NFL research in the ET Parlays Lab

ET Parlays grades NFL player props and shows the reasoning behind every grade. It is a research tool, not a sportsbook: it never takes a bet, holds no money, and picks nothing for you. NFL research is included with ET Plus. This page is the whole inventory: what the NFL product does, where the numbers come from, and what we will not claim yet.

What ET does for the NFL

Eleven surfaces. Each one answers a question you would otherwise answer by opening five tabs.

Which NFL markets it grades

Seven player-prop markets, each read as a probability first and a letter second:

• Passing yards
• Passing touchdowns
• Rushing yards
• Receiving yards
• Receptions
• Rushing plus receiving yards
• Anytime touchdown

Game lines and totals research lives in Team Lab, which is part of ET Pro. ET does not currently rate defensive props such as tackles, and it does not rate kicking.

How much NFL props actually move together, measured on four seasons and 22,256 player-weeks, is in our prop correlation index. How we grade, how much of the board never gets graded, and every place our numbers run hot is written up in full on our methodology page. The NFL record is too young to appear there. The MLB one is not: what ET grades for baseball, market by market, is on the MLB page.

What one of those cards actually holds

Eleven surfaces is a list. The Player Lab card is where most of the work actually sits, so here it is panel by panel.

Behind those tiles is a glossary you never leave the card to read. 24 stats each carry a plain-English definition, the league average, where this player sits against it, and a line on what that stat does to each market. Tap the number and it tells you what it is. Yards per carry says it is rushing yards divided by carries, that the league runs 4.20, and that a back can be efficient and starved at the same time, so read it next to his carry volume. That last part is what most tools leave out, and it is the difference between a stat and a reason.

Underneath all of it is a per-player, per-game table ET builds rather than reads: 35 measurements per player per game, covering how often a player was on the field, how much of his offence ran through him, how far downfield he was used, and how close to the goal line he got the ball. That layer, and the record it produces, is the part worth judging us on.

None of it tells you what to bet. It is assembled so you can disagree with the letter sitting on top of it, which is the only reason to show the working at all.

The case for the number, and the case against it

The Matchup surface does something the rest of the card does not: it takes one projection apart and shows you the argument. Five components, each written as a plain sentence, and each tagged with whether it leans with the number, against it, or does not tilt at all.

Above them sits a verdict on the evidence itself, not on the bet. Strong when several components lean the same way and none lean against. Mixed when they disagree, and it says how many went each way. Thin when the case is built on last season or a handful of games, in which case it tells you to read it as context rather than as a case at all. There is also a state for having nothing to say: when our number sits on the book’s line, the page says so instead of manufacturing a lean.

The sentence at the bottom of that panel is the whole design in nine words: components explain, the ET Read decides. You are meant to be able to look at five rows, disagree with one of them, and discount the number yourself. A product that only hands you a percentage cannot be argued with, which sounds like confidence and is actually just less information.

Role is stable. Production is not.

The reason the card leads with share metrics rather than yards is that they settle at different speeds. How often a team puts a player on the field, and how much of the offence runs through him, is close to a decision the coaching staff makes. What he does with it in one game is mostly variance.

We measured that the hard way. Our receiving projection estimates yards per reception from how deep a player runs rather than from his actual yards per reception, which looks like a shortcut. Tested on the opening weeks of two past seasons, replacing it with the measured number made calibration worse at every setting we tried. Route depth is a stable fact about a role; yards per reception off a handful of early-season catches is mostly noise. In that window the rough stable number beat the exact unstable one.

The same logic is why the opening weeks are the hardest, and why we withhold most of our own board in them. That is written up in why NFL Week 1 and 2 props are unreliable, with the calibration numbers that decided it.

Every reference surface says when not to use it

The four reference tabs all do the same thing structurally: compare a player to a baseline, then tell you when the comparison is not safe to lean on. The guard is the part worth reading.

Each one carries a line telling you how far it can be pushed, and they are the least commercial sentences in the product. The touchdown page says that touchdowns are the noisiest stat in football, that the chances are what persist, and that being ahead of expected is not a fade and being behind is not a buy, because the price already knows the role. The defence page says a leak is a tilt on a player who actually has that role: it nudges his number, it does not make the pick. The role tiles say one game is a blip.

A defence read also refuses to speak for the current season until four games are in, and labels itself last season until then. That is the same policy as the withheld board, applied to a reference table rather than a grade.

One prop is not a slip

Everything above prices a single leg. Put two legs from the same game on one ticket and they stop being independent, which is the thing that quietly decides whether a same-game parlay was ever worth its price.

We measured it on four seasons rather than asserting it. A quarterback going over his passing yards and his WR1 going over his receiving yards land together 1.35 times as often as independence predicts. Two opposing running backs both going over land together 0.855 times as often, the strongest negative in the study. And two of the most repeated beliefs in this category turned out to be nothing at all: two receivers on the same team came out at 0.997, and two touchdown scorers on the same team at 0.994.

One caveat those numbers carry, and it matters on this page in particular. No archive of historical sportsbook lines exists for us to measure against, so every player was judged against his own season median rather than against a posted line. Read them as the direction and rough size of an effect. They are not the multiplier at the book’s number, and they are not the probabilities the reads above are graded on.

The full table, including the cross-game control that shows the method returns 1.00 when nothing is there, is the NFL prop correlation index.

Where the numbers come from

Two different questions, and we answer them differently on purpose.

For grading, we are specific. A read is settled against the week’s final player statistics once the game is final, never against our own estimate of what happened. If the record says a receiver went for 63 yards, the box score says 63 yards, and you can check it in ten seconds. That is the part that decides whether the record below is worth anything, so that is the part we pin down.

For producing a read, we are not. The models run on public league data and a commercial odds feed. We do not itemise them, for the same reason we do not publish the model: naming the ingredients is most of the recipe. Treat that as a real limitation rather than a footnote. You can audit our outputs, not our method, and this page is built so the outputs are enough.

What sits on top is ours: a per-player, per-game view of role and opportunity, and the models that read it. That layer, and the record it produces, is the part worth judging us on.

The record, and how much of it we showed you

Every read is written to the database with its probability, its line and its side before the game starts, so it cannot be edited afterwards. Here is the whole 2026 NFL board so far, read live from that table, weeks 1 through 4.

The second and third numbers are the ones worth reading. In the opening weeks of a season the models lean on usage data this season has not produced yet, so reads that depend on it are held back rather than served. They are still written down before kickoff and still graded, which means we can check ourselves against the reads we did not trust enough to show you. Research is allowed when a recommendation is not.

What we will not claim

Three things, stated plainly so you can hold us to them:

• We have no NFL accuracy figure. The season’s first reads have not been graded. Anyone quoting one for us made it up.
• We do not claim to beat the closing line. We do not store closing prices for NFL props, so we could not check that claim even if we wanted to make it.
• We do not sell picks. ET produces reads. You choose every leg, and you place every bet somewhere else.

Common questions

Try it before you pay for it

The slip checker is free, needs no account, and grades every leg you paste with a letter and the record behind it. It reads MLB props today; NFL legs are graded inside the Lab. If you want the NFL surfaces above, ET Plus covers them.