Back to Blog
UpdatesComplete GuideJune 23, 202610 min read

We Lost $89,814 in Week 1 of the World Cup. Here Is What Happened Next.

OddsFlow lost $89,814 in the first week of World Cup 2026 with a 36% win rate. After model recalibration, we turned it around: +$159,195 in 5 days with a 64% win rate. We publish every loss and every win. This is our transparency report.

OddsFlow

OddsFlow AI Research

OddsFlow Team

June 23, 2026
We Lost $89,814 in Week 1 of the World Cup. Here Is What Happened Next.

We Lost $89,814 in Week 1 of the World Cup. Here Is What Happened Next.

Most prediction platforms would delete this data. We published it.

On June 11, 2026, the FIFA World Cup kicked off in North America. OddsFlow's AI model — the same Dixon-Coles engine that had generated +$638,651 in verified profit across 9,555 bets — entered the tournament with confidence. Seven days later, we were down $89,814.

This is not a marketing story about how great we are. This is a transparency report about what happens when an AI model meets a tournament unlike any other — and what we did about it.

The Honest Truth: We Lost $89,814 in Week 1

June 11-17, 2026 — The Numbers:

MetricResult
Win Rate36%
P/L-$89,814
PeriodJune 11-17 (7 days)
There is no way to sugarcoat a 36% win rate. Our model was underperforming, and every bet was tracked, timestamped, and published in real time on /worldcup/performance.

We did not hide it. We did not delete losing bets. We did not pause the system and pretend nothing happened.

Every single losing bet from Week 1 is still visible on our platform today.

Day 14 — All four matches lost. Total: -$44,210
Day 14 — All four matches lost. Total: -$44,210

Day 15 — Sweden vs Tunisia: -$27,000 in a single match
Day 15 — Sweden vs Tunisia: -$27,000 in a single match

Day 17 — The worst day: Argentina -$17,866, Iraq-Norway -$19,697, France -$11,946
Day 17 — The worst day: Argentina -$17,866, Iraq-Norway -$19,697, France -$11,946

These screenshots are taken directly from our live platform. Every dollar lost is recorded, timestamped, and permanently visible.

What Went Wrong — And What We Did About It

The 2026 World Cup is structurally different from any previous tournament:

  • 48 teams instead of 32 — the largest expansion in FIFA history
  • 16 debuting nations with limited or zero World Cup historical data
  • New venue dynamics — altitude in Mexico City (2,200m), extreme heat in Miami and Dallas, jet lag penalties across 3 time zones
  • Expanded group stage — 12 groups of 4, creating more low-data matchups than ever before

Our Dixon-Coles model excels with historical data. When two teams have played each other or opponents of similar caliber, the attack/defense parameter estimates are precise. But when Saudi Arabia faces a team it has never played in a competitive fixture, the model has less to work with.

What we changed (June 17-18):

  • 1WC-specific recalibration — adjusted attack/defense parameters using tournament-specific data from the first round of matches
  • 2Expanded Monte Carlo weighting — increased emphasis on real-time form signals from the tournament itself, rather than relying solely on pre-tournament estimates
  • 3Confidence penalty adjustments — applied stronger confidence discounts to matchups involving debuting nations with thin data profiles

The recalibration was completed within 48 hours. Not weeks. Not a new model. The same engine, better calibrated.

The Turnaround: +$159,195 in 5 Days

June 18-23, 2026 — Post-Calibration:

MetricBefore (Jun 11-17)After (Jun 18-23)
Win Rate36%64%
P/L-$89,814+$159,195
Net Swing+$249,009
The model did not become a different system. It became a better-informed version of itself.

Key winning signals post-calibration:

MatchMarketResultP/L
Belgium 0-0 IranAsian HandicapWON+$1,810
England 4-2 CroatiaAsian HandicapWON+$1,810
These were not lucky guesses. They were high-confidence signals generated after the model had absorbed real tournament data and adjusted its parameters accordingly.

Day 19 — The turnaround: Morocco +$22,864, Brazil +$16,201
Day 19 — The turnaround: Morocco +$22,864, Brazil +$16,201

Day 16 — Mixed results, but transparency remains: every bet published
Day 16 — Mixed results, but transparency remains: every bet published

Day 18 — England -$27,473, Ghana -$12,885 — we show it all
Day 18 — England -$27,473, Ghana -$12,885 — we show it all

Why We Show Our Losses

This is the part that matters most.

PracticeOther PlatformsOddsFlow
Show losing betsHidden or deletedEvery bet published
Win rate claims"90%+ accuracy" (fake)52.5% (real, verified)
P/L historyOnly show winning streaksFull history including -$89,814 dip
VerificationScreenshots (editable)PDF-verified, timestamped
1X2 market lossesNever mentioned-$1,655 publicly visible
Model failuresNever acknowledgedPublished and explained
We have seen platforms claim 90% or even 95% win rates. That is mathematically impossible in Asian Handicap markets over any meaningful sample size. A 52.5% win rate with proper bankroll management generates significant long-term profit — and that is exactly what our verified record shows.

Our biggest losses are public:

  • Spain vs Saudi Arabia: -$2,000

  • Ghana vs Panama: -$1,877

Our biggest wins are public too:

  • Belgium vs Iran: +$1,810

  • England vs Croatia: +$1,810

If a platform only shows you the good days, you should not trust any of their numbers.

Our Transparency Standard

Every signal OddsFlow generates follows this verification chain:

  • 1Timestamped at generation — each signal is recorded with the exact time it was produced, before the match begins
  • 2PDF-verified — complete bet records are available as downloadable PDF proof
  • 3Published in real time — signals appear on /predictions/live and /worldcup/predictions/live as they are generated
  • 4Settlement is automatic — results are calculated from official match data, not manually entered
  • 5Every loss is permanent — we do not delete, hide, or retroactively modify any bet record
"If we cannot show you the bad days, you should not trust the good days."

This is not a slogan. It is our operating principle. The -$89,814 from Week 1 is permanently visible on /worldcup/performance alongside the +$159,195 recovery. Both are equally important to our credibility.

The Numbers: World Cup 2026

MetricValue
Total Bets134
Matches Covered40
Total Profit+$21,470
Overall Win Rate52.5%
Breakdown by Market:
MarketP/LNote
Asian Handicap+$11,850Our strongest market
Over/Under+$11,275Consistent performer
1X2 Moneyline-$1,655Net loss — publicly visible
Yes, we lost money on 1X2 Moneyline. We publish that too.

The Numbers: Platform-Wide

MetricValue
Total Profit+$638,651
Total Bets9,555
ROI+17.74%
Win Rate54.3%
These numbers cover every bet OddsFlow has ever generated across all leagues and competitions. Every single one is verified and publicly accessible.

See for yourself:



Join OddsFlow Telegram -> — Free AI signals and live match updates

Ask AI -> — Get instant AI analysis on any World Cup match


See Our Full Track Record -> — Every bet. Every loss. Every win. Verified.

Get Started Free -> — Access AI predictions for World Cup 2026 and 19+ leagues. No credit card required.

*All results are verified, timestamped, and permanently published. OddsFlow provides AI-powered predictions for informational and entertainment purposes only. Always bet responsibly.*

#transparency#AI predictions#World Cup 2026#verified results#performance proof#honest predictions#Dixon-Coles

Ready to Try AI-Powered Predictions?

Start your free trial today and see how OddsFlow's AI can help you find value in football betting.

Get Started