- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Winter Olympics
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
re: Sports Betting AI Algorithm Prediction
Posted on 11/12/25 at 2:03 pm to LSUTIGERS74
Posted on 11/12/25 at 2:03 pm to LSUTIGERS74
Show some cashed out tickets.
What probablitiy% and edge combo makes for a A tier
What probablitiy% and edge combo makes for a A tier
0
Posted on 11/12/25 at 2:22 pm to ks_nola
quote:
ks_nola
Great questions! Here's how our tier system works:
TIER CRITERIA
Tier A (71.6% win rate in 2025):
• Win Probability = 60%
• Edge = 25 percentage points
• Spread = 14 points
Example: Alabama -6.5 vs Oklahoma
• Probability: 60.0%
• Edge: 26.3pp (our model says 60%, market implies 33.8%)
• Spread: 6.5
• Result: Tier A
Tier Elite (74.7% historical win rate):
• Home picks only, =57% prob, =15pp edge, =10 spread
Tier C (61.8% win rate): Betting the fade here not the pick shown
• =50% probability, =25 spread, doesn't meet A/Elite standards
Tier D: Not recommended (shown for transparency only)
LINE SOURCE & DISCREPANCIES
We pull lines from CFBD API (Consensus or DraftKings) at opening (Tuesday) - they are supposed to update every morning and night but you found a bug and the new files didn't load the tables with new data (working on this fix now)
Why Alabama shows -6.5 on our site but -7.5 on FanDuel:
• Our line: -6.5 (opening line, Tuesday)
• FanDuel: -7.5 (current line, Saturday)
• Difference: Line moved 1 point due to betting action
This is normal! Lines move throughout the week. Always use your sportsbook's current line when placing bets.
Line movement doesn't invalidate the pick - our edge calculation is what matters. A Tier A pick with a 1-point line move is still a strong play.
Hope this helps! Let me know if you have other questions.
This post was edited on 11/12/25 at 2:26 pm
Posted on 11/12/25 at 2:24 pm to High C
quote:
High C
Yes the algo only likes favorites and is extremely high probability (~85%) on 4 years of data for home favorites. In the 4 years of backtesting it has only ever picked a dog once (went 1-0), I use the tier C for fades as that is where the dogs are ~61.7% win rate betting against the pick.
Posted on 11/12/25 at 3:23 pm to LSUTIGERS74
Cool stuff man. Impressive looking site. Thanks for the trial. I put money on all 3 of the elite plays.
Posted on 11/12/25 at 4:36 pm to High C
Sketchy that no dogs are listed. This is def a square site
This post was edited on 11/12/25 at 4:37 pm
Posted on 11/12/25 at 5:05 pm to CWilken21
There’s an A game tonight for anyone riding
Posted on 11/12/25 at 6:14 pm to LSUTIGERS74
Bama moved to -5.5 not -7.5 but either way seems a one point line move would have some effect on probability and or edge unless you are suggesting people bet moneyline.
Posted on 11/12/25 at 6:45 pm to CWilken21
quote:
ketchy that no dogs are listed. This is def a square site
Yes this is one area where the algo refuses to pick dogs, I have run it mutliple times in various versions to get dog picks and it throws the probability all over the place and win rate goes to less than breakeven. The only way to bet dogs is to fade the Tier C picks which when bet as whole on week by week basis has a +60% win rate. I spoke to an old buddy who ran an algo back at MIT in the early 2000's and he was never able to get there predictions to find dogs either (as an FYI they turned 500 into 500k back before 2005 online gambling ban). The good news is even with no dogs, I will still take ~68% win rate any day
Posted on 11/12/25 at 6:51 pm to LSUTIGERS74
Whatever money you clowns will blow gambling should have just been invested in an index fund.
Posted on 11/12/25 at 7:13 pm to ks_nola
quote:
ks_nola
Great question about the Alabama line movement!
You're right that getting -5.5 is better for betting (easier to cover), but here's what matters for tier classification:
CURRENT Model LINE: Alabama -6.0
I just ran our model with live API data:
• Current spread: -6.0 (Consensus/DraftKings)
• Edge: 25.01pp
• Tier Status: STILL Tier A
LINE COMPARISON
• -6.5 (original): 26.26pp edge ? Tier A
• -6.0 (current model line): 25.01pp edge ? Tier A
• -5.5 (some books): 23.76pp edge ? Tier B
KEY POINTS
Our picks use model lines (Consensus/DraftKings)
Current model line is -6.0, which qualifies as Tier A (25.01pp = 25pp)
Some sportsbooks show -5.5, which is better for betting but drops below Tier A threshold (23.76pp < 25pp)
Tier A requires =25pp edge - this is strictly enforced in the algo.
BOTTOM LINE
Alabama is STILL Tier A based on the current model line (-6.0).
If your book shows -5.5:
• You're getting a better line (easier to cover)
• Real win probability likely ~62-64% (vs 60% at -6.0)
• Technically Tier B by our criteria, but still an excellent play
Either way, it's a strong pick. The -5.5 line is objectively better for betting, even if it drops below the Tier A threshold.
As of note I booked yesterday at -6.5 and would much rather have the -5.5 that ESPN Bets shows currently even in tier B!
Posted on 11/12/25 at 7:17 pm to LSUTIGERS74
I have been using Grok with my super prompt and its being kicking out winners each week
Posted on 11/12/25 at 7:25 pm to Jake88
Jake88
Thanks for sharing this wonderful advice but unfortunately it doesn’t make you any less of a jackass.
Thanks for sharing this wonderful advice but unfortunately it doesn’t make you any less of a jackass.
Posted on 11/12/25 at 7:28 pm to ks_nola
Nor you any less of a fool.
Posted on 11/12/25 at 7:40 pm to SDVTiger
quote:
SDVTiger
Yes with the correct wording in a prompt the AI programs out there have been extremely good at picks...just have to be really careful at the data they pull. Almost all of them will try to sneak in demo, fallback, simulation or predictions into the analysis which spits out a completely unsubstantiated pick analysis. Whats laughable when I first started I was using a prototype LLM from MIT and it was actually masking a random number generator into the code and oddly enough was better than my 1st gen algo backtesting data.
Posted on 11/12/25 at 7:51 pm to LSUTIGERS74
If I had to guess my MLB prompts through chatgpt were about 55-60
Posted on 11/12/25 at 7:56 pm to LSURoss
Wouldn't be shocked at that win rate for MLV. All the professional sports are tough to grade because Vegas is so tuned in but CFB and CBB the algos can find an edge. Noting that I am working on a CBB algo currently and it is completely different and highly selective. NFL is completely off the table even with the large data set, can't even come close to finding a profitable setup.
Posted on 11/12/25 at 8:29 pm to LSUTIGERS74
$13/week seems high and likely a scam
This post was edited on 11/12/25 at 8:30 pm
Posted on 11/12/25 at 9:00 pm to J Murdah
Sure $13 a week can seem high (but considering vig is -110 still a value compared to what you pay the house just to place the bet if you lose) and still cheaper than a damn caniac. I priced this for the average bettor just looking for a transparent data driven approach to making weekly picks with actual backtesting results, without relying on these thousands of posters on X offering free picks or "guaranteed locks" with absolutely no data to substantiate their pick. If you ask my marketing buddies they think it should be priced at $250-500 a month and target sharp bettors that fully understand the data and win rates since ~+67% win rates with back testing data is virtually unheard of, however I differ in opinion and want to help bettors like myself that have terrible gut and intuition when making CFB picks.
Posted on 11/12/25 at 9:15 pm to LSUTIGERS74
Use only tier A and play them all each week?
Posted on 11/12/25 at 9:20 pm to Gings5
quote:
Gings5
Yes thats the system, it doesn't work on individual picks. The model is based on the set for the win rate in tier A.
Tonights pick was NIU vs Mass at -10.5 and covered by a mile. 1-0 to start the week 12 picks.
You can bet other tiers but typically more bets to place to meet WR expectancy. Tier C is all fades, so bet opposites (almost always dogs to get WR of ~61%, last week was 5-1 on those and was an outlier). I use a 2U (tier A) and 1U on all other if I decide to take. For example Alabama current spread at -5.5 falls into Tier B but I know it was a Tier A earlier in the week with -6.5 spread but when spread dropped market probability increased which thru our edge lower and out of our threshold.
Page 2 of 4
Popular
Back to top
Follow TigerDroppings for LSU Football News