Two models. Two prediction markets. One World Cup.
I built Dixon-Coles + Elo from scratch, checked they're calibrated on 2,487 held-out matches, then put them against live Polymarket and Hyperliquid odds.
The market's favourite — France (18%) — is the one bet my goal model says is over-priced. Marking it to market in July.
Full code, rerun it in 30s 👇
https://t.co/sWxaFu9E9F
Two models. Two prediction markets. One World Cup.
I built Dixon-Coles + Elo from scratch, checked they're calibrated on 2,487 held-out matches, then put them against live Polymarket and Hyperliquid odds.
Great read from Two Sigma on regime modeling via machine learning: https://t.co/h5nxy8zUCK
Even though unsupervised, the GMM approach can be used as a label generator for a subsequent multiclass classification (ex ante).
The only real shortcut in life is to understand it backwards.
It's easier to solve a maze backwards. Life is the same.
Learn from people further down the path from you and make their hindsight your foresight.
Be the biggest fan of the people you care about.
Defend them.
Keep the hope alive.
Make them look good.
Catch them when they fall.
Be there when they need you.
Root for them unconditionally.
Help them accomplish their dreams.
Interesting academic paper looking into Bitmex, showing how the perpetual swap leads spot prices and serves as an effective hedge.
https://t.co/nitVvddEFb