I was a broke student
with a spreadsheet.

Two things I have always liked

Sport, and numbers. Not as separate interests, the same interest. The bit of a game I have always found most interesting is the bit nobody shows you: what the market thought, and whether it was right.

I was at university and comprehensively broke. That matters to this story, because being broke is what made me read the fine print. When $20 is a real amount of money, you stop asking “who wins?” and start asking “what am I being charged to find out?”

Then I found expected value, and it broke my brain a bit

The idea is almost annoyingly simple once it lands: you do not have to predict anything. You have to notice when a price is wrong.

A bet is not good because the team is good. It is good because the number is bigger than it should be. Those are completely different claims, and only one of them can be checked.

If that is new to you, I wrote the long version with a coin you can flip yourself: what positive expected value betting is.

I did it by hand, badly

There was no system. There was a spreadsheet, a lot of tabs open, and me typing prices into cells to work out whether bet365 was offering more than the fair price on something. One market at a time. It was slow, it was manual, and I missed almost everything because by the time I had finished the arithmetic the price had moved.

But the principle held. Bet the number, not the team, and the results stop being random in a way you can actually see.

Then bet365 stopped being an option here

Which left the TAB. And the TAB is not the same product: fewer markets, wider prices, and no obvious way for a New Zealander to compare what they are being offered against anything else.

That sounds like the end of the story. It is actually the reason there is one.

One book, no comparison, a margin nobody measures

A market with one dominant bookmaker and no price comparison is a market where nobody finds out what they are paying. So I measured it: a full snapshot of the board, 146,956 prices, and a median house margin of 7.2%. That is on the margin study, with every market browsable by sport.

And a single book with a wide margin still misprices things, often and predictably. Those mistakes were exactly what I had been hunting by hand at university. So I automated it: scanners that price every TAB market against the sharpest books in the world, every fifteen minutes, and flag the ones that are clearly wrong.

What I refuse to do with it

Almost everything in this industry is sold on a screenshot of a winning weekend. I find that genuinely irritating, so this is built the other way round:

• Every pick is graded in public, losses included, on the go-forward book, the markets I still send.
• I report closing line value, not a win rate. A win rate can be made to look like anything by choosing which odds to bet at. Beating the closing price cannot.
• No figure appears that I cannot show you the working for. Where a number is a sample rather than the whole population, it says so.

None of which makes betting a sensible way to make money. It makes it a measurable one, and measurable is the most honest thing anyone in this business can offer you.