Tuesday, December 1, 2009

The Predict-Wager Judging System

The ten-point must system is designed for boxing. It is implemented by boxing commissions, utilized by boxing judges, and produces, when applied to a mixed combat sport, absurd decisions, even when applied correctly. It fails to incentivize proper judging; it incentivizes wins by decision (witness Guida’s transformation). Its one redeeming quality (for those who choose the system) is that it hides corruption and incompetence in its ambiguity.

Admittedly, it’s easy to point out problems. Maybe the ten-point must system is simply the best available. If not, we should stop whining until we offer an alternative. The test, then, is in designing a better system.

But one has already been designed: Pride’s judging system. Pride’s system, however, still produced its share of bad decisions and boring fights. We should consider both as design flaws of any system, and we must design a system that excludes the possibility of such flaws.

Our system will be engineered specifically for mixed combat sports. It will be greatly simplified, due to the simple nature of these sports’ objective: finish the opponent. It will incentivize good judging and legitimate efforts to finish the opponent.

With these objectives in mind, consider the “Predict-Wager System”. A revised description (from Decline and Fall 2):

“’The judge shall be paid in accordance to his return on real-time wagers on the fight. The judge will bet on the likelihood of fighter A finishing fighter B by placing a wager, weighted by his estimation of the likelihood of a finish by fighter A of fighter B. The value of the wager drops (after a calibrated grace period) as time passes, and the value of the chips used in wagering drops as more total chips are wagered. To encourage betting, increasing odds are offered on a fighter during increasing periods in which other judges have not bet on that fighter.

If there is no finish, the judge will be recorded as voting for the fighter on which he bet the most total chips, the fighter who receives the most votes shall win the decision, and the judges shall be paid a fixed rate.’

A judge, witnessing a knockdown, would bet strongly, as he would upon seeing a close armbar attempt. He would bet weakly on takedowns, depending on the perceived danger (or lack thereof) to the bottom fighter. He would bet very little on a fighter like Sherk. Accumulating “points” via lay and pray, or Tim Sylvia style strategies, would become obsolete. The “Fight of the Night” would be easy to calculate, based on total chips bet and their distribution between the fighters.

Skilled judges would be quantitatively identified simply by being those who make the most money – those who saw danger and predicted it without having devalued their chips. Unskilled judges would then be dismissed, or would quit due to low pay. With three judges betting very heavily that a fighter is about to be finished, referees who stop the fight at the correct moment can be easily identified. To greatly increase the action and risk taking of the sport, we would make the fighter’s pay in large part determined by the amount of chips bet on him to finish the fight.

We thus change the incentives: previously having little incentive to judge correctly, the judge is now paid directly in accordance with how well he can predict finishes in MMA, and thus how well he understands the sport. Knowledgeable people then have a strong incentive to become judges; Peoples would lose his incentive to remain one. And fighters would have to be capable of finishing, or at least threatening to do so, and would have a strong incentive to go for the finish, where previously their incentive was to hold down the opponent and not take risks.”

Such a system would require computers, which is likely the only reason it was not previously designed. The number of virtual “chips” is unlimited, but their value declines with every chip bet. The proper numbers for the decline, the grace period, the increasing odds, and the payouts will be determined initially by an economist, and calibrated by experiment, to ensure the judges and the fighters are incentivized in the most efficient way.

To make this all concrete, I will give an example of how to score a fight using the Predict-Wager System, based on a fight I have not previously seen. The value in chips bet, listed in dollars, is simply convention; the quantity represents my estimation of the probability of a finish. “I wager $90 worth of chips,” simply means I believe there’s a 90% chance of a finish (ignoring any betting strategies I may use).

Here’s a quick use of the Predict-Wager System to evaluate Sadollah-Baroni:

Round 1, Minute 1: Baroni comes out strong, swinging wildly. It’s not extremely threatening, but punches are landing hard and one may find its mark. I wager $20 on Baroni.

For the rest of the round, nothing threatening occurs.

Round 2, Minute 3: Sadalloh is opening up on a gassed Baroni with knees, elbows, and kicks; the end seems near. I wager $30 on Sadollah.

Round 3: Baroni seems ready to go at various moments throughout this round: I wager, at different periods, in succession, $40, $50, and $30.

The fight ends. I have wagered a total of $150 on Sadollah and only $20 on Baroni. My vote goes to Sadollah, as do those of the other judges.

Sadollah, however, would likely have finished the fight if he’d had the greater incentives to do so that the Predict-Wager System provides. He was certainly capable of doing so.

In addition to eliminating poor judging and boring fights, the Predict-Wager System will also provide another means of extreme profit: gambling. Judging under the Predict-Wager System becomes a skill-based game of chance, more skill-based than hold ‘em, more entertaining than horse racing. Casinos will hold Predict-Wager tournaments on the nights of fights, and champions will become eligible for a judging spot in the big show.

Unfortunately, the current big show will never use such a system. It is regulated by a bureaucracy, and bureaucracies have two characteristics: 1) they are resistant to change and 2) they abhor accountability. The Predict-Wager System would allow instant evaluation of the competence of the judges. Worse still, it would allow instant evaluation of the competence the one who selects them (Kizer).

Any promoter who does realize the profitability of the Predict-Wager System may contact me for clarification of details or the initial code for the wagering software (contact details in profile).

But in the end, the Rebirth requires far more than simply a new judging system. The Predict-Wager System is but a small part of an entirely new strategy; a business plan and a sport as distinct from the UFC as muay thai is from kickboxing.

This plan will be fully described in my next post.


  1. A brief suggestion:

    Instead of having an unlimited number of chips, have a fixed number of chips per judge. Bets gradually fade in value over time, with the portion that fades returned to the appropriate judge's bankroll in real time.

  2. Excellent observation. That would simplify things considerably... and I will certainly make that revision if I ever get this system in place somewhere.