Who's Actually Sharp on MLB Mainlines After CFTC Recognition?

On June 16, 2026, Novig received CFTC recognition as a true prediction market. ProphetX got there a few days earlier. We're treating June 16 as the split point for both, since that's the date both were operating under the same CFTC recognition Kalshi and Polymarket already had.

Quick context on why that actually matters: CFTC recognition means Novig and ProphetX are now regulated the same way Kalshi and Polymarket already were, as real, federally registered prediction markets instead of sitting in a legal gray area. Practically, that means they can operate nationwide instead of being locked out of most states, which opens the door to a lot more capital and a lot more serious traders showing up. More liquidity generally means sharper pricing, since it's real money competing to find mispriced lines that keeps a market honest in the first place.

So now that all four are regulated the same way, how do they actually stack up on MLB mainline markets? And which book is everyone else still chasing?

This also lands at a good time to ask. The season is wrapping up, which means playoff baseball, fewer games, a lot more eyes on each one, and real money moving accordingly.

We pulled line by line movement data across ten books for the 2026 MLB season and tracked, market by market, who moves first when real information hits and who's just following. This isn't a study of closing lines. It's a study of behavior. Who leads, who copies, who's noise.

The books, and why we picked them:

• Kalshi and Polymarket. The two names most people would call the gold standard of prediction markets right now. If anyone should be sharp on MLB by reputation alone, it's these two.
• Novig and ProphetX. The new arrivals with fresh CFTC recognition. The whole point of this study.
• Pinnacle. The gold standard for sharp lines for about as long as anyone's been tracking sharp lines. The book everyone else gets measured against.
• 4Caster. A global crypto exchange with real liquidity behind it, and a useful third reference point outside the US regulatory framework.
• FanDuel. The best regarded recreational sportsbook in the US, included as our benchmark for what a well run book looks like when it isn't a true sharp reference.
• BetOnline. One of the most widely used offshore books, and a fair stand in for the broader offshore market.
• DraftKings and Fanatics. Two of the most popular recreational sportsbooks in the country. Included as a sanity check, if this method is actually working, these should read as clearly soft, the same way most bettors would already expect.

Ten books, three market types (moneyline, run line, and total runs, each with 1st half and 1st inning variants where that market exists), across our post-ruling window, June 30 through August 7.

How We Got Here

Our first instinct was to run the same playbook that's worked well for us on player props. Find crossed markets, meaning moments where two books disagree enough that betting both sides locks in value, and track which side of that disagreement actually wins over time. On player props, across a huge sample, that method works well. On mainlines it didn't hold up the same way. Player props swing around more and jump bigger than mainlines do. Mainlines are steadier by nature, so they cross less often, and the sample sizes we were working with just weren't big enough to separate real signal from noise. A handful of results one way or the other was swinging entire conclusions.

That pushed us toward a different approach. Instead of grading against final outcomes, which introduces a lot of variance you have to out-sample, we started grading against closing line value, meaning where the market actually settled. CLV is a more stable signal in smaller samples because you're not waiting on a coin flip to resolve. You're just asking whether a book's price moved toward or away from where the market eventually landed.

That in turn led to the framework we actually used for this study. Instead of asking whether a book's price was good at any single moment, we asked who moves the market and who follows them. Every time a book's price breaks away from where the rest of the field is sitting by a meaningful amount, that's an event. From there we look at two things. Does the rest of the market actually come around to that number, and if so, how fast does each other book get there? A book that moves and gets followed is setting the market. A book that moves and gets ignored is noise. A book that's consistently late to react is exactly where the value sits.

One clarification worth making up front, since it changes how to read some of the numbers below: an "event" is a gap opening up between a book's price and the rest of the field, not necessarily that book making a sudden jump. Pinnacle in particular almost never jumps. Its own price barely moves over any given half hour, the single biggest move we found all season was under 5 points. So when Pinnacle's price shows up as "diverging," it's usually not Pinnacle spiking, it's the rest of the market drifting away from a Pinnacle number that's stayed put, and then drifting back. That matters most for the follower table further down.

Who Leads, Who Follows

Here's the moneyline picture across our post-ruling window, roughly 39 days of MLB, June 30 through August 7. "Avg gap" is the average size of the move away from market consensus, given that the price already moved at least 2% to qualify as an event in the first place.

Pinnacle didn't need to be told it's the sharp reference point here. It fell out of the data on its own. It doesn't jump as often as other books do, but when it does move, the rest of the market comes around to its number 83% of the time. That's about as clean a definition of the book that sets the price as you'll find.

Kalshi sits at the opposite end, and not in a subtle way. It moves more than three times as often as anyone else in the study, with bigger average gaps too, and only about one in nine of those moves actually holds up. We'll get into exactly why below, because it's the most interesting result in this whole study, and it's more complicated than a simple "Kalshi is noisy" headline.

Novig and ProphetX land in a real, credible middle tier. Meaningfully behind Pinnacle, but clearly ahead of the noise. That's a legitimate marker of progress for two books that didn't exist as regulated prediction markets a few months ago. DraftKings and Fanatics, our two soft-book sanity checks, land right about where you'd expect a recreational book to land, well behind the real sharp reference points.

Why Does Kalshi Miss So Often?

This is worth pausing on, because on paper it doesn't make sense. Kalshi is arguably the most liquid, most talked about prediction market in the country right now. Moneylines are about as simple a market as exists. It shouldn't be the noisiest book in the study.

We checked the obvious explanations first. It's not a timing issue. Kalshi's moves happen at roughly the same point in the pregame window as everyone else's. It's not concentrated on lopsided lines where a small dollar move swings the price a lot either. That effect is there, but it's minor.

The real answer is simpler than any of that, and it's sharper than we first put it. The median Kalshi move that gets flagged reverses in 35.8 seconds. Not 15 minutes, 35 seconds. Over half are gone within a minute, and about one in eight are gone within 5 seconds. We pulled the raw prints behind these to see it directly: a Kalshi moneyline goes from 1.825 to 1.177 and back to 1.858 inside a 3 second window, backed by a fraction of the size sitting on the ticks around it. That's a thin order book throwing off a bad print and correcting itself almost instantly, not a price sitting exposed. Compare that to Pinnacle, where a flagged move only ever reverses at all 1.6% of the time, and when it does, it takes a median of 20 minutes, which looks like a deliberate correction rather than noise.

One theory worth checking before writing this off as pure noise: what if some of these are real, someone quietly getting a large position down before the market catches on, and the fast reversion is just the surface while the real move plays out over the full pregame window? We tested it directly. Looking only at Kalshi's biggest swings, 5% or larger, the ones that revert within 15 minutes validate by close at basically the same rate as the ones that hold (11.3% versus 10.9%), and the market doesn't drift toward either group afterward. If anything it drifts slightly away from both. That rules out the insider theory. These are just noisy prints clearing against thin size, not someone moving real information through the back door.

There's also a real structural wrinkle specific to Kalshi that's worth being upfront about. Kalshi doesn't run a single moneyline market the way a sportsbook does. It lists each team's contract separately, "will the home team win" and "will the away team win" as two independently traded markets, rather than two sides of one line. On top of that, those two sides aren't really symmetric either, most recreational money piles straight into the YES side of whichever team a bettor likes, while market makers tend to work the NO side with resting limit orders, so even when we match the right YES to the right NO, we're comparing retail-driven pricing on one side against market-maker pricing on the other, not two evenly balanced sides of the same bet. Our pipeline treats those as a matched pair, which is the right way to read them, but it also means the data is more exposed to a specific kind of noise: when a market maker's liquidity briefly pulls out of one side, usually right as a line first gets listed or right around a news-driven repricing, the price can walk through several stale, barely-sized levels in a fraction of a second before the book refills. We found this directly in the raw prints, both sides of a game spiking toward 99% at once for a few hundred milliseconds before snapping back to normal, which isn't a real market state, just a brief empty book. It's the same underlying noise as the 35.8-second reversion story above, just showing up in a different measurement. We're comfortable saying Kalshi's own moves are genuinely noisy. We're less comfortable putting a precise number on exactly how noisy, given how much of the raw data is these brief liquidity gaps rather than real pricing.

It's also worth remembering that moneyline is probably the single most bet on, most liquid MLB market there is. With that much attention on it, it makes sense that most of these books land somewhere in the middle of the pack together. What stands out is that Pinnacle is clearly ahead of that pack anyway.

It's Not Just Moneyline, and FanDuel Splits in an Interesting Way

The same test run across run lines, totals, and the 1st half/1st inning variants of each tells largely the same story. Pinnacle near the top everywhere it shows up, Kalshi noisy everywhere it shows up, for the same reasons above. But one result stands out that you wouldn't see from moneyline alone.

FanDuel behaves like two different books depending on the market. On moneyline it's a credible mid tier performer. But on run line specifically it has more total events than anyone in the study except Kalshi, and validates barely a quarter of the time. Whatever FanDuel is doing well on moneyline, it isn't carrying over to spreads. That actually tracks with what we've seen using FanDuel ourselves. NRFI/YRFI, first 5 totals, and run lines are consistently where we've found real value on FanDuel specifically.

Pulling every market together into one composite ranking:

4Caster is worth calling out here. Thinner sample than Pinnacle overall, but remarkably consistent wherever it shows up, edging Pinnacle out on average rank even though Pinnacle wins on raw volume weighted accuracy. That also matches what we've noticed using SmartStake day to day. 4Caster tends to track Pinnacle closely.

Pinnacle's composite number is also worth explaining rather than just reporting. Pinnacle sets bet limits per market, and those limits vary a lot. Moneyline might carry a $50,000 limit while a market like first half total might only carry $1,000 to $2,000. The markets they're willing to take real money on are the ones they've put real modeling behind, and an 83.3% moneyline validation rate against a 50.3% average across every market makes a lot more sense once you know that. Their biggest, highest limit lines are still the gold standard across sports. The smaller ones just don't get the same attention internally. That said, it's a good sign that Pinnacle still comes out on top even in those lower-liquidity, lower-limit markets, despite getting less internal attention.

DraftKings and Fanatics both land in the same rough neighborhood as FanDuel here, solidly behind the real sharp tier, which is exactly what you'd want to see from a sanity check. If two books everyone already knows are soft came back looking sharp, that would be a real problem with the method. They didn't.

"But Kalshi and Polymarket Are the Most Liquid Prediction Markets"

Fair pushback, and worth addressing directly instead of ignoring it. Kalshi and Polymarket are the most talked about prediction markets in the country, with real, serious capital behind them. Most casual bettors have never heard of Novig or ProphetX. So why would the smaller names be sharper on MLB specifically?

What we found is that Polymarket's noise really does track with public attention. On the most popular games, the ones drawing the most action across the board, Polymarket's own quote activity jumps 14 times versus a quiet weekday game, and its price gets choppier right along with it. That's a textbook signature of casual money piling into the games everyone's watching.

Kalshi doesn't show that pattern at all. Its choppiness is exactly the same on an empty Tuesday afternoon game as it is on a nationally televised one. That's worth sitting with given how Kalshi and Polymarket actually work. Neither one sets a line the way a sportsbook does. Both are peer to peer markets, so the price is really just wherever the last trade cleared. Kalshi not caring how popular a game is tells us its swings aren't coming from casual bettors piling in. It looks more like the liquidity mechanics described above, thin resting size and market makers stepping away for a moment, clearing against whatever's left on the baseball side of the exchange specifically, rather than genuine repricing from real information.

(One more thing worth stating plainly: all of the numbers above are de-vigged before comparison, so spread and fees are not a factor in any of this. This is purely about how each book's price moves relative to the field.)

Where the Actual Value Is

Leadership tables are interesting. This is the part you can use. Restricting to Pinnacle's real, validated moves, here's how often each other book actually catches up to Pinnacle's real number, landing within 1 percentage point of its de-vigged price, not just making some move of its own, and how long that takes:

Worth reading the "minutes" column with the earlier point in mind: since Pinnacle rarely jumps, "catching up" here usually doesn't mean chasing a target that's actively moving. It usually means a book's own price needs to land within a point of a number Pinnacle set a while ago and hasn't moved much since. So a 20-30 minute lag isn't necessarily a book sitting on an obvious mispricing for half an hour, sometimes it's simply how long it took that book to post its next update at all. That's a real limitation of what these lag numbers can tell you in isolation. What they can still tell you is relative: every book here is measured against the exact same target the exact same way, so who's consistently faster than who is a fair comparison, even if the raw minutes shouldn't be read as "time spent knowingly ignoring value."

This is where Kalshi's story actually turns around. On its own, Kalshi's independent moves are too noisy to trust, for the liquidity reasons above. But measured as a follower, catching up to a real Pinnacle move, Kalshi is genuinely excellent. It confirms 85.5% of validated moneyline moves, right behind ProphetX and Novig, and it does it in a median of 6.5 minutes, by far the fastest of any book in the table. Everyone else takes somewhere between 12 and 34 minutes to get there. Kalshi does it in a fraction of that.

That's a real, important distinction. We're not comfortable putting a precise number on exactly how noisy Kalshi's own price moves are, given how much of that data is contaminated by brief, empty-book liquidity gaps rather than genuine repricing. But we're comfortable saying this: a book doing the volume Kalshi does, reacting to real information this fast and this reliably, is not a soft book. The follower data backs that up cleanly, without any of the same data quality concerns, since it's measuring whether Kalshi's price lands near Pinnacle's, not trying to isolate Kalshi's own noisy prints in the first place. Kalshi earns its reputation as a serious, liquid market. It just isn't the one setting the price on MLB mainlines, it's the one reacting to it fastest.

FanDuel's run line result is its own small surprise. It's the noisiest initiator of run line moves in the whole study, but once Pinnacle actually moves, FanDuel is the fastest book in the entire table to land near the new number, under a minute on average. It's a bad leader and a great follower on that specific market, which is a genuinely different problem than being slow across the board, and it's a real source of value if you know to look for it.

BetOnline sits at the bottom of this table most consistently, and DraftKings and Fanatics show up there too on individual markets, DraftKings on run line, Fanatics on moneyline. Those are the spots where a real Pinnacle move sits unmatched the longest, which is where an actual edge lives if you're looking for one. That tracks with what we've seen firsthand using these books ourselves, BetOnline, Fanatics, and DraftKings all tend to hang stale lines longer than the rest of this panel.

The Closing Line

Pinnacle is still the book to set your reference price from. That much hasn't changed. Novig and ProphetX have real, legitimate footing now as credible mid tier books, and once you measure it as catching up to Pinnacle's actual number rather than making some equally large move of their own, both of them hold up well, right alongside Kalshi. That's a real step up from where the market perceived them before regulation.

Kalshi's reputation as the most liquid prediction market in the country is earned, just not in the way you'd first assume. Its own independent moves are too noisy to trust, mostly a product of thin liquidity briefly stepping out of the market rather than real information. But as a follower, it's arguably the best in the whole study, fastest by a wide margin and right up with the top confirm rates. Treat its own moves with real skepticism, and lean on it instead for how fast and how reliably it reacts to everyone else.

FanDuel is worth watching market by market rather than as a single number. Noisy as a mover on run line, but the fastest follower in the whole study on that same market, and a real source of value if you're paying attention to where it shows up.

If you're looking for where the actual value sits, BetOnline is consistently where a real Pinnacle move goes unmatched the longest, and FanDuel deserves a mention too, its run line follow speed alone is worth knowing about even though it's a rough mover in its own right. DraftKings and Fanatics belong in that same conversation, both are consistently among the last to catch up across this study.

One more thing worth keeping in mind as the season wraps up: Pinnacle's limits on MLB mainlines are typically a lot higher for playoff games than they are for a random Tuesday in July. Fewer games, way more attention on each one, and more capital willing to test their number. If their limits climb the way they usually do, Pinnacle's playoff pricing should be even sharper and more trustworthy than what we measured across the regular season here. That makes the gap between Pinnacle and the rest of this list more useful, not less, right when it matters most. If BetOnline, Fanatics, and DraftKings are already hanging stale lines longer than everyone else during the regular season, that's exactly the kind of gap worth watching even closer in the playoffs.

Prefer to test all of this yourself? The full dataset behind this study is public on Hugging Face, so you can run your own version of this analysis from the raw line movements.