Group stage: our simulated portfolio ends down 30%, and we are publishing it
Let us be blunt: across the 2026 World Cup group stage, the Footlab engine's simulated portfolio fell from EUR 100 to EUR 69.63, a -30.37% result. Return on turnover came in at -3.03% over 65 simulated calls. We are showing this losing curve on purpose: a serious analytics tool displays its real results, especially the bad ones. Radical transparency beats a track record reconstructed after the fact to look good.
The real curve, with no misleading axis
The reading is unambiguous. The portfolio climbs to EUR 129.40 on 13/06 (matchday 3) on a strong start, then drops. From 16/06 onward the slope turns durably negative, ending at EUR 69.63 on 28/06. Each point is the end-of-day cumulative capital. The projections are locked before each kick-off: the curve is a deterministic, auditable reconstruction, not a score rewritten once matches are over.
What the model read well, and its blind spots
The engine framed the clear favourites correctly. Across their three group games, the estimated probability it assigned to the outcome that actually happened reached 75% for Argentina, around 60% for France and 57% for Spain: three sides whose defensive solidity (Spain and Argentina, zero or one goal conceded) confirmed the expected level.
The blind spots are just as clear. For Iran (22.9%), Cape Verde (23.2%), Turkey or Portugal (34.9%), the model gave a low probability to what occurred: the prediction-versus-reality gap is widest there. These teams' profiles fooled the engine, sides whose recent results drifted from their Elo level through short-term over or under-performance. Above all, one structural bias stands out: the engine never backed the draw (0 of 65), while a run of draws proved costly, such as Spain - Cape Verde 0-0, Portugal - DR Congo 1-1 or Iran - New Zealand 2-2. In a group stage draws are frequent; ignoring them is the main weakness we have identified.
Why more matches sharpen the projection (and why that is not a promise)
This is not wishful thinking. Mechanically, national-team Elo updates on real results: after 72 recent international matches, the strength gaps between nations are better estimated than before the tournament, when the data mostly came from qualifiers. Recalibration also touches internal parameters, notably the estimated probability assigned to the draw, which this real data lets us revise. That improves the model's calibration going forward; it guarantees no future result. Separating data-driven learning from a results promise is exactly what sets an analytics tool apart from a sales pitch.
The group stage in numbers
72 matches, 215 goals, that is 2.99 goals per game. In attack, Germany, the Netherlands and France lead with 10 goals each. In defence, Mexico and Spain finished the groups without conceding once, ahead of Brazil, Argentina and Portugal (one goal). These factual markers anchor the projections for the knockout stage.
Frequently asked questions
Does Footlab guarantee winnings? No. Footlab is a data-driven decision-support tool; no result is guaranteed, as this openly negative review shows.
How can a 60% hit rate still mean a negative ROI? Hit rate alone says nothing about profitability: 39 winning calls out of 65 do not offset the size of the losing ones. Only ROI, negative here, matters.
What is the Footlab engine? A data model combining normalised Elo, form indicators and estimated per-scenario probabilities (1, X, 2), recalibrated on real results.
Why publish a loss? Because an honest record, even a negative one, is more credible than a cherry-picked track record.
Explore the detailed projections and the engine's method on footlab-data.com. Footlab remains a statistical analysis tool, never an incentive to bet: please play responsibly and keep in mind that no result is guaranteed. If gambling is affecting you, seek support from a national helpline such as BeGambleAware.