How PolyRating Is Calculated

How we treat Polymarket: You pick one outcome (e.g. Yes or No, or one of several options). When the market resolves, we compare the resolved outcome to your predicted outcome. If they match, the signal is correct; if not, it’s wrong. There is no “magnitude” of move—only right or wrong.

How points work: Wins and losses scale with the implied probability (p) of your chosen outcome at lock-in, plus the same time multiplier (T) and E-Factor (EF) used for price signals. Favorites (p ≥ 65%) earn tiny wins (hard-capped). Competitive and underdog picks earn larger risk-based awards; underdogs get an extra win multiplier. Outcome horizons are mildly compressed so a week-long event is not several times a sub-hour one on time alone. All correct wins are capped at 30,000 points. Wrong favorites take a heavy loss multiplier. Daily Boost / Flash can double correct gains; Fame and engagement boosts apply to correct gains only.

In short: asymmetric risk by odds (not a fixed M), with T and EF shared with price signals, so calling a longshot correctly pays more than nailing a near-lock—while the 30k ceiling keeps extreme longshots bounded.

Price signals (LONG/SHORT): M and A depend on the actual price move and how close you were to target; gains and losses are then scaled by market cap (MC). Polymarket (outcome) signals: use asymmetric risk from implied probability p (favorites tiny; underdogs larger), not a fixed M — see “PolyRating for Polymarket events” above. Market-cap weighting does not apply to Polymarket.

• M (magnitude): For price signals, M = 10,000 × |actual % move|. The bigger the move (e.g. +18% → 0.18), the larger M. For outcome signals, risk magnitude R(p) replaces M.
• A (accuracy bonus): When you’re correct and the outcome is within the ±10% accuracy band, you get an extra bonus up to 25% of M. Closer to target = more of A.
• MC (market-cap factor): On price signals, we scale both gains and losses by the asset’s market cap so the same % move on a microcap does not pay or sting like a mega-cap. Moderate curve: mega (≥$50B) 1.0× → large 0.95× → mid 0.85× → small 0.55× → micro (under ~$300M) 0.30×. Missing market-cap data → 1.0× (no accidental scale).
• T (time multiplier): From your signal’s horizon (creation → expiry). Longer horizons (e.g. 1 week, 1 month) use higher T so bolder timeframes can reward more.
• CL (loss coefficient): Used only when the signal is wrong. Loss = −M × MC × CL × T × EF, so losses are scaled by market cap and horizon.
• EF (E-Factor): Depends on your current score (and Prestige). Higher score → lower EF, so the same prediction awards fewer points as you climb. After Prestige, EF stays at the hard high-score rate even though your number resets to 50k — see Catch-up mechanic below.
• FM (Fame multiplier): Optional. When the Fame integration is active, correct-signal gains are multiplied by FM based on your Fame score (1.0× by default, up to 1.50× at 25,000 Fame). FM applies only when the signal is correct and the gain is positive — losses are never reduced or amplified by Fame.
• EB (engagement boost): Because PolyRating is a reputation score, correct signals are amplified by the engagement they earn — up to +100% (a ×2 multiplier). Quality, on-topic discussion counts most, then likes, shares, and views. EB applies only when the signal is correct; it never softens a loss — see “Engagement & reputation” below.

Formulas

When you’re correct (price signals):

MC = 1 for mega-caps, missing market cap, or Polymarket. FM = 1 when Fame integration is off or your Fame is below 500. EB = 1 when a correct signal earned no engagement, up to 2 at maximum engagement. Neither FM nor EB ever applies to losses. MC applies symmetrically to price-signal losses.

When you’re wrong (price signals):

Your new score is the previous score plus this gain or loss, then clamped between 0 and 100,000. At 0, losses don’t go negative.

A −80% move on a bankruptcy microcap is not the same difficulty as a −80% move on a mega-cap. Market-cap weighting (MC above) scales both wins and losses on tiny, ultra-volatile names so the same % move does not dominate PolyRating the way a mega-cap move would.

When a company announces bankruptcy (or must leave Markets for fairness), we can soft-delist the ticker: it leaves browse/search, new signals are blocked, and existing signals for that asset are hidden from the feed. Predictions are not deleted — that remains a hard product rule — but bankrupt names will not keep dominating the competition narrative.

These scoring changes are forward-looking. Past verified scores stay as written history.

Building a community and earning real engagement is hard — and we want to make sure we reward it. A great call that nobody sees builds less reputation than a great call that sparks real discussion. So when a signal resolves correct, we look at the engagement it earned and apply an engagement boost (EB) to that gain — up to +100% (a ×2 multiplier).

This is why growing your audience matters: invite friends and build a following so your calls get seen and discussed. When you’re right, that engagement turns into a bigger reputation climb.

Fame measures participation: profile setup, publishing signals and comments, social engagement, verified referrals, and quality bonuses. Everyone starts at 0 at launch and builds Fame forward from there. It is not a substitute for Polyrating and does not make wrong signals cost less.

You can hide Fame on your public profile in Settings. Hidden Fame still accrues; you just won’t appear on the Fame leaderboard.

Full earning rules: How Fame works.

When you start at 0, predictions are closer to gambling than proven skill — you don’t have a track record yet. So E-Factor is high and correct signals award a lot of points. That lets new analysts get on the board quickly and learn the game without feeling stuck.

As you improve and move past roughly 50k — and especially once you’ve Prestiged — the same quality of call is worth fewer points. You need better and better predictions to keep climbing the leaderboard. That’s deliberate: it stops early lucky runs from permanently locking the top spots.

This is the same kind of catch-up mechanic used in most competitive games: newcomers can rise fast; veterans must keep proving themselves under tougher scoring. Prestige does not reset you to “easy mode” — after Prestige, you keep the reduced high-score E-Factor even while your displayed score is back at 50k.

• Low score: high EF → big gains (and bigger swings).
• High score (~75k+): low EF → smaller gains per win; climbing requires sharper calls.
• After Prestige: EF stays on the hard high-score curve so you can’t farm early-game points again.

Plain version: your profile PolyRating is your long-term career score (0–100k, Prestige, catch-up EF). The monthly season leaderboard is a separate race: it adds up this season’s wins and losses into one net total. Winning big does not erase losses — both count.

From the August 2026 season (starts 5 August 2026, 00:00 UTC), season boards use season points: same call-quality ingredients (magnitude, accuracy, market cap, horizon, boosts, Fame, engagement), but with E-Factor fixed at 1.0 for everyone. Your career rating does not shrink or inflate season points.

July 2026 and earlier seasons are unchanged — they keep summing the old EF-scaled PolyRating deltas. No mid-season rewrite of past seasons.

• PolyRating (overall / profile): career reputation · uses EF from your rating · floored at 0 · capped 0–100k · Prestige.
• Season points (Aug 2026+ boards): net of that month’s verifies · EF = 1 for all · uncapped · same call quality, fair race between veterans and new accounts.
• Why EF=1 on seasons: otherwise high-rated accounts earn fewer season points per win than fresh accounts — which rewards smurf accounts, not better predictions.

So from August onward, a Legend and a Beginner who make the same quality call earn the same season points. The Legend still gains less PolyRating from that call (catch-up EF on career only).

Open the leaderboard to compare seasons. Your profile PolyRating is unchanged by how we rank the monthly race.

• Start: Everyone begins at 0.
• Floor: Your score never goes below 0.
• Ceiling: The maximum is 100,000. As you get close, EF gets smaller so each new win moves the needle less.
• Prestige: When a verified signal brings your score to 100,000, Prestige runs automatically on the server: your score resets to 50,000 and your prestige count increases by one. It happens again each time you reach 100k after climbing back up. See the Prestige section below for the full explanation.

What it is: Prestige is an automatic reset when your PolyRating reaches 100,000. It’s designed so the scale doesn’t dead-end—you get to keep climbing and your profile shows how many times you’ve crossed that threshold.

How it works: When a verified signal pushes you to 100,000, we apply Prestige immediately: (1) your score is set to 50,000, (2) your prestige count (shown on your profile and leaderboard) goes up by one, and (3) you get a notification plus a celebration popup the next time you open the app. There’s no limit to how many times you can Prestige over a career. Your verified signals and history are unchanged; only the current score number resets.

Scoring after Prestige: Resetting to 50k does not restore early-game point gains. Prestiged accounts keep the reduced E-Factor used for high scores, so you still need strong predictions to climb again. That preserves the catch-up balance on the leaderboard.

Why it exists: It gives long-term goals beyond the first 100k, rewards consistency, and lets your prestige count represent how many times you’ve “maxed out.” Once you’ve Prestiged at least once, daily decay also applies when you’re above 80k or idle (see Daily decay), so staying active matters at the top.

Tiers are derived from your current PolyRating (e.g. 0–2k Beginner, 2k–10k Novice, …, 95k+ Legend). They appear next to your score on your profile and on the leaderboard.

If your score is above 80k or you’ve Prestiged at least once, we apply a small daily decay for each UTC day you don’t post a signal. The decay is based on an “average” signal (e.g. a 2% move over a 1-week horizon) and scales with how high your score is—higher score, slightly more decay per idle day.

Post any signal that day (UTC) and we skip decay for that day. So staying active keeps your rating from ticking down.

When your PolyRating is below a set threshold, you can post one Loss Free signal per day (UTC). If it’s correct, you earn 40% of the usual points. If it’s wrong, you lose no points. This lets newer or rebuilding users practice with less risk.

• Catch-up, not a soft reset. Low scores earn points quickly so new players can rise. High scores and Prestige keep E-Factor hard so veterans must keep outperforming — the same catch-up pattern as most competitive games.
• Direction first. Being right on LONG vs SHORT is the foundation. Magnitude and accuracy then determine how much you gain.
• Time matters. Longer horizons (creation to expiry) are rewarded with a higher time multiplier T, so bolder timeframes can pay off.
• Transparent and consistent. The formula (M, A, T, CL, EF) is fixed and matches our public simulator, so you can predict how a signal will affect your rating.
• Bounded and tiered. A 0–100k scale with tiers (Beginner → Legend) and Prestige keeps the system interpretable and gives long-term goals.
• Decay for engagement. Daily decay above 80k encourages top raters to stay active without punishing occasional breaks.
• Fame is separate from skill. Community engagement (Fame) and verified trading skill (Polyrating) use different formulas. Fame may add a small correct-signal bonus when enabled, but it never softens losses or replaces the core M, A, T, EF model.