A collection of the main Italian lottery terms and statistical concepts used in draw analysis.

Ambata: A bet on a single number on one Lotto wheel. Probability: 1 in 18 per draw. The basic bet in Lotto: you win if the chosen number appears among the 5 drawn numbers on that wheel.
Ambo: A bet on two numbers on the same Lotto wheel. You win if both numbers appear among the 5 drawn. Probability approximately 1 in 400 per draw.
Terno: A bet on three numbers on the same Lotto wheel. You win if all three numbers appear among the 5 drawn. Probability approximately 1 in 11,748.
Quaterna: A bet on four numbers on the same Lotto wheel. You win if all four numbers appear among the 5 drawn. Probability approximately 1 in 511,038.
Cinquina: A bet on five numbers on the same Lotto wheel. You win if all five numbers appear among the 5 drawn. Probability 1 in 43,949,268: the rarest combination in Lotto.
Sigizie: A term from Italian lottery cyclometry (Garghella/Tribuzio school). It refers to cyclic configurations of numbers that tend to appear in coincidence with certain periods of the lottery cycle. An Italian lottery tradition concept, without objective probabilistic basis.
Numeri Vertibili: In the Italian lottery tradition, numbers considered 'vertibili' (rotatable) because they tend to appear across multiple wheels or in reciprocal combinations. An empirical cyclometry concept, not supported by formal probabilistic analysis.
Ciclometria lottologica: A system for analysing Lotto draws based on delay cycles and periodic recurrences, developed in the Italian school of Garghella and Tribuzio. It studies eighth-cycles and cadences in the 90 numbers of the Italian Lotto. A concept of the Italian lottery tradition, with no objective probabilistic basis.
Ritardatario (Overdue Number): A number not drawn for many consecutive extractions on a Lotto wheel. A purely descriptive measure of delay relative to theoretical frequency. Note: due to the i.i.d. independence of draws, the delay does not increase the probability of appearing next. There is no 'due' number.
Frequenza (Frequency): The number of times a number (or combination) has been drawn in a given historical period. A historical descriptive measure: it does not imply that more frequent numbers have higher probability in future draws (draws are i.i.d.).
Probabilità (Probability): A measure of the chance that an event occurs, between 0 (impossible) and 1 (certain). In lotteries it is determined by the mathematical model of the game and is constant for every draw, regardless of history.
Valore Atteso (Expected Value, EV): EV = Σ(probability × prize) − cost. For all public games the EV is negative by definition: the house retains a systematic percentage. Example: EV = (0.0012 × 250) − 10 = −€9.70 per draw. No strategy can make the EV positive over the long run.
Legge dei Grandi Numeri (Law of Large Numbers): Statistical theorem: as the number of trials increases, the relative frequency of an event converges to its theoretical probability. The convergence is asymptotic and not compensatory: it does not imply that overdue numbers are 'due' in the next draw.
Legge del Terzo (Law of the Third): Empirical regularity in the Lotto: over a cycle of draws, about two thirds of the 90 numbers are drawn and one third stays absent. It is a descriptive statistical phenomenon observable on historical data, not a prediction: it does not indicate which numbers will be drawn nor increase the chance of winning.
Indipendenza i.i.d. (i.i.d. Independence): Draws are independent and identically distributed (i.i.d.): each draw is an autonomous event with the same probability distribution, without memory of previous draws. Believing otherwise is the gambler's fallacy.
Distribuzione ipergeometrica (Hypergeometric Distribution): A probability distribution describing the probability of k successes in n draws without replacement from a finite population of N elements with K successes. The correct mathematical model for SuperEnalotto (6 from 90), 10eLotto (20 from 90) and MillionDAY (5 from 55).
ECE (Expected Calibration Error): A metric measuring how closely an AI model's stated probabilities match the actually observed outcomes. Predictions are grouped by stated confidence level and, for each group, the average confidence is compared with the real success frequency; the ECE is the weighted average of the gap. Values near 0 indicate realistic probabilities; the calibration threshold is ECE ≤ 0.05. Importantly, ECE does not measure how often the AI is right but how honest it is about its own confidence, and a low ECE does not imply winnings: games remain i.i.d. with negative expected value.
Calibration: The property of a model whose confidence percentages are reliable as real probabilities: a calibrated model stating '70%' is right about 70% of the time. It is measured via ECE. Calibration concerns the honesty of the stated confidence, not the ability to predict random outcomes: for lotteries, calibrated means realistic probabilities, never a guarantee of winning.
Model Reliability: A summary index of how trustworthy the probabilities shown by the AI are, derived from the most recent rolling ECE. Typical states: Calibrated (ECE below threshold), Calibrating (ECE still above threshold, often due to a small sample while converging), Drift detected (marked deviation), Insufficient data (sample below the minimum threshold). Games with fewer draws per week accumulate evaluation data more slowly, so their reliability index is noisier and converges over a longer period.