What is this?

The OV Rating is our estimate of a team’s overall strength, expressed in points relative to the average FBS team. It is an aggregate view built to make sense of preseason win-total markets, schedules, and the other information incorporated into our model. The OV Line asks a narrower question: given what we know about these two teams, what would their matchup-specific spread look like without using the market line for this particular game? It accounts for things like location and the way very large rating differences translate into actual game spreads. That is why the OV Line will not always be a simple subtraction of two ratings. The rating describes the team. The line describes the matchup.

Because football—and the betting market—does not behave quite that neatly. Subtracting the teams’ ratings and adding home-field advantage will usually get you into the neighborhood. But our ratings are designed to represent aggregate team strength, and very large differences between teams do not translate perfectly into game spreads on a one-for-one basis. A 30-point rating difference, for example, may imply something different at the game level than a three-point rating difference. We tested how rating gaps historically translated into real betting lines and use that relationship when presenting the OV Line. So if your arithmetic does not exactly match the displayed line, you have not forgotten how subtraction works. The OV Line is a matchup-specific model estimate, not just ratings gap plus HFA.

Fair question. We could combine every Factor estimate, replace the OV Line with one adjusted number, and ask you to treat that as the answer. But that would hide the most useful part of what we do. Factor estimates are historical averages with uncertainty. Factors can overlap, and the circumstances of one quarterback, weather system, coaching change, or matchup may not deserve exactly the historical average. You may reasonably weigh that evidence differently than we would. So we preserve the OV Line as the baseline before any Factor adjustments, then show the Factors, their estimated effects, the resulting OV Factor-Adjusted Suggested Line, and anything the market still appears to account for that we cannot. Our job is to make the evidence visible and understandable. Your job—should you choose to accept it—is to decide what belongs in your own line.

You can—but we would not tell you that you have created a better prediction by doing it. We provide a rough combined Factor estimate because we know users will reasonably want to understand their total direction and magnitude. If one Factor points three points toward one team and another points one point toward the opponent, it is useful to know that they roughly net to two points. But Factors can overlap. Their historical estimates also contain uncertainty, and we almost certainly have not identified every relevant piece of information. The combined number is best understood as an estimate of how much of the OV-to-market difference might be explainable—not as a new betting line wearing a fake mustache.

It sounds counterintuitive, we know, but there is a reasonable explanation: the two estimates describe different stages of the same market. Remember what Factors are trying to measure—the game-level points that may help explain the difference between the OV Rating-based spread and the market consensus. Before betting opens, sportsbooks use their information—which is almost certainly more complete than ours—to set an opening line that reflects their view of the game, expected customer demand, and an acceptable level of risk. That is the opening effect. After the line opens, the market continues absorbing bets, new reporting, injuries, weather, and the opinions of bettors whom sportsbooks respect. The net change after the opener is the in-week effect. The same underlying condition can matter at both stages without being counted twice: we estimate what was reflected at open, then what changed afterward. We show the effects separately because sportsbooks and the later betting market do not necessarily price the same information in exactly the same way. As always, we give you the evidence and let you decide what story it tells.

Because we do not know everything. We’re not the kind of people who would have known that Blue Horseshoe loves Anacott Steel. The market may be responding to injuries, personnel decisions, matchup details, weather, informed bettors, or other information we have not yet identified or cannot reliably measure. Some Factors may also be more or less important in a particular game than their historical average suggests. And even when multiple Factors are relevant, they may overlap rather than operate independently. An unexplained remainder is not a bug we intend to hide. It is information. It tells you that the Factors we have identified do not fully account for the market’s position—and that you should decide how much confidence to place in the available explanation.

First, we’re trying! OnVarsity is a very small team, and we would rather show no Factor than give you bad information. We are compiling as much usable data as we can, then testing the conditions most likely to have a meaningful game-level effect. Coverage will grow as the evidence does.

It only describes direction. If the OV Line is Michigan −10 and the market is Michigan −7, the market has moved three points toward Michigan’s opponent relative to OV. A Factor that historically points toward the opponent is therefore aligned with market. A Factor pointing toward Michigan is counter to market. Neither label tells you which side is correct. “Aligned with market” does not mean “good,” and “counter to market” does not mean “wrong.” We use the labels to help you understand whether a Factor could explain the observed difference—not to quietly turn the page into a list of picks.

The most honest answer is that we do not always know. A Factor can show a meaningful historical association and still have a wide range of outcomes; a more granular version may eventually explain that variation. The market may also be responding to information we have not captured—or pricing something that history does not support. We show counter-market Factors because hiding them would create false confidence.

In the Ledger. Every live Factor should link to the study supporting it. The Ledger shows what we tested, the historical sample, the estimated effect, the uncertainty and limitations, and—when available—the individual games behind the result. We also include studies that did not support a separate adjustment. That matters because conventional football wisdom contains plenty of ideas that sound obviously true until someone has the poor judgment to test them. You should not have to trust a number simply because we placed it inside a polished box. Follow the link, inspect the evidence, and decide whether the historical comparison makes sense for the game in front of you.

We assume so. But can we prove it to you? OnVarsity is not saying that revenge games or look-ahead games are imaginary. Our claim is much narrower: in the historical data we studied, those labels did not reliably explain an additional game-level difference between the OV Line and the market spread.

Think of it this way: everyone can identify a revenge game before the season. Bettors can account for that matchup when pricing a team’s win total, and that preseason market information helps shape the OV Rating. By the time the game arrives, the narrative’s value may already be embedded in the aggregate market view. So no, you are not necessarily wrong that it matters. Our evidence suggests it may not matter again, in the specific incremental way people often assume.

We try to recreate what could reasonably have been known before kickoff. For an in-season statistic, that means using information from games completed before the week being predicted—not statistics produced by the game we are trying to explain. For injuries, quarterback changes, coaching situations, and similar Factors, we use or reconstruct the information the market could reasonably have had before the game. Most importantly, the OV Line shown for a matchup does not use that game’s market line. If we included it, the model could simply absorb the very difference we are trying to understand. That would make the page look impressively accurate while teaching you almost nothing, which is not our idea of transparency. Historical reconstruction is not perfect, especially when older reporting is incomplete. When we cannot establish what was known or when it became known, that uncertainty belongs in the Ledger rather than underneath the rug.

We probably don’t like them. We like only one team; everyone else is a sad facsimile of a college football program. But more importantly, we don’t think anything. You do—or, more precisely, you and everyone else whose opinions help shape betting markets. The OV Ratings reconstruct that collective view from win totals and schedules. We just tell you what you seem to think, in aggregate. We know you better than you know yourself.

That’s why we built several ways for you to tell us exactly what you think. Head to My Ratings and choose the better team in synthetic matchups, or make picks on real games. Keep telling us who you think is best, and we’ll use nearly the same algorithm that calculates the OV Rating to build your personal ratings.

We hope not! You seem nice. You’re reading this, so you’re definitely someone we would almost probably talk to—and we try not to do that too often. But listen: you don’t have to believe us. Do you believe in yourself enough to take this information and decide what to do with it? If not, start with one of our primers. We promise it will be gentle.

We’re sorry, your call cannot be completed as dialed.

Why are you trying to make us feel feelings again?