A practical Ontario DAILY KENO results tracker with three deterministic choices for every Pick 2–10, exact prize accounting, data-freshness checks, completed-prediction reporting, and budget awareness. The sealed research framework remains available under Advanced Research.
Expected matches = picks × 20 ÷ 70
Historical analysis does not forecast future draw results. Every fixed set of the same size has the same expected matches in a fair draw.
Choice A is the highest-ranked set. Choices B, C, D and E stay within the same ranked pool while progressively reducing overlap. They are alternatives—not higher-odds tickets.
Loading five choices…
Selected: Choice A
Loading fixed selection…
Expected matches—
True chance of any prize—
Prize at selected wager—
Data through—
Every individual number still has a true 20/70 = 28.57% chance of appearing in a fair draw. Historical ranking does not change that probability. Prize amounts shown use the published OLG DAILY KENO fixed prize table; OLG's per-category payout cap can cause proportional reductions in exceptional draws, so verify any real ticket with OLG.
Show full pick 2–10 odds and ranking details
Pick
Five fixed choices
Expected matches
Chance all selected match
Chance of any prize
Why these numbers rank highest
Responsible budget
Plan spending before the draw. Buying more tickets increases total spending and coverage; it does not make each ticket statistically better.
Total planned cost: $1.00
Set a limit you are comfortable losing. This tool does not improve the expected return of DAILY KENO.
Previous completed prediction
No completed future prediction yet.
Completed predictions
This section judges practical results: actual winning numbers, matches, whether the selection won a cash prize, wager, prize amount, and lifetime cost/return. Pending predictions are shown too, but are excluded from completed ROI.
Historical analysis
Most Frequent
Least Frequent
Draws Since Last Appearance
Most Common Pairs
Frequency heatmap
Darker cells appeared more often inside this selected historical sample.
Composition
Sequence behaviour
Rolling backtest
Uses only the deep archive. For every evaluation draw, the fixed shortlist and comparison strategies see exactly the preceding window and never the draw being evaluated.
Run a backtest to compare both match averages and practical prize-winning rate, total prizes, cost, net return and ROI. Results are reproducible after refreshing the page.
Version 5 · V4 sealed research retained
Research Lab
This lab does not assume historical patterns predict a fair lottery. The retained V4 framework separates model selection from a one-time final holdout, and reports uncertainty at the draw level so repeated simulated tickets do not create fake precision.
Fair baseline2.857matches / 10 picks
SelectionValidation onlycandidate pool fixed in code
Final testReveal oncethen holdout is burned locally
Inference unitDrawnot simulated ticket
The local seal prevents repeated final-holdout reveals for the same built-in dataset/session in this browser. Clearing browser storage or editing the source can bypass a local seal; a server/public registry would be required for tamper-proof preregistration.
1 · Preregister and seal an experiment
Choose the design before seeing the final holdout. The seal records the dataset fingerprint, candidate-model set, metric, split boundaries, and settings. Once sealed, these controls are frozen until the active experiment is reset.
No experiment preregistered.
No final-holdout result is visible before preregistration and validation.
2 · Validation/model selection
All fixed candidate models are compared only on the preregistered validation segment. The best challenger is then frozen. This step never evaluates the final holdout.
Preregister an experiment first.
3 · One-time final holdout
Only the Champion and the single challenger frozen by validation are evaluated here. Revealing the result burns the final holdout for this dataset/session in this browser.
Validation must finish first.
Bayesian calibration
Every number starts at the fair 20/70 baseline. Historical evidence moves the estimate only modestly because a prior shrinks noisy frequencies back toward 28.57%.
Diagnostics use the selected session.
Regime & feature diagnostics
Descriptive checks for repeat counts, adjacent pairs, and frequency shifts. They are not activated as predictors unless unseen-data evidence supports them.
Diagnostics will use the recent official-feed history.
Fair-lottery model-shopping benchmark
After the final holdout is revealed, The retained sealed framework repeats the entire validation-selection + final-test workflow on synthetic fair lotteries. This asks how often random histories create an apparent lift at least as large as the observed one.
Reveal the final holdout first.
Prediction ledger methodology
Future predictions are now saved from the Home screen with their exact pick count and wager. Once a matching official result is loaded, V5 records matches, won/lost status, prize amount, cost and net result. Export JSON regularly if you want a durable audit trail because browser storage can be erased.
Prize calculations use OLG's fixed DAILY KENO prize structure for $1, $2, $5 and $10 wagers. Only one prize is available per selection.
Diversified ticket portfolio
If you choose to play multiple tickets, this tool reduces unnecessary overlap while keeping stronger-ranked numbers in the mix. Diversification changes coverage across tickets; it does not improve the lottery's underlying expected return.
Deep archive
— valid historical draws
User-supplied historical archive. It is used for deep backtesting and is not described as current.
Recent official-feed history
— recent draws
Collected from OLG’s public DAILY KENO results feed on August 24, 2026. Verify through OLG.
Known historical gap
The app keeps current analysis and deep-archive backtesting separate. It never treats this gap as a continuous sequence.
Import or export results
Import KenoSpots JSON, the archive CSV format, or CSV with drawNumber,drawTime,n1,…,n20. Accepted imports remain only in this browser.