Answered with numbers you can check — including ours.
Short prices are unforgiving. At odds of 1.30 you must win 76.9% of the time simply to break even, so a handful of upsets erases a long run of wins. The maths gives you almost no margin for a model that is even slightly miscalibrated.
There is also a well-documented market effect running the other way. The favourite-longshot bias means very short prices are often marginally under-priced relative to true probability while long shots are over-bet - but the effect is small, inconsistent across markets, and easily swamped by the bookmaker margin.
The practical problem is that models tend to be over-confident exactly where confidence feels safest. A strong side against a weak one is where rotation, motivation and complacency do the most damage, and those are the factors a goals-based model does not see.
This shows up in published records if the records are honest. Over 2.5 Goals is currently the model's worst published market at -29.2% ROI over 15 settled picks - low-odds markets are typically where a model's small calibration errors cost the most, which is precisely why per-market breakdowns are worth more than a single headline ROI.
MalluSports is a free prediction and analysis service only. It accepts no bets, holds no funds, operates no gambling platform and holds no gambling licence. Betting carries real financial risk. 18+ · begambleaware.org