I model sports props… but strikeouts are my favorite | Owner @AMPAnalytics_ | https://teapaw.org/t.co/7czOwLz7eb
Joined March 2023
- Tweets2.5K
- Following126
- Followers3.1K
- Likes3.3K
Shai 3s 🪜 💰
Duncan 3s 🪜 💰
+220 ✅
+350 ✅
+560 ✅
+750 ✅
AMPanalyticsarena.com/signup
Also added the 3PM projections to the website with a nifty feature, "who's playing"
Think a guy sits tonight before it’s announced? Set him to Out and his whole team reprices instantly... Get there before the books do
7-day free trial: ampanalyticsarena.com/signup
+19,700 banger last night, I never formally "introduced" my NBA model so here we go....One ticket doesn't prove anything, so here’s what does.
My NBA model has priced 92,324 alt 3PM lines across 3 full seasons it never trained on, plus every game this season so far.
When it says a play hits 10% of the time, it hits about 10%. When it says 20%, about 20%, and so on.
That’s the whole game with longshots. A +1900 isn’t a good bet because it might hit, but when the real chance is bigger than the price says, and you only know that if your percentages are honest...
Here’s how it works 🧵
Every play posts live in the VIP Discord: the best price across books, how far the line can move before it’s a pass, units, and the projection.
Also have a bot that grades the plays automatically every night at 3am so we can see how we're doing day in and day out
discord.gg/7JgD56k6f
The probabilities blend my model with the market price, and the model has to "earn its keep"
When it’s higher than the market, those lines hit more often than the market says and when it’s lower, they hit less, tested on seasons it never saw.
Every variable went through the same test and most got cut
Ladders come straight off that, every rung that clears the price bar gets posted, sized by how likely it is: 0.5u on +1, 0.25u on +2, 0.2u on +3.
Over 3 seasons of backtesting, playing the whole ladder this way grew the bankroll faster than one rung per player, every season.
Expect swings... Longshots will lose most nights, but the math works over hundreds of bets and doing this night in and night out, that's what matters.
How a projection gets built:
Minutes first... Injuries, confirmed lineups, rotation changes etc. When news drops, it reruns within seconds.
Then volume: how many 3s his team should take tonight against this opponent, and his share of them.
Then shooting, with hot and cold streaks pulled back toward his real history.
That gives a probability for every line. 3+, 4+, 5+, 6+
