Key takeaways
- The card shows the selection, odds at publication, estimated chance, outcome risk, fair price and a play-from minimum equal to 1 divided by the probability.
- The probability is mostly the market's fair price; the team-form model has a weight of only 0.25 in the regular season.
- The model excludes confirmed lineups and individual starting-pitcher models, so check those yourself before betting.
- Slyx has not demonstrated independent profitability and short samples prove nothing; losses stay publicly visible.
What the analysis card tells you at a glance
A Slyx analysis card answers one question: what is the lowest price at which a given selection is worth considering? Every figure on it is an estimate rather than a certainty, so it helps to know where each one comes from and what is missing from it.
Slyx covers only the NHL, MLB and MLS. Daily lists live under today's picks, and all write-ups are on the analyses page. A detail page (the /analysis/ address plus the analysis number) shows the selection, odds at publication, the model's estimated chance, outcome risk, status, a "why this selection" summary, team form, an odds calculator with implied probability, edge and EV, AI findings with source links and the fair-price line.
A sensible reading order is simple: first the selection and odds, then the fair price, then the estimated chance, the play-from threshold and finally the reasons and sources. If you start with the reasons, it is easy to be carried away by a story and overlook the numbers.
What is free and what is not: one pick per Prague day is published free, the rest are premium. A logged-in free user unlocks one tip per sport, and guests see locked previews. After kickoff the market, pick and odds of premium analyses are revealed, while the reasoning and fair price stay premium. Premium costs 399 CZK per month (30 days, auto-renewal, cancel anytime); see pricing.
Fair price: how it is built and why it anchors everything
Bookmaker odds contain a margin. Slyx strips it out of each bookmaker with a power de-vig and combines the resulting probabilities as a weighted median across books. Sharp books (Pinnacle, Betfair, Matchbook, Smarkets, Circa) carry 3x weight. A market counts only if at least two bookmakers price it completely.
The supported markets are moneyline and totals on .5 lines. MLS moneyline has three outcomes including the draw, so it cannot be read as a yes/no bet, and the three probabilities must sum to 100%.
Fair price is not the truth about a game; it is the market consensus with the margin removed. Why that matters: if your own view differs sharply from the fair price, the sensible question is what you know that the market does not, rather than assuming the market is wrong. More detail is on the methodology page, and probability basics are in the learning guide.
Estimated chance and risk: market plus model
The final probability blends the market fair price with a Poisson model of recent team form. The model weight is 0.25 in the regular season, 0.10 in preseason and 0 when no model is available. The market stays the anchor; the model nudges it.
What the model does not know
The model does not include confirmed lineups or individual starting-pitcher models. That is exactly what you should check yourself, especially in MLB, where the name of the starter changes a game more than in other sports. The market part of the estimate may already reflect this information, but you cannot be sure it reflects all of it.
What the AI step does
Before publication an AI research step looks for injuries, lineups and news with cited sources. Its verdict can only keep, downgrade or veto a pick. It never changes the market, selection, odds, line or probability. Treat the source links in the AI findings as a starting point for your own check, not a final ruling.
How to read the risk
Outcome risk on the card describes uncertainty, not a guarantee. Even a pick rated at 60% loses four times in ten, and it does so even when the estimate is perfectly right. A single loss therefore says very little about the quality of an estimate.
Play-from odds and edge: how a pick gets published
The play-from minimum is 1 divided by the blended probability. If the best available price is lower, the expected value (EV) disappears. The card shows the best odds with the bookmaker, the fair price and this threshold.
To be published, a pick needs EV of at least 1% when a sharp book is present and at least 3% without one. Picks with EV of zero or less, or with an AI veto, are dropped. If fewer matches qualify, fewer picks go out, or none; the limit is never filled artificially. Other caps apply: at most 15 analyses a day and 5 per league, matches must start at least 45 minutes after publication, and each event gets one pick.
Picks appear in three daily editions at 12:00, 16:00 and 20:00 Prague time. Up to two picks per edition are marked "top". The label is an internal ranking within the edition, not a promise of a result.
Worked example, step by step
The numbers below are purely illustrative; they are not a real pick or measured Slyx performance.
- 1Estimated chance: p = 0.55.
- 2Play-from minimum: 1 / 0.55 = 1.818, roughly 1.82.
- 3Offered odds: 1.95. Implied probability = 1 / 1.95 = 51.28%.
- 4Edge: 100 × (0.55 − 0.5128) = 3.72 percentage points.
- 5EV = 0.55 × 1.95 − 1 = 0.0725, or 7.3%.
- 6Full Kelly = (0.55 × 1.95 − 1) / (1.95 − 1) = 0.0725 / 0.95 = 7.6% of bankroll; in practice that is a ceiling, not a recommendation.
At odds of 1.75, EV would be 0.55 × 1.75 − 1 = −3.75% and Kelly 0. Odds of 1.75 sit below the 1.82 threshold, so the pick would not be worth it. This applies to binary markets only; markets with pushes or split settlement work differently.
The calculator on the analysis page performs these same steps. Try it on illustrative numbers before using a real price, and notice how a small change in p moves EV: with p = 0.52 and odds of 1.95, EV = 0.52 × 1.95 − 1 = 1.4%, far below the previous example. That is why edge is so sensitive to errors in the estimate.
What the card does not tell you, and how to read results
The card does not say the pick will win, and Slyx has not demonstrated independent profitability. Average CLV is shown on track record; the per-pick CLV chart there is premium, and the analysis card itself carries no per-pick CLV. CLV is (published odds / closing odds − 1) × 100, using the closing price from the same bookmaker where possible, captured close to the start.
Look at the public ledger. In the last 30 days (to 13 Oct 2026) there were 145 settled picks, 78 won and 67 lost, a 53.8% hit rate, flat-stake ROI of 11.5% and average CLV of 0.9% over 78 picks with a closing price. In the last 90 days, though, 286 settled picks (144 won, 142 lost) give a 50.3% hit rate and ROI of −0.3%. For MLB the 90-day ROI is −6.4% over 227 picks. NHL (50 picks, ROI 25.8%) and MLS (9 picks) are too small to conclude anything. A short period proves nothing.
Predictions are stored before kickoff and evaluated on later data (Brier score, log loss). Brier score is (p − outcome)²; at an estimate of 0.60 a win gives (0.6 − 1)² = 0.16 and a loss (0.6 − 0)² = 0.36 (illustrative). Settlement runs automatically in the morning and evening Prague time, and no model is switched automatically without review.
Every published pick, losses included, is in the history; the weekly reset does not erase them. The bankroll is a model portfolio of flat 500 CZK tickets per analysis, not proof of money wagered. See results for the overview.
A short checklist before you decide
- Compare the live price with the play-from figure; below it, the pick is not worth it.
- Check lineups and starting pitchers with sources outside the model, such as the official league sources summarised on the sources page.
- See whether the price has moved since publication and whether waiting makes sense.
- Read the AI findings and follow the source links.
- Look at sample size, not only ROI.
- Set a limit beforehand. Betting is for adults aged 18+; see responsible betting.
For how picks are chosen, read how Slyx selects picks, and for a general routine the sports analysis checklist.
Reader questions
Why does the card's price differ from the live price at my bookmaker?
The card shows the best odds and bookmaker at the time of publication. Prices move and bookmakers differ. Compare your current offer with the play-from minimum; if it is below that, the model's expected value has gone and you should skip the pick.
Does a higher estimated chance mean a safer win?
No. The estimate is a probability, not a forecast. A pick rated at 65% loses roughly one time in three. What matters is whether the odds beat the fair threshold, which edge and EV check. Judge it over a large sample, not single games.
Where can I see Slyx's losses and results?
In the results section: /historie lists every published pick newest first, including wins, losses and voids, 10 per page. The weekly bankroll reset does not erase losses. The bankroll is a model portfolio, not proof of money actually wagered.
Sources and methodology
Responsible use
For adults 18+ only. This is educational content. Outcomes are uncertain; gambling can cause financial loss. Set time and money limits beforehand and never chase losses.