What we called last week, and how it finished
Seven published assessments, settled in full including the four that lost. A record you can only see when it is good is not a record at all.
Seven assessments were published last week. Three won, four lost, and the week finished slightly down. That sentence is the entire result, and it is worth almost nothing on its own, which is the reason the rest of this piece exists.
The week, in full
The three that won were priced at 2.30, 1.85 and 3.40. The four that lost
were at 1.95, 2.05, 2.60 and 1.70. Nothing about that distribution is
unusual. A set of bets averaging a little over even money should lose the
majority of the time in a given week roughly as often as it wins it, and any
week can look like either a system or a disaster depending on which two results
flip.
The number that carries information
Six of the seven were taken at a price better than the one the same bet closed at. That is the figure worth watching, because it accumulates evidence at a rate results cannot: it is measured on every bet immediately, and it does not care whether the ball went in.
The seventh closed longer than we took it, which means the market moved against our reasoning after we published. Those are the ones worth reading again, and we do read them again.
The argument for weighting it this way is set out in full in why the closing line is the only honest scoreboard.
What this does not tell you
One week is not a sample. It cannot distinguish a method with a small edge from one with none, and anybody presenting a single week as proof of either is telling you about their marketing rather than their model. These are published weekly so that the sample eventually becomes large enough to say something, and so that the losing weeks are already on the record when it does.
Common questions
Do you publish losing bets as well as winning ones?
Yes. Every assessment that was published is settled here, whether it won or lost. A record that only shows the good weeks is marketing rather than evidence.
How should I read a week of results?
Mostly as noise. Seven bets is far too small a sample to say anything about a method, which is why the closing line comparison matters more than the profit figure.
Does a losing week mean the model was wrong?
Not on its own. A set of bets priced at around evens will lose more than half the time reasonably often, and reading a bad week as a broken model is the same error as reading a good week as a proven one.