Frequently asked questions
What the model does, what the numbers mean, and what it deliberately does not claim.
About Kepler
What exactly is Kepler?
Kepler is a quantitative intelligence terminal for football markets. We are not a tipster service. Kepler uses a machine learning architecture and continuous data ingestion to estimate the true probability of football fixtures, and flags it in your terminal when a bookmaker offers a price that implies something different.
How is Kepler different from traditional sports tipsters?
Traditional tipsters rely on subjective analysis, gut feeling or simple statistics. Kepler operates like a quantitative trading desk. The engine processes thousands of data points, strips the bookmaker’s margin, and passes what is left through a strict multi-stage filter. A signal is dispatched only when there is a measurable edge against the market price.
Can retail users access quantitative sports betting models?
Historically this kind of modelling has been closed-door: betting syndicates and funded groups with the capital to build and run it. Kepler exists to make the same approach — de-vigged probabilities, measured edge, published results — available without that gatekeeping. The difference is access, not a different class of mathematics.
The numbers
What is expected value (+EV) and why does it matter?
Expected value is the foundation of quantitative betting. It is positive when the model’s probability for an outcome is higher than the probability implied by the bookmaker’s odds. Taking only +EV positions is what gives a mathematical advantage over a long sample — the same way a book’s margin works in its favour over many bets rather than any single one.
What is closing line value (CLV)?
Closing line value compares the odds you took against the final price before kick-off, which is the sharpest number the market produces for that match. Beating it consistently is the strongest available indicator that a model has a real edge, because it can be measured on every bet rather than waiting for hundreds of results to settle. Our article on CLV sets out the arithmetic and what a good number looks like.
What is a fair probability?
A bookmaker’s odds include its margin, so the probabilities they imply add up to more than 100%. A fair probability is what is left once that margin is removed from one bookmaker’s two prices for the same market. Match pages use Pinnacle where it prices the market and otherwise an exchange, the sharpest prices available, so the figure is the market’s own estimate — the number the model has to beat.
What is xG (expected goals)?
Expected goals measures the quality of the chances a team created, in goals: a side with 1.8 xG created chances that would usually produce about two goals, whatever the score said. Over several matches it describes how well a team is playing more reliably than goals alone, which is why match and club pages show it beside the results.
Match pages and coverage
Which football leagues and markets do you cover?
Kepler analyses fixtures worldwide and filters out thin, erratic competitions where the price moves on very little money. The focus is on leagues liquid enough that a published price is one you can actually get. The model currently publishes goal markets, and others are added as they clear the same walk-forward validation rather than because they would widen the feed.
What does “no signal” mean on a match page?
It means the model has not found a price worth publishing for that match: its probability and the market’s do not disagree by enough. Most matches never get a signal, and that is the filter working rather than a gap in coverage. If one opens before kickoff, subscribers see it in the terminal, and the match page changes to “Signal published” without naming the market or the side.
Which time zone are kickoff times in?
On the public pages — matches, leagues and clubs — kickoff times are in UTC, so a page reads the same wherever it is opened. Inside the terminal, times are shown in your own time zone by default, and you can switch to CEST or UTC in the settings.
Signals and subscription
How many signals will I receive per day?
It varies with the fixture list, and deliberately so. A pick is published only when it clears the confidence floor for its market and the available price still carries an edge, so a quiet midweek can produce nothing at all while a full weekend produces a great deal. We never lower the bar to fill a quiet day — the floors are set by measurement, separately per market, and you can see every published pick on the track record.
Do I need to be at my screen all day to catch the odds?
No. Every open signal stays in your terminal until kick-off, with its price to beat, so you can check in when it suits you. If your bookmaker still offers that price or better, take it; if the price has dropped below it, skip it. The bets you place can be logged in your portfolio.
Can I combine signals in an accumulator?
We recommend against it: bet each signal as a single. A good day is usually three or four winners out of five, not five out of five. As singles that is a profitable day; in a five-fold accumulator one miss loses the whole stake. Every figure Kepler publishes, from the track record to your own results in the terminal, is measured on single bets at one flat unit each, so none of it describes how an accumulator of the same picks would have done.
How does the free trial work?
Every new account gets the whole platform (the live terminal and the bookmaker odds behind every signal) free for 14 days, from the moment you confirm your email. No card is needed and nothing renews on its own: when the days end, choose a plan to keep the signals. Current plans and prices are on the pricing page.
Results and trust
Does Kepler guarantee a profit?
No. No quantitative model can guarantee short-term profit. Football markets carry high variance and drawdowns are a mathematical certainty, not a malfunction. Kepler is built for disciplined staking and long-run measurement — yield, ROI and accumulated edge — rather than overnight results. Every settled pick, winning and losing, is published on the track record.
How is the track record verified?
Every published pick is on the public track record at the price captured when it was issued, and it stays there whether it wins or loses: a published pick is never removed, even if the model later changes its view. Profit is counted at one flat unit per pick, each as a single bet, and closing line value compares each pick’s price with the bookmaker’s closing price, so the edge can be checked bet by bet rather than taken on trust.