The First Ten Bets in a New Market: How Do You Treat Them?

Klaus

Market Sharp
Joined
Dec 17, 2025
Messages
560
Reaction score
2
Points
18
I intend to add the Champions League to the model next season, and I am designing a protocol for the first ten bets. I would like to see yours before I finalise it.
My draft has three parts.
No stake for the first 40 matches. Predictions only, logged with a probability, scored afterwards.
Then the first ten bets at one quarter of the normal Kelly fraction.
At bet ten, a review for process errors: price, settlement, liquidity. Not for edge. Ten bets cannot measure an edge.
The part I am unsure of is the cost. A new market is a cost before it is an opportunity. I would like to know what yours has been, if you have measured it.
 
I have measured it, because I record every entry. Over fifteen years I have entered six new markets: NBA totals, college basketball, MLB sides, the NFL first-half line, hockey totals, and tennis. That gives 60 bets in the first ten of each. They returned minus 9.7%. The next fifty bets in each of the six markets, 300 in all, returned plus 0.8%.
The standard error on 60 bets is about 13 points, so the first number is a direction, not a finding. What I can say with more confidence is why. I went back through the 60 and classified each loss of value. In seven bets I took a price that was not the best available, because I did not yet know which book led the market. In three I misread a settlement rule. That is ten of 60, or 17%, with errors that had nothing to do with whether I could analyze the games.
The first ten bets in a new market do not test your edge. They test your procedure. The cost is roughly one unit per market entered, and I now budget it as a fee.
Klaus, your review at bet ten should be a procedure check. You have identified the right review. The wrong purpose would be to judge the model.
Trust the process, not your gut.
 
Darts.
First time.
Fiver a leg.
Won six out of ten.
Felt like Phil Taylor mun.
Told the lads at the pub I'd cracked it.
Upped it to twenty quid.
Next ten.
Won two.
Never got the six back butt.
Not the money.
The six.
 
As a coach, you never install a new play in the playoff game. You run it in a scrimmage first, at half speed, where it doesn't count, and you watch what the players actually do with it, not what the board says they'll do.
I skipped that step with tennis. I looked at it and thought "two guys, one winner, how hard can it be." I bet my first ten at my normal $500. Three won, seven lost, and I was down about $2,100. I hadn't lost the money on analysis. I'd lost it on not knowing what a retirement does to a bet, or what a surface does to a favorite.
I think there's a middle ground, and it's treating the first ten like a scrimmage. Small, observed, written up afterwards. The film from a scrimmage is more useful than the film from a game, because nobody's performing for it.
 
Here's what nobody wants to hear: I walked into NHL with a toolkit built for football and assumed it would work. Public percentages, reverse line movement, the whole bag. First ten at 2.5% units.
The problem wasn't the analysis. It was that I didn't know which of my sources to trust. In college football I know which site lags and which book copies which. In hockey I had three percentage sources that disagreed by up to twelve points on the same game, and no way to tell which was real. Four of my ten "sharp side" plays had nothing sharp behind them. (I made a thesis out of a book shading a number. Again.)
Vegas knows. It also, annoyingly, knows which of the three sources is right.
 
I work on Pinnacle Terminal (an odds tool), so bear that in mind.

One thing I'd add to Klaus's bet-ten review: log closing line value on all ten, measured against Pinnacle's no-vig close. Ten results are noise, as SharpEddie says, but CLV gives you a reading on every bet, and it picks up the errors in his list. If you took 2.05 while another book had 2.15, it shows on that bet whether it won or lost.

Use the fair close rather than the raw one. Say you took 2.08 and Pinnacle closed 1.95/1.95. Against the raw close that's +6.67%. Take the margin out and the fair close is 2.00, so the real figure is +4.00%. In a market where you don't know yet which book leads, a gap that size can make your prices look fine.

Ten readings won't prove an edge. If most of the ten close against you, though, I'd look at how you're taking prices before touching the model. oli_sussex's point about volume applies here too: a thin close is a weak benchmark, so I'd only lean on CLV where the closing market had proper limits.

There's a free calculator if anyone wants to check their own numbers (no login, batch mode for a list of bets): CLV Calculator - Raw and No-Vig CLV · Pinnacle Terminal
 
I work on Pinnacle Terminal (an odds tool), so bear that in mind.

One thing I'd add to Klaus's bet-ten review: log closing line value on all ten, measured against Pinnacle's no-vig close. Ten results are noise, as SharpEddie says, but CLV gives you a reading on every bet, and it picks up the errors in his list. If you took 2.05 while another book had 2.15, it shows on that bet whether it won or lost.

Use the fair close rather than the raw one. Say you took 2.08 and Pinnacle closed 1.95/1.95. Against the raw close that's +6.67%. Take the margin out and the fair close is 2.00, so the real figure is +4.00%. In a market where you don't know yet which book leads, a gap that size can make your prices look fine.

Ten readings won't prove an edge. If most of the ten close against you, though, I'd look at how you're taking prices before touching the model. oli_sussex's point about volume applies here too: a thin close is a weak benchmark, so I'd only lean on CLV where the closing market had proper limits.

There's a free calculator if anyone wants to check their own numbers (no login, batch mode for a list of bets): CLV Calculator - Raw and No-Vig CLV · Pinnacle Terminal
I just want to note to other posters.
Links are allowed when your comment is useful as the above.
 
lads the dangerous part of the first ten isnt the losses... its the ones you win...
i started tennis at three in the morning on a weekday... first ten... won five... and what i took from those five was that i was good at it... i wasnt good at it... i didnt know the scoring... but five of ten is the best thing that can happen to someone like me in a new market because it hands you the proof you were looking for...
and then you spend the next year trying to get those five back...
the first big win i ever had was a nine leg acca picked by names and colours... it was exactly this... a new market... beginners luck... and i still havent recovered from being right...
 
One must consider that the only time I have behaved in a new market as I advise others to behave was when I came to the Asian handicap, some twelve years ago, I understood nothing of the quarter lines and had the good sense to know it, so for eight weeks I wrote ten hypothetical punts each week, with a rationale beside each, graded them against the closing price, and placed nothing, I was rather pleased with myself until I made my first real ten, at ten pounds apiece, and discovered that I had misread the rule, I had supposed that a draw on a minus quarter line returned the stake, when in fact it loses half of it, four of the ten finished level, I recorded each as a refund, and it was only the bank balance, which is a cruel and exact examiner, that told me I was twenty pounds short, the eight weeks of paper punts had taught me a good deal about the matches and nothing whatever about the settlement, which is the very thing that cannot be learnt from paper, I have since taken to reading the settlement terms of any new market before I read a single team sheet, and to placing the first real punt for a sum so small that I should not mind being told I had it wrong.
 
One unit per market entered is the number I needed. I will budget it as a fee and not count it against the model.
Prof's case is the one my draft does not catch. Paper predictions test the analysis. They cannot test the settlement. I am adding a fourth step: read the settlement rules for every new bet type before the first stake, and place one bet at the minimum to confirm I have understood them.
 
Back
Top
GOALLLL!
Odds