Sup guys,

I’m a data scientist and MMA fan. I’ve been working on a data-driven MMA prediction model and wanted to see if there’s interest here.

Quick breakdown:

- Scraped historical MMA data (over 1M fights)
- Built ELO-like fighter ratings
- Added pre-fight features only (career stats, streaks, age, weight class, nationality, style, etc.)

Trained 3 ML models (Random Forest, XGB, Neural Network)

Compared predictions vs historical odds to find value spots

Backtests on unseen data show >15% ROI so far. Still early (v1), but results have been solid.

Thinking about posting picks / sharing tips here for FG as a starting point to monetize this project, please let me know if anyone is interested


For reference these were the picks for the two events that happened yesterday (ACA and ARES):

event_url event_name event_date fighter_url
https://www.sherdog.com/events/ACA-198-Omarov-vs-Suleymanov-2-110775 ACA 198 - Omarov vs. Suleymanov 2 2026-01-09 https://www.sherdog.com/fighter/Daud-Shaikhaev-161979
https://www.sherdog.com/events/ACA-198-Omarov-vs-Suleymanov-2-110775 ACA 198 - Omarov vs. Suleymanov 2 2026-01-09 https://www.sherdog.com/fighter/Mansur-Khatuev-258221
https://www.sherdog.com/events/Ares-37-Augen-vs-Shiguemoto-2-110495 Ares 37 - Augen vs. Shiguemoto 2 2026-01-09 https://www.sherdog.com/fighter/Sofiane-El-Fatni-442419
https://www.sherdog.com/events/Ares-37-Augen-vs-Shiguemoto-2-110495 Ares 37 - Augen vs. Shiguemoto 2 2026-01-09 https://www.sherdog.com/fighter/Xavier-Lessou-402582
https://www.sherdog.com/events/Ares-37-Augen-vs-Shiguemoto-2-110495 Ares 37 - Augen vs. Shiguemoto 2 2026-01-09 https://www.sherdog.com/fighter/Ylies-Djiroun-86566

I wish you the best but as someone whos been capping MMA for 20 years these never work and don't get too excited until you've ran at least 100+ events. Should track your events on betmma.tips though.

Appreciate the advice and good call on betmma.tips


So far it’s been tested on ~1.6k unseen historical bets and results are positive, but I’m fully aware that past performance doesn’t guarantee anything going forward.

I’ve also seen similar projects but honestly none seemed to be as robust and scientifically sounds as I believe mine to be.

I've taken a very conservative approach during training, prioritizing consistency and downside protection even if that means lower ROI.

edit: Also, I purposefully left out in-cage stats (take downs, strikes landed, etc) which is what people usually focus the most when trying to predict fights.
This data is extremely noisy and reaches a level of detail that cannot be predicted by a model at scale.
My approach is more high level, predicting trends of how fighters progress in their careers and how the market is usually pricing those fights

Anyway, time will tell if I have a golden goose or not lol

This post was edited by ae222 on Jan 10 2026 05:51pm

Picks with positive expected value for UFC 324:


Alex Perez vs Charles Johnson -> odds while picking: +169
Nikita Krylov vs Modestas Bukauskas -> +120
Sean OMalley vs Yadong Song -> -205
Alexander Hernandez vs Michael Johnson -> -161
Jean Silva vs Arnold Allen -> -280

Strong start

Impressive picks for sure. Did you parlay that?

Nope
I’m doing just straight picks to be more on the safe and consistent side

UFC 325 picks (honestly not super confident in this one xD, looking a bit sketchy)

Lawrence Lui Rangbo Sulang
Diego Lopes Alexander Volkanovski
Rafael Fiziev Mauricio Ruffy
Zha Yi Kaan Ofli
Dom Mar Fan Sang Wook Kim
Torrez Finney Jacob Malkoun
Tallison Teixeira Tai Tuivasa

wow super robbery on Yi Zha vs Kaan

for my luck

I also knew it wasnt a good idea to bet against Volk and was not confident on Fiziev vs Ruffy
At least Lawrence Lui bet saved me from a bigger loss