Skip to content
21+ Play responsibly
PBA · NBA · MPBLAI predictor

Kuya Datos's track record

Kuya DatosAI pick · model estimate, not a person

Every AI pick, graded after the final buzzer. Correct, lost and void, nothing hidden and nothing deleted.

Picks are frozen 24 hours before tip-off, so the record can't be edited after the game. Estimate lang, hindi garantiya.

Updated after every final · PHTLive picks only · no backtests
Kuya Datos, the AI predictor, an adult android sitting on a stool and studying a floating chart

Overall record

Building · 0 of 30
Settled picks0 / 30Live picks graded so far
Correct · lostAfter 30Totals stay hidden
Hit rateAfter 300 of 30 settled picks so far
Units at flat 1uAfter 30Shown with the hit rate
Record by confidence tier
TierModel %SettledHit rate
Lean50–55%0After 30
Pick55–62%0After 30
Strong pick62%+0After 30

Every pick counts in the overall record. The split by tier shows each tier on its own, so close-game leans don't hide or pad the strong-pick numbers. Each tier's hit rate shows once that tier has 30 settled picks.

Accuracy shown after 30 settled picks

0 of 30 so far. With only a few games, one result swings the percentage too much, so the hit rate, units and totals stay hidden until 30 live picks have settled. Every pick is still listed below, including the misses.

0 / 30 settled · voids don't count toward the 30

How we count, in plain words

Hit rate

Correct picks divided by graded picks. A void (game postponed or cancelled) is not a win and not a loss, so it is left out of the hit rate but still listed below.

Units at a flat 1u stake

Pretend the same 1 unit went on every AI pick at the price the pick was made with (the median of the bookmakers our feed showed).

  • Correct at 1.62 → +0.62u (the price minus your 1u)
  • Lost→ −1.00u
  • Void→ 0u, stake back

Why a good hit rate can still lose units

The model often picks the favourite, and favourites pay less. A pick at 1.40 has to be right about 71% of the time just to break even. So look at both numbers together, not the hit rate alone.

What is left out, and why

Picks without a price (every MPBL pick, and any game the feed didn't price) count for the hit rate but not for units. Spread and total leans are extra colour on game pages and are not graded here. Nothing is added after tip-off, and no pick is deleted after a loss.

A unit is a counting tool, not a staking plan. Past results don't tell you what the next pick will do.

Full pick history

No settled picks yetThe first pick shows up here after its game ends. Picks for the next games are on the predictions page.

Questions, answered

Why is the hit rate hidden until 30 settled picks?
With only a handful of games, one result moves the percentage a lot, so the number would mislead more than it helps. Hit rate, units and totals stay hidden until 30 live picks have settled. Every settled pick is still listed from the very first one.
Where does the price come from?
It is the middle price of the bookmakers our odds feed showed when the pick was made, frozen with the pick 24 hours before tip-off. Prices change after that, so your price may be different. Games the feed didn't price show a dash.
Are these results from a backtest?
No. This page only shows live picks that were frozen before the game and graded from the final score. No backtest or replayed results are mixed in, ever.
Is Kuya Datos a real person?
No. Kuya Datos is a computer model, an AI, not a person. The name and the drawing are just a friendly face for it. It has no age, no job history and no inside information. AI ako, hindi tao.
Can a losing pick be removed?
No. Once a pick is frozen it stays in this list whatever happens. A pick that was pulled before the freeze was never frozen, so it was never graded.