[{"data":1,"prerenderedAt":416},["ShallowReactive",2],{"post-calibration-score-explained-en":3,"related-posts-calibration-score-explained-en":377,"related-categories-calibration-score-explained-en":398},{"id":4,"title":5,"author":6,"authorImageUrl":7,"authorUrl":8,"body":9,"categorySlug":340,"categoryText":341,"dateModified":342,"datePublished":342,"description":343,"extension":344,"faqs":345,"image":42,"imageAlt":364,"language":365,"meta":366,"navigation":367,"path":368,"seo":369,"sitemap":370,"stem":371,"summary":372,"tags":373,"__hash__":376},"blog\u002Fblog\u002Fcalibration-score-explained.md","Calibration Score Explained: Edge, Hit Rate and the Chart","Bruma","\u002Fimages\u002Fauthors\u002Fbruma.jpg","https:\u002F\u002Fbrumaombra.com",{"type":10,"value":11,"toc":326},"minimark",[12,16,28,31,36,43,56,64,71,74,76,80,87,92,104,107,109,113,119,123,126,132,139,142,149,152,154,158,164,167,171,174,177,183,186,188,192,195,199,202,205,208,211,213,217,220,224,227,234,236,240,243,246,250,253,256,258,262,265,273,281,284,286,290,293,296,306,308,312,315,318],[13,14,15],"p",{},"Some friends are right a lot, and some friends are right about the things that matter. Those aren't the same people. The friend who bets that the sun will rise tomorrow has a perfect record and has told you nothing. The friend who said, back when everyone laughed, that the quiet couple would be the first to get engaged? That's the one whose next call you want to hear.",[13,17,18,19,23,24,27],{},"The calibration section on your KrowdCall profile and P&L tries to tell those two kinds of friends apart. It shows a verdict, two numbers and a chart full of dots. This post explains each part: which of your markets count, how ",[20,21,22],"strong",{},"edge"," and ",[20,25,26],{},"hit rate"," are calculated, how to read the chart, and why the score is careful with you for your first 10 markets.",[29,30],"hr",{},[32,33,35],"h2",{"id":34},"what-does-the-calibration-score-measure","What Does the Calibration Score Measure?",[13,37,38],{},[39,40],"img",{"alt":41,"src":42},"A young woman smiles at her phone while thinking of a calibration chart, with green dots above a dashed diagonal line and one red dot below it","\u002Fimages\u002Fblog\u002Fcalibration-score-explained.png",[13,44,45,46,49,50,55],{},"In forecasting, ",[20,47,48],{},"calibration"," usually means that your probabilities match reality over time: of everything you call 70% likely, about 70% should happen. Our guide on ",[51,52,54],"nuxt-link-locale",{"to":53},"\u002Fblog\u002Fhow-to-make-better-predictions","how to make better predictions"," covers that classic version.",[13,57,58,59,63],{},"KrowdCall can't read your mind, though. It only sees what you do: which side you bought and what you paid. And on KrowdCall, the price is the probability. A YES share at 0.30 Coins means the market gives YES a 30% chance. If you want a refresher on that, see ",[51,60,62],{"to":61},"\u002Fblog\u002Fprediction-market-odds-explained","how prediction market odds work",".",[13,65,66,67,70],{},"So the score asks a slightly different question: ",[20,68,69],{},"when you bought at a price, how often did that side actually win?"," If you keep buying things at 30% and they happen half the time, the crowd was underpricing them and you saw it. If you buy at 70% and they happen half the time, you were paying for confidence that wasn't there.",[13,72,73],{},"That makes the market price your benchmark. The score doesn't compare you to perfection. It compares you to the crowd's number at the moment you made your call.",[29,75],{},[32,77,79],{"id":78},"which-markets-count","Which Markets Count?",[13,81,82,83,86],{},"Before any maths, it helps to know exactly what goes in. KrowdCall turns each market you predicted on into one ",[20,84,85],{},"call",", using these rules:",[88,89],"blog-list",{":items":90,"variant":91},"[\"Only resolved markets count. Open markets wait until they're settled, and canceled markets never count, since nobody was right or wrong\",\"You need to have bought shares in the market. Being invited or adding it to favourites isn't enough\",\"Your call is the side you held more shares of when the market resolved. If you had sold everything by then, it's the side you bought more shares of\",\"The price of your call is the average price you paid for that side, across all your buys\",\"Each market counts once, whether you bought 5 shares or 500\"]","checkmark",[13,93,94,95,98,99,103],{},"There's one more rule, about privacy. Your ",[20,96,97],{},"public profile"," only uses public markets, so friends' private markets never leak into what strangers can see. Your own ",[51,100,102],{"to":101},"\u002Fprivate\u002Fpnl","P&L page"," also counts your private markets, since only you see it. That's why the two can show different numbers, and both are correct.",[13,105,106],{},"Selling deserves a quick note too. If you sold some shares before the end, the call is still about the side you leaned toward, and the price is still what you paid to get in. Selling doesn't rewrite your entry price. It just means you might have cashed out before the result, which shows up in your profit, not in this score.",[29,108],{},[32,110,112],{"id":111},"how-is-edge-calculated","How Is Edge Calculated?",[13,114,115,118],{},[20,116,117],{},"Edge"," is the headline number. Each call gets a small score:",[88,120],{":items":121,"variant":122},"[\"If your side came true, the call scores 1 minus the price you paid\",\"If it didn't, the call scores 0 minus the price you paid\",\"Edge is the average of those scores across all your calls, multiplied by 100 and shown in points\"]","numbered",[13,124,125],{},"Here's Mia's record after four resolved markets:",[127,128],"blog-table",{":headers":129,":rows":130,":highlightCol":131},"[\"Market\",\"Mia's call\",\"Average price paid\",\"Result\",\"Score\"]","[[\"Will the bus be late on Friday?\",\"YES\",\"0.30\",\"YES\",\"+0.70\"],[\"Will Ana finish the marathon under 4 hours?\",\"NO\",\"0.60\",\"NO\",\"+0.40\"],[\"Will it snow before Christmas?\",\"YES\",\"0.80\",\"NO\",\"-0.80\"],[\"Will the group trip happen in May?\",\"YES\",\"0.50\",\"YES\",\"+0.50\"]]","4",[13,133,134,135,138],{},"The scores add up to 0.80. Divided by four markets, that's 0.20, so Mia's edge is ",[20,136,137],{},"+20 pts",". On average, the things she bought came true 20 points more often than their prices suggested.",[13,140,141],{},"Wait, why subtract the price at all? Because that's what the price already promised. A share bought at 0.80 should win about 80% of the time if the crowd is right. Winning it only gets you 0.20 of credit, and losing it costs you 0.80. Buying at 0.30 is the opposite: a small loss if you're wrong, a big reward if you're right.",[13,143,144,145,148],{},"If that sounds like money, it's no accident. A winning share pays 1 Coin, so a call bought at 0.30 that comes true earns 0.70 Coins per share. Edge is basically ",[20,146,147],{},"your average profit per share",", except every market counts the same, no matter how much you put into it.",[13,150,151],{},"That last part matters. Your P&L cares about how much you bet. Edge doesn't. One huge losing bet can sink your P&L while your edge stays healthy, and a lucky big win can flatter your P&L while your edge says your calls were mostly average. Neither number is wrong. They answer different questions: P&L asks how your Coins did, and edge asks how good your reads were.",[29,153],{},[32,155,157],{"id":156},"why-isnt-hit-rate-enough","Why Isn't Hit Rate Enough?",[13,159,160,163],{},[20,161,162],{},"Hit rate"," is the simpler number: how many of your calls came true, out of how many you made. Mia got 3 of 4 right, so her hit rate is 75%.",[13,165,166],{},"It's easy to understand, and that's exactly the trap. Look at two players after 10 resolved markets each:",[127,168],{":headers":169,":rows":170},"[\"Player\",\"Style\",\"Price paid\",\"Right\",\"Hit rate\",\"Edge\"]","[[\"Sam\",\"Always buys the favourite\",\"0.90 every time\",\"9 of 10\",\"90%\",\"0 pts\"],[\"Leo\",\"Hunts for long shots\",\"0.20 every time\",\"4 of 10\",\"40%\",\"+20 pts\"]]",[13,172,173],{},"Sam looks like a genius on hit rate. But he paid 0.90 for things the crowd already rated at 90%, and they happened 90% of the time. He agreed with everyone, and everyone was right. His nine wins earned 0.10 each, his one loss cost 0.90, and it all washes out to zero.",[13,175,176],{},"Leo is wrong more often than right. Yet he keeps finding things the crowd rates at 20% that actually happen 40% of the time. His four wins earn 0.80 each, and his six losses cost 0.20 each. That's 3.20 minus 1.20, or 2.00 over 10 markets: +20 pts.",[13,178,179],{},[39,180],{"alt":181,"src":182},"Two friends look at their own phones: above one floats a YES 90% card with a check mark, above the other a YES 20% card with a star","\u002Fimages\u002Fblog\u002Fhit-rate-vs-edge.png",[13,184,185],{},"That's why the card shows both numbers side by side. Hit rate tells you how often you're right. Edge tells you whether being right was worth anything, given what the crowd already knew. If you only look at one, look at edge.",[29,187],{},[32,189,191],{"id":190},"how-do-you-read-the-calibration-chart","How Do You Read the Calibration Chart?",[13,193,194],{},"The chart turns your calls into a picture. Here's what each part means:",[127,196],{":headers":197,":rows":198},"[\"Part of the chart\",\"What it shows\"]","[[\"Across (price paid)\",\"The average price you paid for your calls, from 0% to 100%\"],[\"Up (came true)\",\"How often those calls actually came true\"],[\"Dashed diagonal line\",\"The fair line: calls that came true exactly as often as their price said\"],[\"A dot\",\"Your calls in one price range (0-10%, 10-20% and so on), placed at their average price and their hit rate\"],[\"Dot size\",\"How many calls are in that range: bigger dots carry more weight\"],[\"Dot color\",\"Green when those calls beat their price, red when they fell short\"]]",[13,200,201],{},"So a green dot sitting well above the line around 30% means: when you bought things at around 30%, they came true much more often than that. That's the zone where you spot what others underestimate. A red dot below the line around 80% means the opposite: your confident buys came true less often than you paid for.",[13,203,204],{},"A few shapes come up often:",[88,206],{":items":207,"variant":91},"[\"Dots hugging the line: you mostly agree with the crowd's prices, and the crowd is doing fine. Your edge will sit near zero\",\"Green dots on the left, near the line on the right: you're good at spotting underrated long shots, and your favourites are priced about right\",\"Red dots on the right: you tend to buy favourites that are overpriced. Maybe the group is overconfident, or you buy after the price has already run up\",\"One big dot, nothing else: you always buy in the same price range. The chart can only judge the prices you actually trade at\"]",[13,209,210],{},"On a computer, hovering over a dot shows its exact numbers: the price range, how many calls came true, and out of how many.",[29,212],{},[32,214,216],{"id":215},"what-does-the-verdict-mean","What Does the Verdict Mean?",[13,218,219],{},"Above the numbers, the card sums everything up in one sentence. It follows a simple rule:",[127,221],{":headers":222,":rows":223},"[\"Edge\",\"Verdict\",\"Tone\"]","[[\"+5 pts or more\",\"Ahead of the crowd\",\"Green\"],[\"Between -5 and +5 pts\",\"About even with the crowd\",\"Neutral\"],[\"-5 pts or less\",\"The crowd has been right more often\",\"Red\"]]",[13,225,226],{},"The 5-point band in the middle is deliberate. An edge of +2 or -3 pts is well within normal noise, and it would be silly to call one person a forecaster and the other a fool over it.",[13,228,229,230,233],{},"There's one more rule, and it's the subject of the next section: ",[20,231,232],{},"with fewer than 10 resolved markets, being ahead only counts as \"so far\"",". The card says you're ahead of the crowd so far, but that it's early days, and it doesn't mention the points or praise a talent you haven't had time to show yet.",[29,235],{},[32,237,239],{"id":238},"why-are-your-first-10-markets-early-days","Why Are Your First 10 Markets Early Days?",[13,241,242],{},"Okay, this is the part that keeps the score honest. A single call can swing your edge enormously. Buy YES at 0.51 once, get it right, and your edge is +49 pts. Get it wrong and it's -51 pts. Neither number says anything about you. It's a coin flip with decimals.",[13,244,245],{},"How much does luck move things? A rough rule: for calls around 50%, the noise in your edge is about 50 points divided by the square root of how many markets you have.",[127,247],{":headers":248,":rows":249},"[\"Resolved markets\",\"Typical swing from luck alone\"]","[[\"1\",\"About ±50 pts\"],[\"4\",\"About ±25 pts\"],[\"10\",\"About ±16 pts\"],[\"25\",\"About ±10 pts\"],[\"100\",\"About ±5 pts\"]]",[13,251,252],{},"That's a rule of thumb, not an exact law, and calls at very high or very low prices are a bit less noisy. But it shows the shape of the problem. Even at 10 markets, an edge of +10 pts could still be mostly luck. At 100 markets, it probably isn't.",[13,254,255],{},"So why 10, and not 50? Because the point of the score is to be fun and useful among friends, and a friend group might resolve a few markets a month. Ten is where the numbers stop being pure noise and start suggesting a direction. The note under the chart says it plainly: it gets reliable after about 10. Treat that as the moment the score becomes worth discussing, not the moment it becomes the truth.",[29,257],{},[32,259,261],{"id":260},"why-measure-against-the-crowd-at-all","Why Measure Against the Crowd at All?",[13,263,264],{},"Here's a fair objection: isn't the market price just everyone's guess? Why should beating it mean anything?",[13,266,267,268,272],{},"Because that guess is usually decent. A market price combines what everyone in the group knows and is willing to put Coins behind, which is the whole idea of the ",[51,269,271],{"to":270},"\u002Fblog\u002Fwisdom-of-the-crowd","wisdom of the crowd",". Beating a price set by people who know the topic is hard. If you manage it consistently, you probably know something they don't, or you think about it more clearly.",[13,274,275,276,280],{},"There's a subtle catch, though. Your own purchase moves the price. On KrowdCall, prices come from an automated market maker, explained in ",[51,277,279],{"to":278},"\u002Fblog\u002Flmsr-explained","LMSR for beginners",", and each share you buy costs a little more than the last. So the average price you paid already includes the push you gave the market. Big purchases in small markets get slightly worse prices, and that's fair: the score judges the price you actually paid, not a better one you wish you'd had.",[13,282,283],{},"And small groups have their own risk. If everyone shares the same blind spot, the crowd's price can be wrong in a predictable way, and beating it is easier. That's still real skill, but it's skill against your friends, not against the world.",[29,285],{},[32,287,289],{"id":288},"how-can-you-improve-your-score","How Can You Improve Your Score?",[13,291,292],{},"There's no trick that guarantees a better edge, and gaming it doesn't make sense either: the only way to raise it is to make better calls. These habits usually help:",[88,294],{":items":295,"variant":91},"[\"Only buy when you think the price is wrong. Agreeing with the crowd isn't a mistake, but it earns nothing on this score\",\"Ask what the price is missing. Local knowledge, like how often your friend actually shows up on time, is where small groups misprice things\",\"Be wary of expensive favourites. A 90% share wins small and loses big, so you need to be right more than 90% of the time for it to pay\",\"Look at your red dots. If they cluster in one price range, that's the habit to check first\",\"Keep predicting. More resolved markets make the score more honest, whatever it ends up saying\"]",[13,297,298,299,303,304,63],{},"For a deeper look at deciding when and how much to buy, our ",[51,300,302],{"to":301},"\u002Fblog\u002Fprediction-market-strategy","prediction market strategy guide"," turns these ideas into practical rules. And the habits behind good forecasts, like base rates and updating, are in ",[51,305,54],{"to":53},[29,307],{},[32,309,311],{"id":310},"what-the-score-can-and-cant-tell-you","What the Score Can and Can't Tell You",[13,313,314],{},"Back to the two friends from the start: the one who's always right about obvious things, and the one who's right when it's surprising. The calibration score is built to tell them apart. Hit rate rewards the first friend. Edge rewards the second, because it only gives credit for being right more often than the crowd expected.",[13,316,317],{},"It still has limits. It can't see the reasons behind a call, it can't tell skill from luck over a handful of markets, and it only judges the price ranges you actually trade in. A great edge over 8 markets is a promising start, not a reputation.",[13,319,320,321,325],{},"But over time, it answers a question that's surprisingly hard to settle in a group chat: when you disagree with everyone, how often are you right? If you'd like to start building your own answer, ",[51,322,324],{"to":323},"\u002Fmarkets","find a market"," where you think the crowd has it wrong.",{"title":327,"searchDepth":328,"depth":328,"links":329},"",2,[330,331,332,333,334,335,336,337,338,339],{"id":34,"depth":328,"text":35},{"id":78,"depth":328,"text":79},{"id":111,"depth":328,"text":112},{"id":156,"depth":328,"text":157},{"id":190,"depth":328,"text":191},{"id":215,"depth":328,"text":216},{"id":238,"depth":328,"text":239},{"id":260,"depth":328,"text":261},{"id":288,"depth":328,"text":289},{"id":310,"depth":328,"text":311},"explainers","Explainers","2026-11-13T00:00:00Z","How the KrowdCall calibration score works: which markets count, how edge and hit rate are calculated, how to read the chart, and why your first 10 markets are early days.","md",[346,349,352,355,358,361],{"question":347,"answer":348},"What is the calibration score on KrowdCall?","It is a section of your profile and of your P&L that compares the price you paid for each of your calls with whether it came true. It shows two numbers, edge and hit rate, a chart of your calls grouped by price, and a one-sentence verdict on whether you are ahead of the crowd, about even or behind it.",{"question":350,"answer":351},"How is edge calculated?","For every resolved market you predicted on, KrowdCall takes the side you ended on and the average price you paid for it. A call that came true scores 1 minus that price, one that didn't scores minus the price. Edge is the average of those scores across your markets, shown in points, so +12 pts means your calls came true 12 points more often than their prices said.",{"question":353,"answer":354},"Why is my edge low when my hit rate is high?","Because hit rate ignores what you paid. If you mostly buy favourites at 85% or 90%, you should be right most of the time just to break even. Edge subtracts the price, so a 90% hit rate on 90% calls is an edge of about zero: you agreed with the crowd, and the crowd was right.",{"question":356,"answer":357},"Which markets count toward my calibration?","Only resolved markets where you bought shares. Open and canceled markets don't count. Your public profile only uses public markets, while your own P&L also includes your private markets, so the two can show different numbers.",{"question":359,"answer":360},"Why does KrowdCall say it's early days?","With few resolved markets, luck dominates. One lucky call at 50% is worth 50 points of edge on its own. Until you have about 10 resolved markets, the verdict says you're ahead of the crowd so far rather than praising a skill, and the card shows a note about it.",{"question":362,"answer":363},"Is the calibration score the same as a Brier score?","No. A Brier score grades probabilities you state yourself. KrowdCall's score uses the prices you chose to pay, so it measures how your calls did against the crowd's price at the time, not how accurate your own numbers were. Both reward the same habit: buying only when you think the price is wrong.","An illustration of a young woman on a couch smiling at her phone, thinking of a chart where green dots sit above a dashed diagonal line, next to a card that reads +12 pts.","en",{},true,"\u002Fblog\u002Fcalibration-score-explained",{"title":5,"description":343},{"loc":368},"blog\u002Fcalibration-score-explained","The KrowdCall calibration score compares the price you paid for your calls with how often they came true. Edge is the average gap between the two, in points: positive means you bought things the crowd underpriced. Hit rate is the share of calls you got right. The chart groups your calls by price, and the score needs about 10 resolved markets before it says much.",[374,375],"Prediction Markets","Strategy","TbDFEFKmTjpHVc58yq_qPp805fiOUDABK-leKiUoMPM",[378,384,388,393],{"title":379,"description":380,"image":381,"categoryText":382,"path":383},"Wedding Prediction Game: 12 Kind Guest Questions","Run a wedding prediction game guests love: 12 kind yes\u002Fno questions timed to the ceremony, speeches and dance floor, plus how to get the couple on board.","\u002Fimages\u002Fblog\u002Fwedding-prediction-game.png","Tips","\u002Fblog\u002Fwedding-prediction-game",{"title":385,"description":386,"image":387,"categoryText":341,"path":270},"Wisdom of the Crowd: Why Groups Often Predict Better","Wisdom of the crowd explained: Galton's ox, the four conditions a crowd needs, how crowds fail, and how to use your friend group to make better forecasts.","\u002Fimages\u002Fblog\u002Fwisdom-of-the-crowd.png",{"title":389,"description":390,"image":391,"categoryText":382,"path":392},"Office Prediction Game: A No-Money Team Guide","Run an office prediction game with no money: HR-friendly rules, questions you can check and a simple private setup for your team with free virtual Coins.","\u002Fimages\u002Fblog\u002Foffice-prediction-pool.png","\u002Fblog\u002Foffice-prediction-pool",{"title":394,"description":395,"image":396,"categoryText":382,"path":397},"Election Night Party: A Friendly Prediction Game","Host an election night party with a kind, non-partisan prediction game for any country: questions on timing, turnout and counting, plus ground rules.","\u002Fimages\u002Fblog\u002Felection-night-with-friends.png","\u002Fblog\u002Felection-night-with-friends",[399,405,411],{"name":400,"slug":401,"path":402,"count":403,"image":404},"Guides","guides","\u002Fblog\u002Fcategories\u002Fguides",14,"\u002Fimages\u002Fblog\u002Fgoal-accountability-with-friends.png",{"name":406,"slug":407,"path":408,"count":409,"image":410},"Comparisons","comparisons","\u002Fblog\u002Fcategories\u002Fcomparisons",4,"\u002Fimages\u002Fblog\u002Fgroup-chat-polls-vs-predictions.png",{"name":382,"slug":412,"path":413,"count":414,"image":415},"tips","\u002Fblog\u002Fcategories\u002Ftips",26,"\u002Fimages\u002Fblog\u002Fprediction-game-ideas-for-friends.png",1795196975523]