Beat the book.
Back the model.
A per-league prediction model, fit walk-forward on years of results and calibrated against the closing line. Every call is settled in public — 87% of high-strike picks land — across 401 settled calls.
14-day free trial · no card · one tap on Telegram
18+ · Please gamble responsibly · BeGambleAware.org · past performance is no guarantee
Recent Hits
Settled calls · 81% landed · past performance is not a guarantee of future returns · 18+
Real picks. Public record. A measurable edge.
Tips on your phone, daily.
One tap on Telegram — no card, no email, no commitment. 14 days, free.
Every pick lands in your channel before kickoff — with the reasoning and the best available price.
Stake to your bankroll with the built-in Kelly calculator, at your own bookmaker. We publish tips — we never touch your money.
Validated in four leagues.
The competitions our Draw No Bet model is validated in — walk-forward tested across three seasons, every single call shown below as a result.
Each square is one Draw No Bet call — green won, orange lost; draws excluded (Draw No Bet returns the stake). All four seasons (2022–26) are walk-forward, out-of-sample backtests in the validated raw-probability band 0.55–0.80. The strike rate on each card is this season's live settled results. Past performance is not a guarantee of future returns.
World Cup 2026
The first 48-team World Cup runs across the United States, Canada and Mexico — 104 matchesover five and a half weeks. The single biggest run of football we'll ever price in one window.
International football rewards a different model than the leagues, so we built one for it. The instant the outright and match markets open, we read them across every major book and strip out the bookmaker margin — then fuse that sharp consensus with national-team Elo and a goals model trained on decades of international results. The output: a vig-free fair line on every nation and every fixture.
It runs on its own ledger, never folded into the club track record — every World Cup call shown in full and judged on its own results.
- Jun 11Opening matchEstadio Azteca · Mexico City
- Jun 11–27Group stage48 teams · 12 groups of 4
- Jun 28 – Jul 3Round of 32
- Jul 4–7Round of 16
- Jul 9–11Quarter-finals
- Jul 14–15Semi-finals
- Jul 19FinalMetLife Stadium · New Jersey
Every fixture priced, the margin stripped out.
We read the outright and match markets across every major book, strip the bookmaker margin, and blend in national-team Elo and a goals model — a vig-free fair line on every nation and fixture. Settled outcomes appear below; the live fair line is for subscribers.
78 / 95 match calls · 82%
Each row is a match we priced before kickoff — the score, our call, and whether we got it right. Nothing hidden.
- 2026-07-1019:00Z2026-07-10 · 19:00ZSpainvsBelgiumMatch: SpainGoals: Under 2.5✗2–12–1✓
- 2026-07-0920:00Z2026-07-09 · 20:00ZFrancevsMoroccoMatch: France or Draw(double chance)2–02–0✓
- 2026-07-0720:00Z2026-07-07 · 20:00ZSwitzerlandvsColombiaMatch: Draw or Colombia(double chance)0–00–0✓
- 2026-07-0716:01Z2026-07-07 · 16:01ZArgentinavsEgyptMatch: Argentina3–23–2✓
Every call, settled in public.
Settled picks only · past performance is not a guarantee of future returns
How often the model's most confident calls land. Short-priced favourites at average odds 1.21 — wins very often, but each win pays small.
HIGHER CONFIDENCE, higher strike.
The model is well-calibrated when its stated confidence matches the actual hit rate. Each bar should land above its confidence band; bars below 50% are coin-flip territory.
The 60% line marks the typical strike rate of a tipster service quoting short-odds favourites. Above it is the product promise.
Where the model is strongest.
- 1X282.4%match result161picks·131/28
- Draw No Bet72.0%home or away · void on draw1904picks·1028/400+452v
- Double Chance77.2%two of three outcomes2862picks·2180/645+1v
- O/U 2.573.5%total goals over / under34picks·25/9
- BTTS0.0%both teams to score3picks·0/1
| Group | Picks | Strike | W–L |
|---|---|---|---|
1X2 match result | 161 | 82.4% | 131/28 |
Draw No Bet home or away · void on draw | 1904 | 72.0% | 1028/400+452v |
Double Chance two of three outcomes | 2862 | 77.2% | 2180/645+1v |
O/U 2.5 total goals over / under | 34 | 73.5% | 25/9 |
BTTS both teams to score | 3 | 0.0% | 0/1 |
- Serie A77.4%I1710picks·483/141+58v
- La Liga76.6%SP1617picks·436/133+48v
- Primeira Liga79.5%P1520picks·376/97+47v
- Belgian Pro League62.9%B1205picks·100/59+46v
| Group | Picks | Strike | W–L |
|---|---|---|---|
Serie A I1 | 710 | 77.4% | 483/141+58v |
La Liga SP1 | 617 | 76.6% | 436/133+48v |
Primeira Liga P1 | 520 | 79.5% | 376/97+47v |
Belgian Pro League B1 | 205 | 62.9% | 100/59+46v |
The whole truth, even when it costs us.
Every bet has its own line: win more than 1 ÷ its odds and you profit. Win less — however high the strike looks — and you lose. The graph shows that rule in its simplest form; the table below holds our own surfaces to it, including the ones it condemns.
Here is the one rule we'd want every punter to know before trusting anyone's numbers — including ours. A bet's break-even strike rate is one divided by its decimal odds. Win more often than that and you profit; win less and you lose — no matter how spectacular the win-rate looks on a screenshot. Strike rate on its own is meaningless. Strike rate against its break-even line is everything.
“Ninety percent winners” is a real number you can hit — at odds so short it loses money anyway. It's the easiest figure in tipping to make look incredible, and some services are content to leave it at that. We hold ourselves to a harder standard: a strike rate only means something with its odds and its line beside it.
So here is ours — all of it. Our own model produces surfaces that look far better than the ones we actually sell. Below the graph, we grade them against the line and tell you why they never reach you — even though they'd make prettier adverts.
breakeven % = 1 ÷ decimal oddsClears the line with room to spare — the multi-season walk-forward record against Pinnacle's closing price across La Liga, Serie A, Liga Portugal and Belgian Pro. This one we send.
Looks unbeatable — but at these odds the line sits higher than the strike, so nine-in-ten landing is a small, grinding loss. Do the subtraction before we do: 1 ÷ odds. We show these on the dashboard, labelled strike, not yield, and never sell them as value.
High strike, still a loss: the model sits on the wrong side of the closing price here. Published openly, ungated, labelled breadth, not edge.
A 74% bet that prints, against a 90% bet that bleeds. Same model. Opposite verdict.
stack ~90% legs and the strike collapses to ~73%; on −EV legs the loss compounds. The one rosy acca number was a selection artefact. Binned.
can't even be validated — no honest historical odds — and it's the sharpest market in football. We don't sell what we can't prove.
Showing the bets we don't take costs us — the losing surfaces above would sell better dressed up as winners, and some in this market are happy to sell them that way. We publish them instead, with the maths that rules them out, because a record you can check is worth more than a screenshot you can't. That trade is the whole product. The slate stays small because the line is honest.
The graph's two bets are illustrative examples, not our data. The table's strike rates and yield are the settled multi-season walk-forward record (2022–26), measured against Pinnacle's closing line. Break-even = 1 ÷ average decimal odds; cushion = strike − break-even, shown only where a single representative price applies (coverage spans many prices, so it is judged on realised yield). Double Chance figures are read from the same record as the dashboard's strike-rate board. Past performance is not a guarantee of future results.
Beyond the majors.
Competitions we price and settle in public but don't sell — the Asian and American year-round leagues plus the Scandinavian summer, covering the European break. Same model; strike rate on par with the majors, but the yield isn't there — so breadth, not edge picks.
Settled walk-forward results. These leagues sit outside the headline yield — published for coverage breadth, not as edge picks.
Winners and losers, in full.
Nothing hidden, nothing cherry-picked.
High strike rate. Every high-confidence call (model probability ≥ 85%) across all 9leagues and all 5 markets. Lands at the rate the numbers say it should — short odds, almost always right. One pick per fixture, the model's most confident call on each.
- May 25, 2026BundesligalostSC Paderborn2:1WolfsburgDouble Chance — Draw or Away33/1001.33
- May 24, 2026Serie AlostMilan1:2CagliariMatch Odds (No Draw) — Home37/2001.18
- May 24, 2026Serie AwonNapoli1:0UdineseDouble Chance — Home or Draw2/151.13
- May 18, 2026Premier LeaguewonArsenal1:0BurnleyMatch Result — Home3/251.12
| Date▼ | Fixture· | Tip | Price· | Result |
|---|---|---|---|---|
| May 25, 2026 | SC Paderborn2:1Wolfsburg Bundesliga | Double Chance — Draw or Away | 33/100 | lost |
| May 24, 2026 | Milan1:2Cagliari Serie A | Match Odds (No Draw) — Home | 37/200 | lost |
| May 24, 2026 | Napoli1:0Udinese Serie A | Double Chance — Home or Draw | 2/15 | won |
| May 18, 2026 | Arsenal1:0Burnley Premier League | Match Result — Home | 3/25 | won |
For the curious.
Skip if you trust the numbers.
Every tip is generated from data alone — years of historical match results across each league, re-weighted daily so recent form counts more than ancient history. No tipsters, no hot takes, no team loyalties — only what the numbers support.
The model writes a probability for every market on every match — 1X2, Draw No Bet, Double Chance, Over / Under 2.5, Both Teams To Score. From that single forecast we publish two surfaces — for two different kinds of punter.
High strike rate — every high-confidence call (model probability ≥ 85%) across all 9 leagues and all 5 markets. Lands at the rate the numbers say it should: roughly 85% of these settled picks have won, on slim positive yield. Short odds, almost always right.
High yield — our walk-forward backtesting asked where do we measurably beat the closing line? The answer is Draw No Bet in La Liga, Serie A, Liga Portugal and Belgian Pro, at medium confidence, in the medium-odds band — a consistent positive yield across three backtested seasons. Smaller slate, harder calls, but the price is right.
The test that matters — every edge is measured against Pinnacle's closing line, the sharpest price in world football, where the pros bet. We don't beat soft averages; we beat the closing price — and we only publish the leagues and markets where, fit walk-forward on data strictly before kickoff, we provably do. We never train on the day we pick. That discipline is why the slate is small and the record holds.
And for every published pick we shop the market and surface the best price across the major books — so you back the sharpest number on offer, not whatever your one bookmaker happens to show.
The dashboard above shows every tip from this season — winners and losers, in full view. Nothing is hidden, nothing is cherry-picked. The strike rate is what it is.
Refer. They subscribe. You earn.
Your referral link lives in your account. Share it — and when the people you refer subscribe, you earn a recurring commission. No selling, no spin. The same edge that sharpens your bets pays you back.
- 01Share your private referral link — it's waiting in your account.
- 02They sign up and subscribe through it.
- 03You earn a recurring commission on what they pay.

Straight answers.
How do I get the picks?
Daily on Telegram — every pick posts to your channel before kickoff, with the reasoning and the best available price. Subscribers also see the full slate on the website ledger.
Do I place bets here?
No — Kestra isn't a bookmaker. We publish the model's picks; you place your own bets at your own bookmaker. We even surface the best price across the major books so you back the sharpest number on offer.
Can I try it before paying?
Yes — a 14-day free trial, one tap on Telegram, no card and no email. Watch the picks land for two weeks before you decide anything.
Can I cancel?
Anytime. It's a 30-day rolling subscription — no lock-in, no contract.
How do I know it's real and not cherry-picked?
Every single call settles in public — win or lose — on the ledger above. The model is fit walk-forward (never trained on the day it picks), and every edge is measured against Pinnacle's closing line, the sharpest price in football.
What happens on a losing run?
No model wins every week, and we never pretend otherwise — the losses are on the public record too. The edge is proven over thousands of graded bets, not a hot streak, and the built-in Kelly calculator helps you size every bet to your bankroll. Only ever stake what you can afford to lose.
Today's slate is published.
Start your free trial.
Subscribers get the day's high-confidence picks — fresh picks each morning, in the dashboard and via Telegram. Every settled outcome from past picks is already public — the rest is what you get when you sign in.
Start the free trial to unlock fixtures, selections and odds.