Live · updated Oct 4, 8:11 PM CT · 25 top picks today
Sports predictions, graded in public.
Put your money where your mouth is.
We're sports fans who build our own prediction algorithms for tennis, the NFL, college football, golf, the NBA, NHL, UFC, MLB, soccer and chess. Every
pick sits next to the live market price and gets graded after the game, wins and losses. Take it to the group chat,
talk your trash, and settle it for real with your crew, before the first serve and all game long.
The top of each of our boards: the spots where our algorithms see the most
value against the price. Our best effort to hand you an edge. Start here, add what you know about the players, make your call.
Next up: who our models favour and how strongly. Alongside, the calls that just settled, every one of them,
misses included. Tennis last 7 days: 1678–900 (65% right).
✓Caroline Werner 76%Caroline Werner vs Ekaterina Yashina · won
✓Jose Pereira 55%Arklon Huertas Del Pino Cordova vs Jose Pereira · won
✗Lois Boisson 59%Lois Boisson vs Leolia Jeanjean · Leolia Jeanjean won
✓Tatjana Maria 83%Madison Brengle vs Tatjana Maria · won
✗Gustavo Heide 78%Guido Ivan Justo vs Gustavo Heide · Guido Ivan Justo won
✗Abdullah Shelbayh 57%Abdullah Shelbayh vs Dylan Dietrich · Dylan Dietrich won
✓Joaquin Aguilar Cardozo 69%Joaquin Aguilar Cardozo vs Santiago De la Fuente · won
✗Francisco Rocha 70%Francisco Rocha vs Oleksii Krutykh · Oleksii Krutykh won
Win probability = our model's chance for the side it favours. Not a price and not a bet recommendation.
Sports are better with your crew
Settle it with your friends
Watching alone on a couch is the fake version. The real one is the group chat blowing up on a break of
serve, the buddy who swore his quarterback would show up, the co-worker you owe lunch. The athletes are part of your
crew's week. Make the call, back it, and let the game decide.
Pick'em trash talk. She'll take your pick'em money. Nicely. Challenge Greta →
1
Make the callStart from our pick or fade it. Your read on the
players is the part no algorithm has.
2
Send it to the chatTalk your trash before the game. Receipts
count here too.
3
Back it togetherFliff is a social sportsbook: play with your
friends and sweat it together, before the game and live. Or trade the price on Kalshi.
Fliff · social sportsbook · 18+
100% deposit match
Up to $100 in Fliff Cash on a $9.99+ first deposit. Sweat the picks with your crew.
30-day window to make your $9.99+ qualifying deposit
Play for fun, with what you can afford to lose. 21+ recommended; Fliff is 18+ and not in every state.
The boss fight
The market is the final boss
Every price on the board is the whole world's best guess with real money behind it: sharps, syndicates,
bots, your co-workers, the guy in your group chat. That's the boss. Here is where our algorithms stand against it in
every sport, graded in public. Left of the white line the market is winning; right of it, we are. Nobody has reached the
green flag yet. That's what makes it the real game, and why your intel, your smarts and your bots matter.
Tennis
-12.0 mB n=2,126
NFL
-32.1 mB n=25
College Football
-26.1 mB n=10,931
Golf
-5.2 mB n=2,348
NBA
-7.6 mB n=1,402
NHL
-4.7 mB n=1,468
MLB
-0.6 mB n=4,056
UFC
-42.6 mB n=156
Soccer
-4.2 mB n=6,495
Chess
-0.2 mB n=1,162
← the market is winningwe're winning →
White line = the market price. Green diamond = our bar. mB = thousandths of a Brier score, the standard accuracy grade for probabilities. Details on
the track record.
Live track record · tennis
How good are our numbers, really?
3,202matches graded, last 30 days
64.1%picked the winner
0.219Brier score (lower is better; a coin flip is 0.250)
Against the market 6 hours before start: behind the market by 8.7 mB (95% band -13.9 to -3.4), n=1,277.
When we say 70%, do they win 70%? Dots on the line are honest numbers.
Dots below it mean we were overconfident, and right now we are.Week by week vs the market. Brier score on the matches both of us priced. Higher is better on this chart.
our modelthe market, same matches
Every settled ATP, WTA and Challenger prediction from our live log. Updated every cycle; misses stay up.
These numbers grade our raw model. The board's headline win % leans on the market as well and is graded separately
from 2026-10-04 (4 graded so far; shown here at 30).
Full ledger · raw JSON.
Where we beat the market, and where we don’t
Our predictions match or beat the prediction market in some slices, not across the board. Out of 69 cuts we grade, we hold our own in 9; the market wins the other 60. On Tennis Challenger, Tennis ITF, College basketball, Golf, NBA, NHL, Soccer and UFC the market is clearly the better forecaster, and the band says so. Across the whole board, Chess and MLB are level with the price or better. The wins cluster in the same places: when we stay within 10 points of the price and when we call it close to a coin flip. When we break hard from the market, the market is usually right. So read our number as a second opinion that sharpens close calls, not a reason to bet against the price. None of these leads is proven yet: every band still crosses zero, and we will say so until one doesn’t.
Tennis ATP: we say under 60%. Ours 0.253 vs market 0.258, edge +4.8 mB (95% band -11.9 to +21.4), n=187.
our modelthe market, same games
Tennis ATP: differ from market <10pp. Ours 0.229 vs market 0.233, edge +3.7 mB (95% band -2.6 to +9.9), n=266.
our modelthe market, same games
Chess: all matches.replay Ours 0.066 vs market 0.066, edge -0.2 mB (95% band -2.4 to +2.0), n=1,162.
our modelthe market, same games
Chess: we say 70%+.replay Ours 0.064 vs market 0.064, edge -0.2 mB (95% band -2.3 to +1.9), n=1,142.
our modelthe market, same games
Chess: differ from market <10pp.replay Ours 0.060 vs market 0.061, edge +0.6 mB (95% band -0.7 to +1.9), n=1,060.
our modelthe market, same games
MLB: all matches.replay Ours 0.244 vs market 0.243, edge -0.6 mB (95% band -1.9 to +0.7), n=4,075.
our modelthe market, same games
MLB: we say under 60%.replay Ours 0.248 vs market 0.248, edge +0.5 mB (95% band -1.0 to +2.0), n=3,084.
our modelthe market, same games
MLB: differ from market <10pp.replay Ours 0.244 vs market 0.243, edge -0.4 mB (95% band -1.6 to +0.8), n=3,997.
our modelthe market, same games
UFC: differ from market <10pp.replay Ours 0.259 vs market 0.264, edge +5.3 mB (95% band -11.6 to +22.1), n=47.
Too few games per week to chart (a week needs 15+).
Every slice (72 rows, the losing ones included)
Group
Slice
n
Our Brier
Market
Edge mB
95% band
Picked winner (us / mkt)
Tennis ATP
all matches
395
0.231
0.229
-1.3
-11.6 to +9.0
58% / 60%
Tennis ATP
we say 70%+
101
0.186
0.182
-3.6
-19.4 to +12.3
75% / 75%
Tennis ATP
we say 60-70%
107
0.234
0.224
-9.8
-29.0 to +9.3
64% / 62%
Tennis ATP
we say under 60%
187
0.253
0.258
+4.8
-11.9 to +21.4
45% / 51%
Tennis ATP
differ from market <10pp
266
0.229
0.233
+3.7
-2.6 to +9.9
57% / 59%
Tennis ATP
differ from market 10pp+
129
0.234
0.222
-11.5
-40.2 to +17.2
60% / 62%
Tennis Challenger
all matches
600
0.251
0.226
-24.9
-36.4 to -13.3
56% / 64%
Tennis Challenger
we say 70%+
134
0.245
0.221
-24.6
-51.0 to +1.7
62% / 63%
Tennis Challenger
we say 60-70%
136
0.263
0.246
-16.9
-43.0 to +9.2
53% / 60%
Tennis Challenger
we say under 60%
330
0.248
0.220
-28.2
-42.7 to -13.8
55% / 65%
Tennis Challenger
differ from market <10pp
301
0.235
0.233
-2.1
-8.1 to +4.0
62% / 61%
Tennis Challenger
differ from market 10pp+
299
0.267
0.219
-47.8
-69.8 to -25.8
51% / 66%
Tennis WTA
all matches
584
0.202
0.196
-5.8
-12.7 to +1.0
67% / 68%
Tennis WTA
we say 70%+
221
0.131
0.130
-1.3
-9.8 to +7.1
84% / 83%
Tennis WTA
we say 60-70%
182
0.242
0.237
-5.3
-19.5 to +8.9
61% / 60%
Tennis WTA
we say under 60%
181
0.247
0.236
-11.9
-25.5 to +1.6
52% / 57%
Tennis WTA
differ from market <10pp
424
0.196
0.193
-2.5
-6.7 to +1.7
68% / 67%
Tennis WTA
differ from market 10pp+
160
0.218
0.203
-14.6
-37.1 to +7.9
63% / 70%
Tennis ITF
all matches
2,059
0.232
0.180
-52.2
-60.2 to -44.3
61% / 73%
Tennis ITF
we say 70%+
328
0.186
0.152
-34.1
-51.0 to -17.1
75% / 76%
Tennis ITF
we say 60-70%
681
0.234
0.178
-55.7
-69.7 to -41.7
63% / 74%
Tennis ITF
we say under 60%
1,050
0.245
0.189
-55.6
-67.1 to -44.1
56% / 71%
Tennis ITF
differ from market <10pp
535
0.223
0.218
-4.6
-9.1 to -0.1
63% / 63%
Tennis ITF
differ from market 10pp+
1,524
0.235
0.166
-68.9
-79.4 to -58.4
60% / 76%
College basketballreplay
all matches
5,239
0.172
0.168
-4.2
-5.7 to -2.6
73% / 73%
College basketballreplay
we say 70%+
2,964
0.119
0.116
-3.0
-4.6 to -1.4
85% / 85%
College basketballreplay
we say 60-70%
1,100
0.235
0.228
-7.7
-12.0 to -3.5
62% / 62%
College basketballreplay
we say under 60%
1,175
0.247
0.244
-3.8
-7.8 to +0.3
55% / 54%
College basketballreplay
differ from market <10pp
4,683
0.165
0.163
-2.1
-3.2 to -1.0
74% / 74%
College basketballreplay
differ from market 10pp+
556
0.230
0.209
-21.5
-32.9 to -10.2
64% / 66%
Chessreplay
all matches
1,162
0.066
0.066
-0.2
-2.4 to +2.0
92% / 92%
Chessreplay
we say 70%+
1,142
0.064
0.064
-0.2
-2.3 to +1.9
93% / 93%
Chessreplay
we say 60-70%
16
0.170
0.156
-14.2
n too small
81% / 81%
Chessreplay
we say under 60%
4
0.259
0.312
+53.0
n too small
25% / 50%
Chessreplay
differ from market <10pp
1,060
0.060
0.061
+0.6
-0.7 to +1.9
93% / 93%
Chessreplay
differ from market 10pp+
102
0.126
0.117
-9.1
-29.8 to +11.6
85% / 84%
Golfreplay
all matches
5,402
0.121
0.120
-1.3
-2.1 to -0.5
81% / 81%
Golfreplay
we say 70%+
3,696
0.074
0.073
-0.8
-1.5 to -0.2
91% / 91%
Golfreplay
we say 60-70%
728
0.217
0.215
-1.3
-4.5 to +1.9
66% / 66%
Golfreplay
we say under 60%
978
0.231
0.228
-3.0
-5.5 to -0.6
55% / 58%
Golfreplay
differ from market <10pp
5,312
0.120
0.119
-1.0
-1.7 to -0.4
81% / 82%
Golfreplay
differ from market 10pp+
90
0.213
0.195
-17.8
-44.0 to +8.4
66% / 68%
MLBreplay
all matches
4,075
0.244
0.243
-0.6
-1.9 to +0.7
56% / 55%
MLBreplay
we say 70%+
62
0.140
0.138
-2.1
-9.7 to +5.5
85% / 85%
MLBreplay
we say 60-70%
929
0.238
0.233
-4.2
-6.9 to -1.6
61% / 61%
MLBreplay
we say under 60%
3,084
0.248
0.248
+0.5
-1.0 to +2.0
54% / 53%
MLBreplay
differ from market <10pp
3,997
0.244
0.243
-0.4
-1.6 to +0.8
56% / 55%
MLBreplay
differ from market 10pp+
78
0.254
0.241
-12.3
-37.8 to +13.1
60% / 51%
NBAreplay
all matches
1,402
0.208
0.201
-7.6
-12.1 to -3.0
68% / 67%
NBAreplay
we say 70%+
533
0.159
0.150
-8.6
-14.4 to -2.9
80% / 80%
NBAreplay
we say 60-70%
412
0.233
0.222
-10.1
-18.7 to -1.5
63% / 62%
NBAreplay
we say under 60%
457
0.243
0.239
-4.0
-13.6 to +5.5
58% / 58%
NBAreplay
differ from market <10pp
1,040
0.206
0.203
-3.1
-5.9 to -0.2
68% / 67%
NBAreplay
differ from market 10pp+
362
0.215
0.194
-20.5
-36.0 to -4.9
69% / 67%
NHLreplay
all matches
1,475
0.248
0.243
-4.7
-7.1 to -2.4
54% / 56%
NHLreplay
we say 70%+
46
0.244
0.232
-12.3
-24.9 to +0.2
63% / 63%
NHLreplay
we say 60-70%
372
0.233
0.230
-3.4
-7.8 to +1.0
63% / 63%
NHLreplay
we say under 60%
1,057
0.253
0.248
-4.9
-7.7 to -2.0
50% / 53%
NHLreplay
differ from market <10pp
1,423
0.248
0.244
-4.0
-6.2 to -1.9
53% / 56%
NHLreplay
differ from market 10pp+
52
0.240
0.216
-24.1
-56.0 to +7.7
63% / 65%
Soccerreplay
all matches
9,951
0.202
0.198
-3.6
-4.7 to -2.6
70% / 70%
Soccerreplay
we say 70%+
5,490
0.177
0.175
-2.2
-3.1 to -1.2
76% / 76%
Soccerreplay
we say 60-70%
2,322
0.219
0.214
-4.8
-7.3 to -2.3
67% / 67%
Soccerreplay
we say under 60%
2,139
0.246
0.240
-6.1
-9.2 to -2.9
56% / 57%
Soccerreplay
differ from market <10pp
9,066
0.198
0.196
-2.0
-2.8 to -1.3
71% / 71%
Soccerreplay
differ from market 10pp+
885
0.236
0.216
-19.8
-28.4 to -11.2
62% / 65%
UFCreplay
all matches
156
0.240
0.198
-42.6
-68.4 to -16.9
53% / 70%
UFCreplay
we say 70%+
2
0.298
0.266
-32.5
n too small
50% / 50%
UFCreplay
we say 60-70%
33
0.234
0.198
-35.9
-91.1 to +19.3
64% / 70%
UFCreplay
we say under 60%
121
0.241
0.197
-44.6
-74.4 to -14.9
50% / 70%
UFCreplay
differ from market <10pp
47
0.259
0.264
+5.3
-11.6 to +22.1
45% / 57%
UFCreplay
differ from market 10pp+
109
0.232
0.169
-63.3
-98.8 to -27.8
56% / 75%
Market = Kalshi price 6 hours before the scheduled start; Brier score, lower is better; edge = market Brier minus ours, in thousandths. "We say" = our favourite's probability; "differ" = gap between us and that T-6h price. Level or ahead = edge ≥ −1, n ≥ 30, and a band not wholly below zero; a rule fixed before this page was built. Tennis is graded on forecasts logged before each match. Replay marks a sport graded by re-running its model forward through past games against the same T-6h price: no future data, but not a forecast we published at the time. Raw JSON.
Free · no spam · leave any time
Join the IPS Crew
We're a crew of sports nerds, traders and bot builders taking on the betting market in public. Join and you get:
The morning drop — our top picks across every sport, before the market moves.
Friday receipts — what we hit, what we missed, and what the market knew that we didn't.
First in — crew members get the first invites to the IPS Discord and the crew pick'em league when they open.
Small samples swing hard: a +9% ROI on 28 bets is not proof of anything yet, and we say so.
Paper results are simulated (CFTC Rule 4.41) — see the disclosures below.
FLIFF SOCIAL SPORTSBOOK · 18+
100% Match up to $100 in Fliff Cash · $9.99+ first deposit
Sweepstakes odds for ATP & WTA tennis in 32 US states. Redeemable for real cash prizes.
Referral link. Not available to US users. We compare Polymarket's prices with Kalshi's; it does not influence our picks.
For bot builders
Plug our numbers into your bot
Every number on this page is a free JSON feed. No key, CORS on. Point Claude, ChatGPT, Cursor or a
Python script at it, paper-trade first, and put your agent on the board.
/predictions.jsonTennis — every ATP/WTA match, quarter-Kelly, Kalshi tickers
Every tout on the internet is up 74 units. Nobody shows the losses.
We started IPS to do the opposite: build real models, put our number next to the market's number, and grade every
single call in public — so you can see exactly how good (or not) we are, and get sharper watching us work.
Show the receipts.Every pick is logged before the game. Nothing gets deleted.
The market is the boss.Beating your buddy is easy. Beating the price is the game.
Build, don't just tail.Our feeds are free. Fork the bot. Make it better than ours.
Want a bigger role? We're looking for volunteer analysts (one sport, one note a week) and
moderators for the coming Discord. Raise your hand
21+ recommended · informational research only. Model probabilities are estimates, not betting or
financial advice. Prediction markets and sweepstakes involve real risk of loss. Need help? 1-800-GAMBLER.
Regulatory Disclosures, Safe-Harbor & Terms of Use
1. Analytical & Research Use Only: iPredictSport is a quantitative research platform publishing statistical estimates for educational, informational, and analytical purposes only. iPredictSport is not an investment adviser, commodity trading advisor (CTA), registered broker-dealer, or sports wagering operator. Nothing contained on this site, in our API, or in our newsletters constitutes financial advice, investment recommendations, or sports betting advice.
2. Hypothetical & Simulated Performance (CFTC Rule 4.41): Performance figures for our autonomous paper agent and proposed bet track records are simulated, hypothetical, and based on paper audits. Unlike an actual performance record, simulated results do not represent actual trading, may not reflect market liquidity, exchange slippage, or trading bans, and cannot account for psychological risk factors. No representation is made that any user will or is likely to achieve profits or losses similar to those shown. Past performance does not guarantee future results.
3. Market Risk & Capital Loss: Trading binary contracts on prediction markets (such as Kalshi or Polymarket) or placing sports wagers involves substantial financial risk, including the possible loss of 100% of staked capital. Odds and spreads fluctuate continuously. Users assume sole responsibility for their own trading decisions and bankroll sizing.
4. Responsible Gaming & Age Requirements: Fliff sweepstakes is available to legal residents aged 18+ in eligible states. Financial prediction exchanges (including Kalshi) require 18+, and commercial sportsbooks require 21+, subject to jurisdiction. If you or someone you know has a gambling problem and wants help, confidential crisis counseling and referral services can be accessed 24/7 by calling 1-800-GAMBLER (1-800-426-2537) or visiting ncpgambling.org.
5. Affiliate & Compensation Disclosure: iPredictSport may receive affiliate commissions, referral fees, or cost-per-acquisition (CPA) compensation from links to partners (including Fliff, Kalshi, and Polymarket) at no additional cost to you. For Fliff, attribution requires entry of promo code IPREDICT or IPSLIVE in the mobile app. Fliff is a sweepstakes sports product for adults 18+ and is void where prohibited by law (18 excluded states). This commercial compensation does not alter our objective quantitative probabilities, model outputs, or editorial gates.
