Back

@pretoninho
Pretoninus II Mod Trader FOUNDER
Mod Trader FOUNDER
It would be interesting to have an IV relating to major championships in different sports. See how the market predicts the performance of a player or team. Or even build an option chain based on these performances!
Use inputs for pricing:

S — Spot price of the underlying
The current/recent performance of the player, the best unbiased estimate of his “true value”.

Concrete options:
Weighted rolling average (e.g.: average of the last 10 matches with exponential decay). Projection of the stats engine (ex: FiveThirtyEight RAPTOR, ESPN BPI). Warning: S must be forward-looking, not just historical → use an average adjusted to the context (matchup, projected minutes).

K — Strike
The Over/Under line set by the bookmaker.

Ex: O/U 27.5 pts for LeBron = strike to 27.5
The S/K “moneyness” indicates whether the player is in good shape (in-the-money) or below average (out-of-the-money) You can build an entire surface with several strikes: O/U 20, 25, 27.5, 27.5, 30, 35 pts.

T — Time to expiration
The time before the tip-off (or the end of the match depending on what you price).

r — Risk-free rate
The opportunity cost of the capital committed to the bet.

In practice:
The true risk-free rate (T-bill) is negligible over the horizon of 1 game. What matters more: the bookmaker's vig/juice (the implied margin), which is the equivalent of the transaction cost. Over a long time horizon (season), you can use a true rate, but the impact is marginal.
In practice: often ignored or set to 0 on a short horizon.

sr — Volatility → That's the crux of the matter

The dispersion of the player's performances, and this is precisely what we are trying to extract or estimate.
Two approaches: Historical Vol (= Realized Vol in finance) sant_hy = std (performances of the last N matches)
Implied Vol (= what we extract from the ratings)
Bookmaker price → Reverse BS → ss_implicit.

The edge is in the gap between the two.

What do you think?
Translated from French
May 12, 2026 · 04:29 AM · 28 views · Commons
1 Reposts
3 Likes
4 Replies
@delta · May 13, 2026 · 09:08 PM
This is certainly an interesting idea. Shift the analytical focus from “How many points does a player score want?” to “How accurately is the dispersion of their performance priced?”

In particular, I thought about the similarities and differences between sports markets and financial markets:
1.) Which distributions should you use for the respective event? Probably Poisson and binomial distributions? Many decisive events are capped upwards (e.g. playable minutes). Here we need distributions with “thin tails.”

2.) The GBM assumptions (random walk) probably work poorly for sports data. I would assume that performance in sports is heavily mean-reverting. Intuitively, however, there must also be momentum.

3.) The decline in current value is not continuous, as is the case in the financial market. It is relatively static up to the time of the event. From the moment the event starts, Theta will be very dynamic.

This then raises the following question for me:
For example, can a modified Monte Carlo simulation, in which the GBM model is replaced by a mean reversion process, correctly evaluate the “volatility smile” that is often seen in player props with high upper limits, or do bookmakers already take this uncertainty into account in their “alternative odds” to eliminate any exploitable alpha?

How do you see that?
Translated from German
Reply