UFC 285 Breakdown

Predictions Overview:

To level set before diving into the predictions, the odds model and plain model predictions from WolfTicketsAI this week are:

Given that it is a launch week for the blog they have all been provided, going forward they may just cover what is being explored.

For this initial post, I'll dig into just one prediction: Jones vs Gane.

Jones vs. Gane

Before writing more seriously, let me get my personal opinions out of the way first. I am not a fan of Jon Jones, his behavior in and out of the Octogon for years has solidified this opinion.

Despite my own biases, both models are predicting a Jones victory and with his previous record at light heavyweight(LHW) it is not hard to see why. However, the fight isn't going to be at light heavyweight, it is going heavyweight(HW). Going up in weight can often be a challenge, below I'll explore some interesting stats that could be relevant to how this fight plays out.

Initial Stats

Here is a full breakdown of how WolfTicketsAI sees them relative to their HW peers:

Weight Class Averages is the result of collecting all the most recent stats of fighters who had at least one fight at the weight class in the last 3 years.

Something to note here THESE STATS DO NOT APPLY FOR JONES AT HW He's never fought there before, so we cannot yet know how he might perform.

What we can say though is that his initial height and reach put him at a slightly above average initial frame for the weight class, and he will be tested to see how additional muscle and weight enables him to perform against larger opponents.

Trends

Jon has always been known for someone who would deliver a pace but he has been accused of slowing down recently, the question is can we actually see that over his career?

WolfTicketsAI leverages stats from 2014 onward as that is when MMA started to mostly resemble its modern incarnation. For his career of 2014 to 2023, his striking volume per minute looks like this:

That's a bit of a drop but nothing drastic, but I also can reflect these stats to be a record of his performance at that time, relative to his peak, to help show relative decline:

Note that Jones is trending in 2023 at an all time max of -4% his career peak in terms of strikes per minute, that's not a huge problem but now we should look at Gane in context as well:

This is interesting, there is a larger drop from Gane over his career, actually dropping by more than 10% of his peak recently from fight with Ngannou.(You can ignore the big drop at the start, it is an issue with my skills at matplotlib and transforming dataframes, I'll get better.)

Percentages and ratios are nice, but raw numbers are important for fights, speed is binary:

Gane is delivering a matching pase on total strikes and a higher pace on significant strikes, and he's been doing this at Heavyweight already. This leads to the most important question here:

Can Jones Keep His Pace at Heavyweight?

I really have no idea here, my heart, brain, and gut say no, his last performances were lackluster, but he's had time to train and plan his climb to heavyweight and we should get our answer soon.

Other Journeys to Heavyweight

Below is a table of every fighter at Light Heavyweight who went up to Heavyweight from 2014 to present:

When breaking this down into percentages, 38.46% won and 61.54% lost.

In this list, the only example against top tier competition was Daniel Cormier going against Stipe Miocic and winning in 2018. Interestingly enough, DC lost both rematches with Stipe, ending their trilogy with a 1-2 record.

This view of how fighters have done going up in weight for a particular weight class would definitely be useful as context to temper model expectations in the future, especially for lower weight classes where the gap is not so large in terms of total pounds.

Final Verdict

The models are always going to have blind spots and this is certainly the case: * The model does not know that the stats were computed for a lower weight class. * I do not have a great way of skewing stats for a fighter based on weight class changes. * Jon fought mostly at a different era in MMA without many of his peers being well rounded as modern fighters. The stats again can be misleading if we assume they will stay the same against Gane.

I look forward to seeing how this fight plays out but this bout has given presented a platform to dig into some healthy skepticism of the models and has added to my notes for future improvements in later generations.

DO NOT BET ACCORDING TO THE MODEL ON THIS FIGHT.

Personally, I may place a small wager on Gane despite the predictions that are posted, if I do it will exist for all in my bet tracking later.

What's Next?

These kinds of articles are designed to help me with my thinking on understanding the predictions of the models and to better guide decisions into WolfTicketsAI itself. Once understood they can be automated into the site and available on demand for all future events, to stay up to date on everything that is available and to get more insightful data on each prediction, start your free trial at https://wolftickets.ai.

Thanks

— Chris