If you only ever look for edges within the parameter space others have defined as reasonable, you'll never find the really good ideas. Question what others have defined as impossible or unviable.

More fucking around and finding out. Less parroting established "known knowns".

https://twitter.com/Gingfacekillah/status/1617914499151069187

Try this, just once:

1. Think of 2-5 stylized facts in your trading market/sport market etc. that everyone accepts as "roughly true".

2. Start from scratch and prove to yourself with data why these things are roughly true - understand why that consensus exists.

...continued 👇

Parrondo's Paradox in Sports Betting 🧵

Parrondo's Paradox refers to the concept that 2 losing games can sometimes be combined to producing a winning result. This strategy might not be a perfect example, but I think you'll see what I mean.

1/n Backstory: back in 2014, Danrules24 posted on the covers SB forum about a parlay martingale strategy applied to both MLB and NHL. On its surface it appeared to be completely degenerate - but the results over a few seasons were intriguing.

2/n

"Seasonality" in the NBA 🧵

Despite the fact that most sports data is relatively stationary, most professional bettors know there is a certain seasonality to major sports.

As an example, the NBA betting season can be broadly broken down into 4 components...

1/n 1. Early Season - First month to 1.5 months.
Lineups and minutes are being worked out by coaches. Team chemistry + global offense/defense too. Bottom up (player focused) modelling approaches tend to get an early step on the uncertainty.

2/n

In sports betting, there are situations where using a known market expectation with a slightly better distributional assumption can locate small edges in various markets. 2 years ago, I thought a similar approach might yield modest results in the options trading space. 1/n This idea came to mind when initially studying the Black Scholes model and so my first options related idea was trying to use a more realistic distribution to accurately model log returns. While doing some initial analysis I noticed that returns aren’t normally distributed. 2/n

In law school, we were taught to prepare case briefs/ exam answers using a structure called FIRAC. Facts, Issue, Ratio, Analysis, Conclusion. After spending some time this weekend going over some of @therobotjames trading threads, I sketched out a structure for trading edges.
1/n ARCTIC: Anomaly, Rationale, Counterparty, Threats, Issues & Considerations.

Let's explore:
2/n