) × (− s) = s × [ p × g − (1 − p) ]
That is the correct expected value formula for PolySouq. An equivalent and useful check on your arithmetic is EV = p × payout − s, that is, the expected value of what reaches your account minus what you paid. The two forms always give the same number, and using one to verify the other is an excellent habit that exposes calculation errors immediately. To get the expected return as a percentage, divide by your stake: EV% = p × g − (1 − p).
From this equation we extract the single most important number in the file: the breakeven probability, the minimum probability that makes your decision break even — neither profitable nor loss-making. Set EV = 0 and solve for p to get p* = 1 / (1 + g). The practical rule is simple: enter only when your estimate p is greater than p*, and the wider the gap between them, the greater the value. How wide that gap needs to be is tabulated in a later section.
This is the executable recipe. You can run it in your head in under a minute, or build it into a simple spreadsheet that becomes your own expected value calculator. The steps are ordered, and skipping any one of them is the source of most of the errors we catalogue later in this guide.
An important note on the first step: the numbers you read are a snapshot in time. The pool keeps moving until the market closes, which means the final g at settlement may differ from the g you calculated on. This is a structural feature of a parimutuel pool and there is no way around it; we devote a full section to it later. The practical takeaway is that you treat EV as a live estimate rather than a final number, and you prefer wide edges over marginal ones, because only wide edges can absorb the pool moving against you.
A note on the fourth step: if you cannot state the reason for your estimate in a single sentence explaining why you are more accurate than the rest of the participants, you probably are not, and the right decision is to pass. Passing is a legitimate decision with an expected value of exactly zero — and zero beats a negative number. To understand what moves market estimates in the first place, review why prediction market odds change.
Take a binary market on the TASI index: "will the index close above a specified level by month end?" Assume the pool before you enter is split as follows: Yes = 6,000 USDC and No = 4,000 USDC, a total of 10,000. You have studied the index's seasonality and foreign flows and believe the probability of "No" is higher than the pool reflects, so you decide to commit 100 USDC to "No".
Step 2: after adding your position, your side's pool becomes S_win = 4,000 + 100 = 4,100, the opposite pool stays at S_lose = 6,000, and the total is 10,100.
Step 3: g = 0.9 × (6,000 ÷ 4,100) = 0.9 × 1.4634 = 1.3171. That is, you earn
Implied probability: your side's share of the pool is q = 4,100 ÷ 10,100 = 40.59%. That is the crowd's estimate. But your breakeven point sits above it because of the commission: p* = 1 ÷ (1 + 1.3171) = 43.16%. In other words, you do not merely need to be more accurate than the crowd — you need to be more accurate by more than 2.57 percentage points.
Steps 4 and 5: your independent estimate is p = 55%, clearly above 43.16%. Substituting:
EV = 100 × [ 0.55 × 1.3171 − 0.45 ] = 100 × [ 0.7244 − 0.45 ] = 100 × 0.2744 = +27.44 USDC
Check with the alternative form: 0.55 × 231.71 − 100 = 127.44 − 100 = +27.44. The two numbers match, so the calculation is sound. The expected return is 27.4% of the amount committed. In this example the index closed below the target level, the market settled in favour of "No", and 231.71 USDC reached your account. But note carefully: winning here did not prove your calculation was right — it would have remained a good decision even if it had lost, which is exactly what the third example demonstrates.
This is the example that exposes the single most common mistake of all: confusing a "likely forecast" with a "valuable forecast". Take a market on OPEC+ decisions framed as: "will the group approve a production increase at this month's meeting?" The pool before you enter: Yes = 9,000 USDC and No = 3,000 USDC, a total of 12,000. You also favour "Yes" and commit 200 USDC.
After your entry: S_win = 9,000 + 200 = 9,200 and S_lose = 3,000, total 12,200. So g = 0.9 × (3,000 ÷ 9,200) = 0.9 × 0.3261 = 0.2935. What reaches your account on a win = 200 × 1.2935 = 258.70 USDC, a net profit of just 58.70. The reason is obvious: your side is crowded and the opposite pool is small, so the prize you will be splitting is tiny relative to how many of you there are.
Implied probability and breakeven: q = 9,200 ÷ 12,200 = 75.41%, and p* = 1 ÷ 1.2935 = 77.31%. Your independent estimate is p = 72% — a very high probability by human standards, meaning you expect to win in roughly seven cases out of ten. And yet:
EV = 200 × [ 0.72 × 0.2935 − 0.28 ] = 200 × [ 0.2113 − 0.28 ] = 200 × (−0.0687) = −13.74 USDC
Check: 0.72 × 258.70 − 200 = 186.26 − 200 = −13.74. Identical. So this decision loses 6.9% of the amount on average even though it will win most of the time. Small frequent gains do not offset large rare losses here, because the market prices the outcome above your own estimate. The correct decision: pass. Do not enter a position because "its chance of success is high"; enter it because p exceeds p*.
And note the other face of the same market: whoever believes the probability of "No" exceeds their own breakeven is the one with value here. The opposite side's breakeven is calculated the same way after adding their position to the "No" pool. This symmetry is what makes a parimutuel pool a market rather than a game: each side has its own breakeven point, and the gap between it and your estimate is the only source of value.
This example serves two purposes: generalising the formula beyond two outcomes, and showing the case everyone dealing with EV must understand — an arithmetically sound decision that ends in a cash loss. Take a match-outcome market in the sports section with three outcomes. The pool before you enter: Team A win = 5,000, draw = 2,000, Team B win = 3,000, a total of 10,000. You believe the draw is underpriced and commit 150 USDC to it.
The generalisation: in a multi-outcome market, S_lose is the sum of all the other pools. After your entry: S_win = 2,000 + 150 = 2,150 and S_lose = 5,000 + 3,000 = 8,000, total 10,150. So g = 0.9 × (8,000 ÷ 2,150) = 0.9 × 3.7209 = 3.3488. What reaches your account on a draw = 150 × 4.3488 = 652.33 USDC, a net profit of 502.33.
The calculation: q = 2,150 ÷ 10,150 = 21.18%, and p* = 1 ÷ 4.3488 = 22.99%. Your independent estimate is p = 30%, roughly 7 percentage points above breakeven:
EV = 150 × [ 0.30 × 3.3488 − 0.70 ] = 150 × [ 1.0047 − 0.70 ] = 150 × 0.3047 = +45.70 USDC
An expected return of +30.5%. Then the match ended with a Team A win. The market settled, and you lost the full 150 USDC. Was the decision wrong? No. The decision was right and the outcome was bad, and those are two different things. Imagine ten similar decisions under the same conditions: you win three of them for 3 × 502.33 = 1,506.99 and lose seven for 7 × 150 = 1,050, netting +456.99 — roughly 10 × 45.70, exactly as the formula predicted. The catch is that the ordering of those ten is random, and the seven losses may come first.
An operational note specific to football markets: you can cancel your position before kick-off and get your full amount back. That option is available only before the match starts, and after it there is no exit at all — the position runs to settlement. So if your core inputs change (an injury, a line-up, the weather) before kick-off, that is the only window in which to recalculate and act.
The implied probability is what the pool says about the probability of an outcome, and it is simply that outcome's share of the total pool: q = S_win / (S_win + S_lose). Many people stop here and compare their estimate directly against q, and that mistake costs money. The correct comparison is not with q but with the breakeven probability p*, which includes the effect of the commission. Substituting for g in terms of q gives a direct and convenient form:
p* = q / ( 0.9 + 0.1 × q )
This means you can calculate the breakeven point from the displayed percentage alone, without knowing the amounts. A quick example: if your outcome's share is 50%, your breakeven is 0.50 ÷ (0.9 + 0.05) = 52.63%. You need an estimation edge of 2.63 percentage points over the crowd just to break even. The table below computes that "burden" across the whole range:
The table reveals an important pattern: the commission burden in percentage points peaks near the middle of the range (about 2.6 points at 50%) and shrinks at the extremes. In practice this means finely balanced markets are the hardest in terms of the estimation edge required, while long-shot or near-certain outcomes need a smaller edge in points — but be careful, a smaller edge in points at the extremes can be a larger one in relative terms: seeing 12% where the market sees 10% is a claim that you are 20% more accurate in relative terms, and that is a heavy claim.
There is a technical point worth noting in multi-outcome markets: the displayed percentages may not sum to exactly 100% when only the leading outcomes are shown, or because of rounding. When calculating, always rely on the safe rule S_lose = total − S_win rather than adding the percentages one by one. To see how markets are resolved and how these numbers are applied at settlement, review how PolySouq settles markets transparently.
The table below is the closest thing to a ready-to-use expected value calculator. The columns are the pool's implied probability q after your position is added, the rows are your own estimate p, and the values inside the grid are the expected value in dollars per 100 USDC committed. To read the result for any other amount, simply scale: a 250 USDC position gives 2.5 times the number shown.
Three readings deserve a pause. First: the cell at p = 50% with q = 50% equals −5.00, not zero. When your estimate matches the crowd's exactly, you are not at breakeven — you are losing by the size of the commission. This is the "tax" paid by anyone trading without an edge, and it alone explains why event trading cannot be a source of income for anyone without an independent estimate.
Second: the left-hand column (q = 20%) produces large numbers because its profit factor is g = 3.6. But be careful — those numbers are conditional on your estimate actually being correct. Estimation error on long-shot outcomes is the most common of all, because the estimate there rests on small historical samples. Expected value rewards you for accuracy and punishes overconfidence just as sharply.
Third: the p = 80% row ends with a negative value at q = 80% (−2.00). Once again: a high probability does not mean positive value. The dividing line in every column is the breakeven point from the previous table — every cell above it is positive and every cell below it is negative. If you want to build your own version in a spreadsheet, the only formula you need in each cell is = 100 × ( p × 0.9 × (1−q)/q − (1−p) ).
This property is entirely absent from fixed-odds models and it is one of the defining features of a parimutuel pool: your position size changes your return. Every dollar you add to your side raises S_win and lowers g — you are splitting the same prize with a bigger version of yourself. The result is that the expected return as a percentage falls as your position grows, while the total expected value rises at first, peaks, and then turns negative.
Let us measure it with numbers on the first example's pool (your side 4,000, the opposite side 6,000, and your estimate p = 55%):
Read the last two columns together. The relative return falls without interruption from 28.3% to below zero, while the absolute value peaks near a position of about 1,140 USDC and then retreats. At 4,000 USDC you have pushed your breakeven from 43.16% to 59.70% — meaning that with your position alone you converted a good decision into a losing one. This is not theory: if you represent a meaningful share of a small pool, you are trading against yourself.
A necessary warning: the size that maximises expected value is not the correct size. Maximising EV ignores volatility, the chance of losing streaks, and the fact that your estimate p may itself be wrong. Position size must be derived from capital-management rules — a small fixed percentage of your balance, a cap on exposure to any single event, and a cap on exposure to a shared factor (several markets that all move with the price of oil in the oil and energy section, for instance). Review risk and capital management before you translate any number in this guide into a real amount.
The practical conclusion: always calculate EV at the size you actually intend to execute, not at a small hypothetical size. And if you find that the size your capital rules call for kills the value through dilution, that is a signal the pool is too small to absorb you, and that waiting or reducing the position is the right decision.
Behaviourally, this is the most important section in the guide. Positive expected value is not a promise of profit — it describes a theoretical average that materialises only across a large number of independent decisions. The third example above produced EV = +45.70 and ended in a full 150 loss, and that is the natural, expected behaviour of a decision with a 30% chance of success: the most likely outcome of any single such decision is a loss, and the value comes from how large the wins are relative to how rare they are.
Three sources make the gap between the calculation and reality wider than beginners assume. First: volatility and sample size. Even with a genuine edge, consecutive losing streaks are statistically common. Seven losses in ten decisions at a 30% probability is not exceptional — it is the expected distribution, and it may be preceded or followed by entirely different stretches. Anyone measuring their performance over ten decisions is not measuring skill; they are measuring luck.
Second: estimation error itself. The formula assumes p is correct. In reality p is your own estimate, exposed to overconfidence, confirmation bias, and reading news stories as if they were data. If you think your edge is five percentage points when it is in fact zero, you are paying the commission on every decision and calling it "bad luck". The only remedy is recording your estimates in advance and comparing them against actual outcomes across dozens of decisions — that is, calibration — not just recording your profits.
Third: pool movement after your entry. The g you calculated on is a snapshot. If additional money flows onto your side before close, your actual return falls below what you computed; the reverse is also true. You cannot exit or cash out your position mid-event to lock in a gain or cap a loss — the position runs to settlement, with the sole exception of the pre-kick-off cancellation window in football markets, which returns your full amount. For exactly this reason, marginal gaps between p and p* are not worth executing: a half-point margin can be erased by a single inflow onto your side.
The mental rule that protects capital: judge the decision by its inputs, not by its outcome. A positive-EV decision that ended in a loss remains a good decision you should repeat; a negative-EV decision that ended in a profit remains a bad decision you should stop making. The money, meanwhile, is real, and losing the full amount committed is a possible outcome in every position.
We see the following errors repeatedly from people importing EV formulas from sources designed for different models. Each one changes the final number materially, and some flip its sign from negative to positive — the most dangerous kind, because it pushes you into a losing position while you feel confident.
The first mistake in the table is the most widespread because it comes from foreign content written for order-book models. The difference between it and the correct calculation is not cosmetic: in a crowded pool the wrong model can hand you a calculated return several times the real one. The sixth mistake — deriving p from the pool — is the most insidious, because it looks methodical and objective while it is in fact a mathematical guarantee that your expected value will be negative over the long run.
A simple protective habit: before any execution, write four numbers on one line — g, q, p*, p — then the fifth, EV. If you cannot write the first four, you do not yet know what you are doing. That line takes under a minute, and it is the difference between systematic event trading and guesswork. For the mechanism from the ground up, review how prediction markets work.
The question is legitimate and deserves a structural answer, not a slogan. On a traditional betting site, your counterparty is the company itself: it sets the odds, embeds a margin in its own favour, and profits directly when you lose. Its interest lies in your loss, and the odds displayed are not an estimate of the truth but a price engineered to secure a guaranteed margin regardless of the outcome.
In the parimutuel pool model that PolySouq runs on, that counterparty does not exist. The platform takes no opposing position and sets no probability; the displayed probabilities are simply participants' shares of the pool — the market's opinion about itself. Those who split the losing side's money are the participants who estimated the outcome correctly, not the company. The platform's role is purely operational: aggregate the pool, resolve the outcome, and distribute it by the published formula. That is precisely why we describe the activity as event trading: you estimate a probability, compare your estimate against a displayed collective estimate, and execute when you find a gap — the same methodology used in derivatives pricing.
The second difference lies in the platform's revenue source. The 10% commission is not taken from your stake nor from the total pool, but from the losing pool only. Two consequences follow that exist in no betting model: if nobody stands on the other side, there is no losing pool, so everyone gets their money back in full and the commission is zero. And if nobody backed the outcome that occurred, the market is voided, all stakes are returned, and the commission is zero as well. The platform earns only when there are actual profits to be split.
The third difference is the arithmetic solvency guarantee. Every market is checked so that the sum of payouts plus commission equals exactly the total staked into the pool — no more and no less. There is no open-ended liability on the platform and no promise of a return the pool does not cover, which is what makes the mechanism auditable arithmetically by any participant. Questions of religious ruling have their own separate place, and we addressed them in is prediction market trading halal?.
One thing remains to be said plainly: a different structure does not remove the risk. The money is real, in USDC on the Polygon network, losing the full amount committed is a possible outcome in any position, and nothing in this guide constitutes personal financial advice.
Calculation without execution discipline is worth nothing. Before every position, run this checklist in full. If any item fails, the default decision is to pass — and passing has an expected value of zero, which beats every negative number in this guide's tables.
The money mechanics are simple and direct. Account creation is free. Deposits are made by sending USDC on the Polygon network to a personal deposit address tied to your account, with a minimum deposit of 0. There is no wallet connection and no browser-extension signing — the model is a deposit address you send to. Withdrawals are made in USDC only, to a Polygon address you provide; no bank transfer, no IBAN, no card withdrawal. Full details are in deposits, withdrawals and fees, and step by step in how to withdraw USDC.
On refunds, the rule is one and the same: refunds are always full, never partial. If the operator cancels or voids a market, all amounts are returned in full. If nobody stands on the other side, everyone gets their money back and the commission is zero. If nobody backed the outcome that occurred, the market is voided and all stakes are returned. And in football markets, cancelling your position before kick-off returns your amount in full.
Risk statement: event trading on PolySouq is conducted with real money, and losing the entire amount committed to any position is an entirely possible outcome. There is no demo account, no virtual money and no signup bonus, and you cannot exit or cash out a position mid-event. Positive expected value does not guarantee a profit on any single decision, nor over any limited number of decisions. Everything here is general educational content and does not constitute personal financial or investment advice, and it does not take account of your circumstances, objectives, or ability to bear a loss. Do not commit money you need.
If you want the full path from zero — from understanding the mechanism to your first position — start with opening an account and your first trade, then apply this guide's method to a single market with a small amount until you are satisfied that your arithmetic produces the same numbers you see on the market page. Browse what is available through the markets page and pick a market where you hold genuine knowledge, not one that merely looks exciting.
The formula is EV = s × [ p × g − (1 − p) ], where s is your position size, p is your estimate of the probability that the outcome occurs, and g is the profit factor calculated as g = 0.9 × (S_lose / S_win). This equation is derived directly from the real settlement formula payout = stake × (1 + S_lose/S_win × 0.9). An equivalent form for checking is EV = p × payout − s, which always gives the same number. Do not use the fixed-odds formula ("buy at 0.40, collect 1.00") — it is designed for a completely different order-book model and does not apply to a parimutuel pool.
Five steps. First, record the total staked on your side and on all other sides combined. Second, add your amount to your own pool, so S_win = the previous pool + your amount. Third, calculate g = 0.9 × S_lose/S_win. Fourth, estimate your probability p from data entirely independent of the pool. Fifth, substitute into EV = s × [p × g − (1−p)], then confirm that p is greater than the breakeven point p* = 1/(1+g). Example: your pool at 4,100 against 6,000 gives g = 1.3171, so with an estimate of 55% and an amount of 100, EV = 100 × (0.55 × 1.3171 − 0.45) = +27.44 dollars.
The breakeven probability is the minimum probability that makes your expected value zero, calculated as p* = 1 / (1 + g), or directly from the displayed percentage as p* = q / (0.9 + 0.1q), where q is your outcome's share of the total pool. Example: if your outcome's share is 50%, your breakeven is 52.63%; if it is 70%, your breakeven is 72.16%. The practical rule is to compare your estimate against p* rather than against q, because q ignores the effect of the commission. The commission burden in percentage points peaks near the middle of the range at about 2.6 points at 50%, and shrinks at the extremes.
No, and the difference is structural rather than linguistic. On a betting site the company is your counterparty: it sets the odds, embeds a margin in its own favour, and profits directly from your loss. On PolySouq that counterparty does not exist: the platform takes no opposing position and sets no probability, and the displayed percentages are the participants' own shares of the shared pool. Those who split the losing side's money are the participants who estimated the outcome correctly. Commission is taken from the losing pool only, so if nobody stands on the other side, everyone gets their money back in full and the commission is zero. The activity is event trading: estimating a probability, comparing it against a displayed collective estimate, and executing when a gap exists.
Because expected value is a theoretical average across a large number of decisions, not a prediction of the outcome of a single one. In the worked example in this guide, the expected value was +45.70 dollars on a position with a 30% chance of success, and the most likely outcome of any such decision is a loss — the value comes from how large the wins are relative to how rare they are. Imagine ten similar decisions: you win three for 1,506.99 and lose seven for 1,050, netting +456.99 exactly as the formula predicted, but the ordering of those ten is random and the seven losses may come first. Judge the decision by its inputs, not by its result.
Commission is taken from the losing pool only, which is why the 0.9 factor sits attached to the S_lose/S_win ratio and not to your stake. Your original amount returns to you in full on a win, on top of your share of the profits, and the platform earns only from actual profits. Its arithmetic effect is to lift the breakeven above the implied probability: at a 50% share, the breakeven rises to 52.63%, meaning you need an estimation edge of 2.63 percentage points just to break even. The common mistake is deducting 10% from your stake or from the total pool — both give the wrong number.
Yes, and this is a structural feature of a parimutuel pool. The number you calculate is a snapshot built on the pool figures at that moment, but flows continue until the market closes. If extra money comes in on your side, S_win rises and the profit factor g falls, so your actual return is lower than you calculated; if money flows onto the opposite side, your return rises. This is why marginal gaps between p and p* are not worth executing: a half-point margin can be erased by a single inflow. Always prefer wide gaps that can absorb the pool moving against you.
Yes, directly — and this is what distinguishes a parimutuel pool from fixed-odds models. Every amount you add raises S_win and lowers the profit factor g, meaning you split the same prize with a bigger version of yourself. In the worked example in this guide, the expected return falls from 28.3% on a 50-dollar position to 14.4% at 1,000 dollars, then flips to −7.9% at 4,000 dollars because the breakeven rises from 43.16% to 59.70%. Even so, the size that maximises expected value is not the correct size — position size must be derived from capital-management rules, not from maximising the number.
Account creation is free. Deposits are made by sending USDC on the Polygon network to a personal deposit address tied to your account, with a
No. There is no selling or cashing out of a position mid-event; the position runs to settlement. The only exception is football markets, where you can cancel your position before kick-off and get your full amount back — that is the only window to re-estimate and act if core inputs such as injuries or the line-up change. In practice this means the entry decision is the only decision you own, which is why the arithmetic must be settled before it and not after. There are also no stop-loss tools, no leverage and no margin on the platform.
You can build your own in a simple spreadsheet in minutes, and all you need is four input cells and three formulas. The inputs: your side's pool before you enter, the other sides' pool, your amount s, and your estimate p. The formulas: g = 0.9 × S_lose / (your side's pool + s), then p* = 1/(1+g), then EV = s × (p × g − (1 − p)). If the interface shows percentages instead of amounts, convert them with S_lose/S_win = (1 − q)/q. For a quick per-100-dollar calculation straight from the displayed percentage, use EV = 100 × ( p × 0.9 × (1−q)/q − (1−p) ).
The implied probability q is what the pool says: your outcome's share of the total amounts staked, that is, the participants' collective estimate. p is your own estimate, and it must be derived from entirely independent data — statistical history, official statements, economic releases, technical inputs — not from the percentage displayed in front of you. If you take p from the pool itself, you are reasoning in a circle and the arithmetic result will always be negative by the size of the commission. The only source of value in event trading is the gap between p and p*, and if you cannot state a clear reason why you are more accurate than the rest of the participants, you probably are not, and the right decision is to pass.
Disclaimer: Prediction markets are a legal and legitimate way to trade information about the outcomes of future events. However, trading carries risk and you may lose the full amount you trade — so only trade what you can afford to lose. This content is educational and is not financial or investment advice.