Generation method · history-shaped
Deep AI
Relationship-weighted — favouring numbers that have historically turned up together, kept in balance.
Neural nets and the company numbers keep
A neural network is layers of simple units joined by weights — numbers that start random and are tuned during training until they encode the structure in the data. Deep learning stacks those layers so each extracts a more abstract representation than the last. It is genuinely powerful wherever real structure exists: language, images, speech.
The relevant idea here is the embedding, made famous by word2vec: words that appear in similar contexts get similar vectors. Applied to draws, the analogue is co-occurrence — which numbers tend to turn up together. That is computed with plain market-basket analysis: for every pair of balls, count the draws containing both, and rank the pairs by frequency.
No signal to learn in a fair draw
In a fair lottery the draws are independent, so any pair that looks “hot together” is a chance artifact — textbook spurious correlation, the kind that inevitably appears when you compare enough data. Our tendency to see meaning in it has a name: apophenia, perceiving patterns in random noise.
Machine learning needs patterns to learn from. Given none, a deep net overfits: it fits past draws almost perfectly and still performs no better than chance on the next one, because it memorised noise rather than learning a rule. The technology is real; here it can only mirror randomness back at you.
How we borrow its shape
The Deep AI method draws from genuine quantum entropy, then biases the pick toward numbers that have historically co-appeared in the real archive, kept balanced. Real randomness, read through a lens of past draws — offered as a curiosity.
- Favours numbers that have often co-appeared in past draws
- Reads pair frequencies from the real historical archive
- Keeps the result balanced rather than lopsided
Sources & further reading
- IBM — What are word embeddings? — Embeddings and the distributional idea behind word2vec and co-occurrence models.
- Turing — market basket analysis — How “which items appear together” is computed: pair frequency, support, confidence, lift.
- Statistics By Jim — spurious correlation — Why apparent associations arise by pure chance — the honest frame for lottery “pairs.”
- World Lotteries Association — random chance is the essence of the lottery — Draws are truly random and independent; no exploitable patterns.
- Britannica — Apophenia — The human tendency to perceive meaningful patterns in random data.