Shots too random


Some games are pure luck (such as a coin toss), some are pure strategy (such as chess), but there’s also a middle ground. Some games can exist that combine luck and strategy, this is where PP sits. The existence of RNG within PP doesn’t mean that strategy becomes meaningless; a player who plays well will have more occasions where they beat the odds than a player who plays badly.

A player who succeeds despite a lack of ability has been lucky, and a player who fails despite a strong ability has been unlucky. But that’s for just individual situation, over the course of a full campaign the player with ability will win out, whereas the player who lacks ability will need to strengthen.


I didn’t run into this issue in my play throughs so far, as the targeting reticule does a pretty good job of indicating that a shot has a really low chance of connecting… so I just wouldn’t take that kind of shot and instead attempted to move into a better, closer (and if possible higher) position. Meanwhile the AI would try its shot against my guy on its turn, missing wildly and inflicting just 1 point of damage (i.e. practically nothing), which was quite amusing to be honest. So it could be that what’s needed are two things; 1) the interface communicating the shot effectiveness may need to a bit clearer and 2) upping the range of some of the weapons a bit to make them a bit more accurate at long ranges. I prefer option 1 because if a player wants to piss ammo away that won’t hit the target, I say let them. Just like the AI wants to do. It’s a strategy game after all and if a player is using the wrong strategy…


Actually I think the number of aliens is probably the most obvious thing to change for an easier difficulty mode; just take a few of them off to give people who aren’t quite so good at these types of games a fairer chance. Over time they’ll learn and then be able to set the difficulty back up.


The likelihood of 4 misses happening in a row is just the same in the system that you you’re describing as a pure RNG system, and in either system the spread of 4 misses across 16 shots will have a random pattern.

The difference comes in what happens following those 4 misses, with an RNG system, the percentage to hit remains at 75% for every future shot regardless of hits and misses. With the example that you’re giving the percentage will fluctuate. Both systems work fairly because every player is treat the same, but in the example that you’re giving the player who has full knowledge of how that system works will then game the system:

Player A has made a high number of hits in a row, he knows that his next set of shots are more likely to miss than they would normally be, what is he going to do, he isn’t going to take them. More likely he’ll hunker down and shoot at a wall until he’s had enough misses to raise his hit chance again. The opposite is also true, after a number of misses a player will be more likely to then take those unlikely to hit shots because they’ve started becoming likely. You could sit at the back on the map taking starting off taking a 5% shot in the knowledge that eventually that percentage is going to raise to almost certainty.

With an RNG system, there’s strategy because you have to play in such a way that you position yourself in place where you will increase your percentages and minimise risk. That to me is an ideal system, it makes a game fun by creating tension with the possibility that your best laid plans can still go awry. Once you start dealing with certainties and guarantees and loaded systems the game just gets boring from my POV, I find no no point in playing something if I know that there’s 100% certainty of events happening.


Yes, I too said shots were most likely to fall on (or close to) the yellow rim, but that doesn’t mean the probability distribution is denser there, it’s just because the area of the centre and up to, say, r distance away is much smaller than the area of the band of distance at most r from the yellow circle.


Ok, yeah I used the wrong word. Substitute in ‘potential’ rather than ‘ability’ since ability involves the player’s skill and decision making too.

The point I’m trying to get across about RNG though is that a 75% chance is not always a 75% chance. It might be in the computational side of things but in practice a 75% chance can result in 100% misses, 50% misses, no misses, etc.

If a player is told that they have a 75% chance but then miss 2-3 shots then it doesn’t matter what’s on the computational side of things. They’re not seeing a 75% chance, they’re seeing a 33% chance. Which will obviously make the player feel cheated, and for good reason.
If you want to make a game with percentage chances as an indication to the player then the percentages have to be accurate. This means you have to prevent bad luck from making them inaccurate.

Correction, ‘this is where PP should sit’. If the luck part can override all strategy and make it meaningless then it’s not a game in that middle ground, it’s a game skewed toward luck. PP should be the opposite. It should be skewed toward strategy over luck, with luck taking on a much smaller role. This could mean making the guns way more accurate so that they have a base 85-95% chance to hit what they’re aiming at or some clever background processes that make the chances to hit reliable.

You say that over the course of a campaign the player with strategy will win while the other will lose. This isn’t the case. One bad mission can ruin a whole campaign. Lose your best team with all your best gear and you’re stuck with rookies for the next mission that will likely be just as hard as the previous one. You’ll be down on resources from having to re-create good gear and you’ll have lost out on whatever mission reward was available.
Someone with mediocre strategy who never has a terrible mission will likely win compared to a person with good strategy that hits a devastating mission.

Not quite. You see, it’s capped. In a true RNG system every chance would have a 25% chance to fail. This means that 5 in a row could fail simply by the computer rolling a number lower than 25 in it’s RNG multiple times. In a capped system like the one I’m suggesting the chance is reduced every time a miss occurs because there are less misses compared to successes left in the available pool.
So 12 hits, 4 misses means a 75% chance. 12 hits, 3 misses means an 80% chance. Etc.
It’s taking the initial hit chance and carrying it through for all the calculations, rather than doing each one independently. Which means that it’s far more reliable.
Of course, similarly if you get a lot of hits, 4 hits, 4 misses being left in the pool would make it a 50% chance to hit. So it doesn’t benefit the player in any way, it just ensures that hits and misses are balanced out between sets of 16, rather than the player being able to roll infinite misses despite the unlikelihood.

Your argument about gaming the system is true. Players could look into it, find out how it works and then know when they’re certain to land their shots. However players could also use cheat engine to give their soldiers infinite health, or their base infinite resources.
I would also suggest having the pool reset at the start of each mission.
Plus, it would take a LOT of calculation on the players side. I mean, this kind of system would have to account for all percentage chances in one pool. I’ve not studied statistics so I can’t think up a system like that off the top of my head but it would have to be something that accounts for modifiers like cover and whatnot. For example, if you had a 75% base chance to hit but then the enemy is behind cover so that goes down to 25% chance to hit, if the hit landed then you’d have to remove a bigger amount of ‘hit’ from the chance pool.


I’m sorry, I have no idea what you’re saying here.
What is the “area of the band of distance”. Also, “the area of the centre and up to the radius away”? That doesn’t really make any sense… Why would you add an area and a radius?

Pretty sure you’re overthinking this. This is the kind of bell curve it seems they use, with the middle point being the ring of the yellow circle.
So, you can see that the area with the most likelyhood of the bullets landing is just on either side of the centre of the curve. With about 66% of bullets spanning either side of the yellow rim.


I’ve ran into issues where aiming a sniper at a (very) distant enemy wouldn’t highlight any body parts, despite a definitely uninterrupted LoS. Both circles entirely covered crab body parts, but somehow it wouldn’t recognize it as an enemy. These shots always fail to do any damage.

However, accuracy isn’t the cause here. Could either be a bug, a hidden maximum range on weapons that I’m not aware of yet, an engine limitation or anything else.


I’ve had this exact issue. I started a post on it but I can’t remember what I called it. I put it under bugs and I think it’s due to the hitbox of the enemy not being lined up with the visual animation of the enemy. So you see the enemy there but they’re actually crouching behind cover or are a space over or something.


That’s the very definition of 75% chance. Every single roll has a fixed outcome (75% chance) and is totally independent from the previous. Expecting that odds will even out on a small series is the first and worst statistical fallacy, and the reasons players feel cheated. It really is simple misunderstanding of probabilities. thinking "oh, I’ve been unlucky on these last four 75% shots, the fifth HAS to hit is mathematically unsound, and by pandering to that crowd and introducing compensation algorithms in number generators, game developers are reinforcing that idea, which isn’t a good thing imho.

Not for good reason, they’re just confusing probabilities and outcome. Once again, there is no reason for the outcome to match the initial probability on a small series. You can’t look at the outcome of a series, calculate the percentages of successes and say “here, because this series had a 46.5% of success, the initial chance was 46.5%, not 60%”.

By the same logic, we should also prevent good luck from ruining the accuracy. It’s been done, doesn’t mean that I like it. Many games cheat to help the player (even those where you’d least expect it) and if numbers aren’t exposed, then so be it, but if a game is based on rolls, displays the probability of success somewhere on screens and then takes a dump on this number and wipes its arse with the “%” sign, then what’s the point?

That’s another fallacy. Competent streamers have demonstrated consistent win rates in the harder modes of XCOM, Xenonauts or any other similar games. They do experience strokes of bad luck and overcome it. They do experience bad missions and climb back from them, or bail out if it starts to stink. I’d argue that the game IS about managing bad luck, knowing when to bail, knowing when NOT to take a shot.

That’s the first thing you’ll notice when watching people discovering tactical games and more experienced ones playing it. Beginners will take a 65% shot and hope that it hits, while vets will only do it if they have no choice, but will prefer to try and get a better position (and a better percentage), break LoS, hunker down and wait for a better shot, reposition for a guaranteed damage grenade or anything else. Or if they have to take that 65% chance, they’ll ensure that they aren’t in a position to get rekt too hard if it fails, because it’s likely that it will.


It’s not about what’s right and wrong. It’s about what makes a game fun. Plus, saying that statistics are set before the action happens is kinda silly. statistics are reflective. They describe the frequency of one occurrence over another. We can program a computer to select a random number between 0-100 and then only proceed to the ‘hit’ reaction if it’s below 75, while going to the miss reaction if it’s over 75, but how does the computer select the number? It bases it of a number of other variables. Variables which are not random but that are always changing themselves and so the result of the equation seems to be a random number. However, if these variables were involved in repeating processes that caused them to be within a consistent range that caused the random number generator to constantly output answers above 75 then the player would constantly miss.
My point is that we can’t take into account every variable, and neither can game designers. So it’s impossible to state that there’s a 75% chance of landing a hit unless you engineer it so that there absolutely can’t be less or more than a 75% chance to hit.
It’s not always a misunderstanding of probabilities. I mean, if the player misses 3/4 shots then that shows a 25% chance to hit. If it then averages out to a higher number with more shots then great but sometimes players don’t get more than 4 shots in a mission with each character, or that character might die from missing too many shots and not being able to take out surrounding enemies. Which means that for that mission that character did have a 25% hit rate, not a 75% one as they were supposed to.
See, the problem is that you’re thinking purely mathematically. Mathematics is a tool to better understand the real world. It often simplifies things so that they’re easier to compute. When actually applying that mathematics it doesn’t always give the best results.

“chance” percentages are always a rough estimation. There’s never any guarantee that reality will follow the pattern. So actually, even though it was designed to give a 75% chance it might have instead been a 45% chance.

Yeah, I literally pointed this out earlier when I said that if the player hit 8 in a row out of the 12 in the series then they would have a 50/50 chance of hitting and missing next time even if it said 75%.

EXACTLY! :stuck_out_tongue:
How are you not getting this? I’m trying to give the idea of a system where probabilities are reflected by an overarching certainty. Rather than having RNG “take a dump on this number and wipe it’s arse with the “%” sign.”
It’s exactly what you’re talking about there that I’m suggesting a system to avoid. The game gives you a 75% chance and then you miss 3 times in a row. That’s the system “taking a dump” on that number and just screwing you over despite what it promised you: a 75% chance.

I didn’t say that it would always happen, so it’s not a fallacy. The possibility of it happening is still there. What about an above average player that isn’t a pro? They get hit with a terrible mission and it could ruin an otherwise manageable campaign for their skill level.
My point is that extreme bad luck isn’t fun.
By all means, I think Phoenix Point should put in a hardcore mode with true RNG and very little safety nets, because that’s what would make it hardcore. I wouldn’t want to see that in the ‘normal’ or ‘easy’ game modes.

“Or if they have to take that 65% chance, they’ll ensure that they aren’t in a position to get rekt too hard if it fails, because it’s likely that it will.” So 65% shots are unlikely to hit now? You realise that’s 2/3, right? That’s a pretty good rate and it means shots are likely to hit.


But not as likely as 75% shots are to… :wink:


What makes a game fun is subjective option, we all have our own opinions.

2+2 will always equal 4, and a 75% chance to hit something will always be a 75% chance to hit.

Saying that you’d prefer 2+2 to equal 5, or for a 75% chance to suddenly become 80% because it would be more fun is a subjective option that you’re entitled to have, but personally I prefer my maths to work mathematically within games. When games start lying to me about number I really don’t find that fun, I find it broken.


I think you’re confusing percentages with statistics…

Statistics are reflective, they record events that have already happened, they can even compare actual outcomes with expected outcomes. But the game deals with percentages, and percentages aren’t reflective, the give only the expected outcome, the ‘to hit’ roll works based on percentages, not on statistics.

And I think that this is where you’re confusing it. With a 75% chance to hit the expected chance of landing a hit will always be 75% - The result can be either a miss or a hit, and over time the statistical results can vary from 75% (over a long period of time they will trend to 75%) but the percentage chance to hit was and remains 75% regardless of what the actual results were.

A 75% chance to hit stands for each individual shot, it doesn’t stand for a ‘over the course of a game 75% of my shots will actually hit’ and I think in wishing that it should stand for that you’re then possibly mistaking what that 75% chance actually informs you of.


I’d argue that putting all your best team and best gear into a single mission without backup soldiers prepared to step in should you lose that first team is a poor strategy to begin with, regardless of what system is in place.


That’s a fair point, I was forgetting that your sample would reduce in the system that you’re proposing.


With only 3 shots you don’t have statistically significant data regardless of whether those shots all hit, missed, or you got a mixture of hits and misses, you need a larger sample size. If they take 30 shots and miss them all, then I’d say you’ve got a problem, but the chances of that happening are infinitesimally small. (It’s 1 in 1,152,921,504,606,850,000)


This is false though. 2+2 is always 4 but 75% chance to hit will rarely come out as 75 hits per 100 shots.
It’s not saying 2+2 should equal 5. It’s saying 4+0 should equal 4, or 1+1+1+1 should equal 4, or 2^2 equals 4. It’s the exact same percentage chance of landing the shot, it’s just made reliable rather than hoping random chance will reflect the hit chance the soldiers were designed to have.
It wouldn’t suddenly become 80%. It would be 75% chance to hit over all the soldiers shots rather than on each shot.

Do you realise that a percentage is a statistic? The game gives you a statistic that represents how often your soldier should land a hit if it were to run 100 simulations. That’s the statistic and it’s displayed as a percentage.

I’m not confusing anything. I know what a 75% chance represents in a game like Xcom or phoenix point. It represents a random number generator which has been designed to throw out number between 1 and 100. It then takes that number and checks to see if it’s above or below the threshold for the action being performed in the game, in this example 75. if it’s above 75 it sends a negative feedback, a miss. If it’s below 75 then it sends a positive feedback, a hit.
The problem is that if the RNG keeps throwing up numbers higher than 75 then it doesn’t matter what the percentage chance to hit was. It could have been 20%, 50%, etc, it would have missed all the same. My solution means that the percentage chance players are shown actually means something, rather than just relying on chance to even out the hits and misses to equal the ratio they were supposed to.

If all the shots someone takes in a game at 75% chance to hit, don’t average out at about 75% landing the hit then there is clearly something wrong with the way the game was designed. People take a lot of shots in a game. If the statistical evidence doesn’t match up with the predicted chance over that size of sample pool then it’s clearly not a 75% chance in the first place.

You could say that but by taking along rookies to level them up, or by leaving behind your best gear in case you fail, you’re actually increasing the opportunities for something to go wrong and for you to fail the mission. Even if you go for lower difficulty missions to level up the rookies they usually have less of a reward, which puts you behind on another front.

The problem there is that you don’t get a larger sample size than about 12 shots on every mission. Plus, having 3 misses in a row is far worse for a player than having 3 misses spaced out between successful shots or multiple turns. It’s not about pure mathematical principles it’s about the impact on the game as well.


It will. If not 75 out of 100, then 750 out of 1000. That’s why it’s 75%. Unless the implementation of the pseudo-random algorithm is flawed, but then it’s a coding error, not a fundamental design flaw.

You can even have “true random number generator” in computers, it just takes more effort/money than what a simulation usually needs. A decent read:

I say that a coin toss has 50% chance to be head. I say that according to statistics, the 4 toss were all tail, so it has 0% chance to be head. Statistic is something which already happened, chance is predicting the future. You can show statistics with percentage, but that’s not (theoretical) chance.

That’s why there are algorithms/functions which gives you pseudo-random numbers with good enough randomness. Getting 5 out of 10 shots instead of 7.5 doesn’t mean it’s bad. Getting 50 out of 100 maybe.

Nope. All it says is that there were not enough “rolls” to have the chance even itself out. That’s like saying toin coss is badly designed because after one round it’s not 50-50. The winner will say it’s perfectly designed because he could (in theory) lose, but the loser will say Best of 3/5/7/etc.!

You may say it’s not fun for the loser side, but then don’t use randomness in a game. If you KNOW that at the end of a match, both of you will get 50-50, what’s the point of the rng in it? Same for shooting chance, if you know you will have X shots out of Y, it’ll become Into the Breach instead of XCOM. Both are enjoyable, but they use RNG in different ways.

My problem with all the % is that the players expect anything bigger than 75% to be 100%, so unless you make them feel that, they will complain. Even if the RNG was perfectly fair, and over 10k shots you got 7.5k, they will scream when they had 5 missed shots in a row.

The problem is not with the game, it’s with the expectation of the players. Most of them don’t even want to have missed shots, only from enemies. They want “100%” shots, “usually I hit 50%” shots, and “never hit 20%” enemy shots. You can’t make a “fair” game with those expectations. The whole debate is not about true randomness in shots, it’s about managing players… Which I hate, if I see % I want to have it fair, and not “fixed”. For Fixed chance, I play Into the Breach. There I have 100% shots.


The game doesn’t give you a statistic, it gives you a percentage chance. It doesn’t become a statistic until the event happens and is recorded. Not that I’m for or against your idea but there is a difference.

Now one could say that statistically 75% of your shots should land, but what is left off (as its implied) is the “if you were to take an infinite number of shots and record the data.” Though this talk is going more down a symantic rabbit hole at this point.

As for your “groups of 16” idea, I’m leaning more against it and here’s why. If I were to take 12 shots and 11 of them landed then that would mean that I would only have a 25% chance to hit a target with my next shot. If that shot landed then my next three shots would all have a 0% chance to land in order to meet the 75% per 16 rule. At which point I could be forced to miss a far more critical shot just because of the quota system of hits/misses. Likewise I think such a system would be horribly messy once we start throwing in all the different variables that come from angles, range to the target, record keeping, etc.