• REGISTER NOW!! Why? Because you can't do much without having been registered!

    At the moment you have limited access to view all discussions - and most importantly, you haven't joined our community. What are you waiting for? Registration is fast, simple and absolutely free so please, join Join Talking Horses here!

AI Cheltenham Festival Project - Crowdsource?

It can look like learning, but it isn’t. If it refers to something from a previous chat, that’s stored context or memory, not the model improving itself.

Training would mean the underlying system changes because of your input. That doesn’t happen. You’re seeing recall or pattern matching, not genuine learning.
Ah, right. I understand now.
 
Correct me if I am wrong but has not Claude just picked from the top 2 (possibly favs) at the time of you running your request in each race?
 
Doesn't look to be that simple, though odds/market position seem important. Claude does output the reasoning, example below:

APPLYING TO THE CARD
HorseNotesVerdict
Saratoga (IRE)O'Brien trainer ✓, Walsh jockey ✓✓, 11-6 weight ok, 31 days since run ✓, OR 130 ✓, 6/1 odds (borderline)⭐ Top pick
Winston Junior (IRE)Kennedy jockey (1 win, 14% SR) ✓, 11-7, 52 days ✓, 6.5/1 odds, OR 131⭐ Strong fancy
Harwa (FR)Flanagan jockey (trending big-priced winner) ✓, 11-8 weight ✓✓, 35 days ✓, 21/1 (good odds bracket)⭐ Each-way value
Glen To Glen (IRE)Slevin jockey ✓✓, O'Brien trainer ✓✓, 11/1, OR 129, 93 days (slightly long)EW interest
Bibe Mus (FR)Nicholls trainer ✓, Twiston-Davies jockey (1 win), 11-3, 15/1, 3 days — too quick ✗Swerve
Manlaga (IRE)Henderson/Boinville, 8/1 — Henderson is 0/7 in trends, avoidSwerve

Summary: Saratoga (O'Brien/Walsh — the winning combination 2x in last 4 years) fits the profile almost perfectly. Winston Junior is the main danger. Harwa at 21/1 is the each-way value play given the Flanagan/big-price trend angle.
--------------------------------------------------------------------------------------------------------------------------
 
Back
Top