Bet tracking template

Chapter 30 told you what a record needs to contain and why a mental one will not do. This appendix is that record, laid out to fill in: fourteen columns defined one by one, a monthly audit ritual, and a worked specimen month so you can see the whole thing populated before you populate your own. Rule one: keep it the same way every time, including the month that embarrasses you, precisely the month the template exists for.

The fourteen columns

Date, competition, market. The basics, logged before you forget the context, not reconstructed from memory or a bank statement.

Selection. The exact bet, worded as you'd need to word it to settle a dispute with the bookmaker.

Price, fractional and decimal. Both forms, always. Fractional is this book's default; decimal is what you need for the CLV column below, which is arithmetic decimal odds do more comfortably.

Bookmaker or exchange. Where you struck it. Different firms move at different speeds; a price is meaningless without knowing whose board it came from.

Stake. Level stakes throughout a segment, or Chapter 29's fractional Kelly figure if that's your method, but the same method every time within any one record. Chapter 30 already showed what happens to a yield when the stake quietly creeps up once the numbers turn good.

Your probability. Written down before you looked at the price, not after. This is Chapter 4's habit: the single column that separates a genuine value judgement from a bet reverse-engineered to look clever in hindsight.

Fair, de-vigged closing price. The market's best available estimate, stripped of its overround, at the moment the match kicked off. Chapter 8 is the method; this column is where the answer lives.

CLV. Closing line value, the fast test from Chapter 30: your decimal price taken, divided by the fair closing decimal price, minus one. A positive number means you beat the closing market; a negative one means the market closed better than you started.

Result. Settled, not pending, not "should have won."

Running profit and loss, and yield. P&L in your stake unit, yield as P&L over total staked, the two numbers that eventually decide whether any of this was worth doing.

Segment tag. A competition, a market type or a model version, whatever grouping lets a losing angle show up on its own rather than hide inside a healthy total.

The monthly audit ritual

Once a month, read the record rather than just adding to it. Two habits do most of the work.

Yield against CLV-predicted yield. Your average CLV is not literally your yield, but over a large sample the two should sit in the same neighbourhood, because the closing price is the market's best working estimate of fair odds. Compare them by segment. A yield running well ahead of a modest average CLV is most likely a good month, not a good method, exactly the gap Chapter 30 warned against celebrating. A yield lagging a healthy average CLV, in a sample nowhere near the sample-size wall, is not a crisis, just noise still settling; the honest response is to keep recording, not raise the stake.

The kill-list rule. Retire a segment when its average CLV has run flat or negative across a run long enough that Chapter 30's arithmetic would expect noise to have cleared, whatever that segment's yield says meanwhile. Do not retire one on a losing month alone; that is what variance looks like inside a genuine edge too. Do retire one that has had its fair run and still shows the market beating you to the price. That distinction, applied honestly, is the discipline this appendix gives you a template for.

A worked specimen month

The month below is an illustrative specimen built for this appendix, not a real record. Stakes are one point, taken here as ten pounds; all figures recomputed in full.

Total staked £100, total P&L +£13.11, yield +13.11 per cent, strike rate 6 from 10. Average CLV across the ten bets is +1.60 per cent, positive and pointing the same way as the yield, the shape a genuine small edge produces before the sample is large enough to prove it. Note the 23 August line: a winning bet sitting on negative CLV, the market having moved against the price between bet and kick-off. One line like that means nothing. A whole segment reading that way, once it has run its course, is the kill-list rule earning its keep.

Ten bets settles nothing on its own, and this specimen is not offered as though it did; it exists to show the template filled in, ahead of filling in your own. The real test starts at row eleven, and it does not finish for a while yet.