The myth of the ATGM reroll

You’ll commonly hear from people that ATGMs have lower than the displayed accuracy due to rerolls. Apparently during ALB ATGM was far stronger, then in some WRD patch they were nerfed, and one of the nerfs was to add these rerolls. I’ve seen different explanations for the specifics of the bug/mechanic, but the most common one by far is this:

The ATGM rolls for accuracy periodically, say every two seconds. If it fails the roll, it hits the ground or flies off. If it passes, it keeps flying until it reaches the target or it has to roll again.

If this explanation were true, this would mean that slow ATGMs are much less accurate than described. For example if a TOW 2 has to roll three times, it would have not 70% accuracy but 70%^3 = 24%. This is huge.


But if you actually go and test the ATGMs, you’ll find that their accuracies are more or less the same as what the game says above the weapon [and the numbers above the weapon are what you’d expect given the unit card and veterancy]. It’s a really easy thing to test – the difference between 70% and 25% accuracy will be clear as day within ten shots. Even if a missile only rerolls once it would be pretty clear if Eugen are trying to scam us.


So if testing shows the accuracies are as described, why does the myth persist?

One reason is that rerolls do exist. There’s a variable in the mod tools for them and it’s been confirmed by Eugen that a reroll mechanic is in the game, albeit one that doesn’t influence accuracy. In fact, the ATGM reroll bug may have existed in exactly the way the myth describes it for some time in the WRD beta, or at least a similar bug was in effect. Regardless, the intended role of rerolls is not as an arcane nerf. They’re meant to allow for missile accuracy to change if the shooter gets shot at and demoralized while the missile is in flight. More on this at the end of the post, from Eugen themselves.

Another reason is that the myth gives us a good “just-so” story for why slow ATGMs suck. If you know from experience that slow ATGMs are pretty bad, hearing a theory whose consequences are that slow ATGMs are really inaccurate would line up perfectly with your experience. But slow ATGMs are bad because it’s much more likely for the launch vehicle to get killed or panicked, or for the target to reach concealment, not because their accuracy is fake.

A third reason is that people don’t test things. Which is fine, nobody has the time for it. People who do find the time once in a while should share their results. But I didn’t have a blog when I first heard about this, more than two years ago.


Finally, here are some quotes from FLX, an Eugen employee:

You don’t have to get series of hit to actually it the target because if you roll a hit after having rolled a series of miss the missile will fly toward its target and hit it.

I change single roll to multiple rolls for different reasons :
– you cannot keep your static accruacy if you fire then move
– you can be stressed during missile flight
– you cannot predict if a missile will hit or miss from the start
– visually it’s rather cool

Compared to WEE, missiles cannot crash into the ground if a dice roll is a miss, only out of control missiles can hit hte ground.
To keep an accurate hit probability with multiple rolls you need as many roll as possible so I changed the check interval from 20 or so seconds ( = longer then the missile flight time) to 0.2 which is the maximum that the game authorize.
It works perfectly but there is a rule I misunderstood which says :
“If a hitroll is a miss in the final approach of the missile (calculated in 3D frames) then the missile is deflected to prevent accidental hit”

[Note from honhonhonhon – the rest of the quote describes a bug that has since been fixed, and may be what spawned the myth. It’s an old quote that I included because it describes the motivations behind the reroll system. Don’t get confused.]

What I didn’t know is that if you have multiple dice rolls in the final approach the missile is deflected anyway even if you have successes after an initial miss.

So with 0.2sec between rolls slow missiles could have 2 or 3 rolls during final approach which means you had to make a series of sucess to avoid being deflected.

Let’s take an exemple :
– Final approach distance is 500m
– Dice roll occurs every 0.2sec
– missile speed is 500m/s

hit hit hit hit miss = your missile miss it’s normal
miss hit hit hit hit = your missile is deflected anyway even if you have had 4 success after a miss.

So the only way to get a hit in my example is to make 5 hit in a row.

The faster the missile the less dice roll in the final approach, the more accurate the missile the more chance to make a series of success.

You don’t have to get series of hit to actually it the target because if you roll a hit after having rolled a series of miss the missile will fly toward its target and hit it.

In WEE the accuracy of ATGM was wrong because when you rolled a miss the missile got deviated in random direction and if it hit the ground it was destroyed.
Since there is 50% chance that the deviation is toward the sky and 50% toward the ground the practical accuracy was lower than expected.
With WALB we brought a new feature that prevents a missile from hitting the ground when the roll is a miss befor it hits or overfly its target which brings back the accuracy to what is displayed.

That’s the theory. In practice some elements can make your accuracy lower than displayed :
1) You get stunned during missile fly = you lose control
2) Target disappear for more than one second during missile fly = you lose control
3) You get stressed during missile fly = your next roll will have lower hit probability
4) Your missile deviate so much from its trajectory after multiple miss that when the last roll is a hit the missile cannot turn enough in the few meters remaining to actually hit its target.

Point 1 2 3 are designed that way and controlled.
The number of time point 4 appear is much harder to calculate. Your missile have to score multiple miss in a row and the random deviation must be several time in the same direction. Then the last roll that is a hit has to occurs close enough to the target and the missile rotation speed has to be low enough to prevent the missile from hitting the target hitbox.

To minimise this case the faster the missile the smaller the maximum deviation angle of the missile is.

[ Source: , pages 2 and 3, ]


Actually, one more thing. FLX’s explanation does leave some room for unexpected misses. As he says, if a missile rolls multiple misses in the same direction, it may stray too far from the target and be unable to turn fast enough to perform a hit rolled at the end. But this is very unlikely. I believe a forum member detected the effect after a lot of testing, and it was as expected within the single digits, but I can’t find the thread [and it is as of yet unreplicated, because nobody has tried it].