How Capy Works

sportsbooks price every game with a built-in margin. most bettors never question whether that price is fair. capy does.

we pull live odds from major sportsbooks and compare them against Pinnacle — the sharp market reference used by professional bettors worldwide. when a book is offering more than the true probability implies, that gap is positive expected value (+EV).

capy is a pricing tool. no forecasts, no picks. just the math on whether a price is fair relative to the sharpest line in the market.

expected value (EV) measures how much a bet is theoretically worth per dollar wagered, relative to the true probability of the outcome. a +2% EV bet means that, on average, every $1 risked returns $1.02 before variance. that edge is small. it only compounds into profit over many bets.

capy derives EV using Pinnacle's two-sided moneyline as the reference:

EV is calculated only on moneylines (head-to-head markets). spreads and totals are excluded — the fair-value calculation is less reliable on those markets because half-point differences and alternate lines create too much ambiguity in matching Pinnacle's reference.

EV shown on capy is a snapshot at the time the data was last refreshed — typically within the past hour. live odds move constantly. a +2% EV opportunity may no longer exist by the time you see it.

what this means in practice

• —edges are small. typical +EV plays run +1% to +5% — not +20% or +30%.
• —most bets will still lose. a +3% EV edge on a ~50/50 moneyline still loses roughly half the time.
• —value accumulates with volume. a handful of +EV bets proves nothing. hundreds do.
• —short-term results are noise. a hot or cold stretch of 15–20 bets tells you almost nothing about edge.

a sharp line is a price set by a market that takes large bets from professional bettors without restricting or banning them. because sharp bettors bet into correct prices aggressively, books that accept them move to accurate lines faster. the result is a lower margin and a more reliable fair-value reference.

Pinnacle is used as capy's reference because it meets all three criteria that matter for this purpose:

• —lowest margin. Pinnacle's vig is typically 2–3%, compared to 8–10% at mainstream recreational books. a lower margin means the embedded fair probability is closer to the true market price.
• —accepts sharp action. professional bettors can bet Pinnacle at high limits without being limited. this means Pinnacle lines are stress-tested by informed money, not just recreational action.
• —widely used as a benchmark. Pinnacle is the standard reference used by sharp bettors, odds traders, and analytics tools globally. using the same reference allows meaningful comparisons over time.

capy uses Pinnacle data via the Odds API (EU region). this is not a direct integration with Pinnacle's live feed. prices are cached and refreshed on a scheduled cron — usually hourly.

small timing gaps between the cached price and the live Pinnacle price are normal. the displayed Pinnacle odds may occasionally differ by a few cents from what you'd see if you checked Pinnacle directly at that moment.

closing line value measures how the Pinnacle line moved between the time a pick was logged and the moment the game started. it compares the implied probability at placement against the implied probability at close.

CLV is considered the most reliable process-based metric in sports betting because it measures whether you are consistently getting better prices than the sharp closing market — not whether you won. wins and losses are heavily influenced by randomness on any individual bet. CLV is much harder to fake over a large sample.

a bettor with consistently positive CLV is getting better prices than the market ultimately settles at. over a large enough sample, that advantage shows up in results. a bettor with consistently negative CLV is consistently getting worse prices than the market — a sign of timing lag or poor line shopping, not edge.

what this means in practice

• —positive CLV is the signal to watch. consistent +CLV over a large sample is more meaningful than short-run roi.
• —a winning record without positive CLV may be luck. a losing record with positive CLV may be variance.
• —CLV only becomes reliable at scale. below 30 picks, a single outlier dominates the average.
• —use CLV to evaluate process, not to predict future results.

at typical edge sizes (+1.5% to +5% EV), variance dominates results for a very long time. the intuition most people have about "how many bets is enough" is usually off by a factor of 10.

rough benchmarks for moneyline bets near even money:

a +30% roi from 8 picks is a coin-flip result — it would appear frequently from a completely random strategy. capy suppresses green/red color coding on roi and win rate until minimum sample thresholds are met, but even after those thresholds are crossed, the numbers reflect recent performance under variance, not confirmed edge.

the metrics that carry the most signal at small samples are process-based ones: EV at pick time and CLV. both measure whether you captured a good price relative to the sharp market reference. they do not depend on game outcomes and are therefore much less distorted by short-run variance.

sports betting edges are real but small, and they exist within a high-variance environment. a consistent +3% EV edge is strong. it will still produce losing months, losing streaks of 10 or more bets, and stretches where results feel completely random — because they partially are.

what capy can tell you:

• —which sportsbook is offering the best available price on a given game right now
• —whether a price is above or below the sharp market's implied fair value
• —whether tracked picks are beating the closing Pinnacle line over time
• —how pick performance compares across sports, EV buckets, and pick types

what capy cannot tell you:

• —who will win a specific game
• —whether any individual bet will be a winner
• —that a positive-EV pick has more than a marginal edge over 50/50
• —that short-run roi or win rate reflects real edge

use capy to evaluate whether you are getting good prices. evaluate performance on process — EV at pick time and CLV — not on short-run wins and losses.

this is for:

• —users who want to understand what the price actually means, not just which team to back
• —users comfortable with variance — losing stretches included — and willing to evaluate process over short-term results
• —users thinking in terms of hundreds of bets and long-term edge, not this week's record

this is not for:

• —users looking for a guaranteed edge or a service that picks sides for them
• —users betting based on narrative, matchup feel, injury reports, or gut instinct
• —users expecting consistent short-term profit from a small sample