CageIQ Who wins? Beat the AI
Charles Radtke vs Stephen Thompson prediction
UFC 334 · Saturday, November 14, 2026. Our model gives Charles Radtke a 54% chance to win.
Breakdown
Radtke and Thompson are close to opposites in how they work. Thompson is the volume striker, landing 4.01 significant strikes a minute to Radtke's 2.66, and he has three inches of height and reach on him. Radtke is the steadier all-rounder. He absorbs less (2.62 a minute vs 3.01), shoots 1.85 takedowns per 15 minutes at 41%, and holds control for 47% of his fight time. Thompson rarely wrestles and has defended only 63% of takedowns against Radtke's 75%. Both men have been knocked out twice, and Radtke's finish rate is higher (39% vs 27%).
Our model leans Radtke at 55%, mostly because of wrestling and control. His takedown game and 75% takedown defence give him a way to neutralise Thompson's range, and his 67% smoothed win rate and two-fight win streak beat Thompson's three-fight skid. Thompson's higher Elo (1575 vs 1545) keeps it close. The market sees it the other way, with Thompson at 54%, a nine-point gap. The likely reason is that Thompson's 22 UFC fights against top competition carry weight the model may undervalue. The model includes layoff and age, but may underweight Thompson's 490 days out and age of 43.8, as well as Radtke's age of 36.4 and 217 days off.
Key question: can Radtke get Thompson to the mat often enough to blunt his volume, or does Thompson's range keep the fight standing?
Written by AI from the numbers and the scouting notes below. It doesn't make predictions of its own.
How it ends
Most likely result: Stephen Thompson by KO/TKO (25%).
KO / TKO49%
Radtke 24%Thompson 25%
Submission12%
Radtke 9%Thompson 3%
Decision38%
Radtke 21%Thompson 18%
Win chance × how that fighter tends to win, against how the other tends to lose (knockdowns, submission attempts, takedowns vs. strikes absorbed and times stopped), plus the division. Tested on 2,506 fights (2022–2026): more accurate than division averages in 5 of 5 years.
What the market says
| Radtke | Thompson | |
|---|---|---|
| CageIQ model | 54% | 46% |
| Fans | ||
| Sportsbooks (median, no vig) | 46% | 54% |
| Kalshi$2k traded | 48% | 52% |
| Polymarket$751 traded | 50% | 51% |
Across 3 sportsbooks, Radtke is priced between 46% and 47%.
When our model picks a different winner than the opening line, the line has moved our way 62% of the time and away 31% (659 UFC fights, 2022–2026). It's not a betting edge: by fight night the price has usually caught up. How we measure it.
Where to bet
Best price right now at sportsbooks licensed in the US, UK or EU (as of Nov 7, 18:54 UTC). Prices move, so check before you bet.
Radtke2.02 (+102)at FanDuel · US →Typical price 2.03.
Thompson1.77 (−130)at FanDuel · US →Typical price 1.77.
All 3 sportsbooks
| Sportsbook | Licensed in | Radtke | Thompson |
|---|---|---|---|
| FanDuel | US | 2.02 | 1.77 |
| BetOnline | — | 2.11 | 1.77 |
| Marathonbet | — | 2.03 | 1.70 |
Books without a link aren't licensed in a major regulated market, so we don't link to them or count them as the best price.
Plain links: no sportsbook pays us. Bet only where it's legal for you, 18+ (21+ in most US states). Our win chances aren't betting advice: the betting market picks winners better than our model (see our record).
Tale of the tape (UFC fights only)
| Charles Radtke |  Stephen Thompson | |
|---|---|---|
| Last 5 (newest first) | WWLWL | LLLWL |
| Record outside the UFC | — | 5-0, 3 finishes |
| UFC record | 5-2 | 12-9-1 |
| Rating (Elo) | 1545 | 1575 |
| Sig. strikes landed / min | 2.66 | 4.01 |
| Sig. strikes absorbed / min | 2.62 | 3.02 |
| Striking accuracy | 52% | 46% |
| Striking defence | 55% | 55% |
| Takedowns / 15 min | 1.85 | 0.28 |
| Takedown defence | 75% | 64% |
| Submission attempts / 15 min | 0.50 | 0.02 |
| Knockdowns / 15 min | 0.45 | 0.46 |
| Finish rate | 39% | 27% |
| Streak | W2 | L3 |
| Days since last fight | 217 | 490 |
| Age | 36 | 44 |
| Height | 5'9" | 6'0" |
| Reach | 72" | 75" |
Rates for fighters with few UFC fights are blended toward league averages. The pick is logged with a timestamp before the event and graded on the track record, win or lose.