What is the optimal portal level to glyph hack - an analysis of my Aurora challenge data

DrThodDrThod ✭✭✭

Cross post from Reddit:

In mid June I presented a small study investigating the effect of glyph hacks and especially the speed bonus on the chance to get a key for a P1. You can read about this study here: https://drthod.com/investigating-level-1-portal-key-and-glyph-hacks/

I have now used the data I gathered during the Aurora Challenge to come up with 3 simple rules to find the optimal portal level to key hack.

  1. You need to glyph hack
  2. Go for the highest portal level where you reach 90%+ success rate
  3. If in doubt, go for a lower level

Rule 1 is obvious. Please don't **** me for this one so soon after Aurora - I know of many agents who promised not to glyph hack for a long time.

Rule 2 incorporates the balance between higher drop rates for more glyphs vs smaller speed bonus and the very real and very harsh penalty for failed glyph hacks.

Rule 3 is there because we tend to overestimate our skills. In addition if our skills lead to the total amount of keys to be the same then the lower level portal is the preferred one to hack. Why? It is likely better to gain at least 1 key instead of a higher chance to get 3 keys while also the chance of 0 keys rises.

The full article with tables and more numbers is here: https://drthod.com/what-is-the-optimal-portal-level-to-key-hack/

Comments

  • ClumsyNinjaClumsyNinja ✭✭✭
    edited August 2019

    Nice study, this must have been a nightmare for your patience!

    I think the more keys you are seeing at higher levels is to do with the percentage bonuses.

    P1 perfect hack bonus is 38% meaning you need 62% speed bonus to get 100% (easy, 100%)

    P2 perfect hack bonus is 60% meaning you need 40% speed bonus to get 100% (easy, 100%)

    P3-5 perfect hack bonus is 85% meaning you need 15% speed bonus to get 100% (trivial, 100%)

    P6-7 perfect hack bonus is 120% meaning you need 80% speed bonus to get 200% (hard, say 100%)

    P8 perfect hack bonus is 162% meaning you need 38% speed bonus to get 200% (easy, 200%)

    I think this last percentage (100% or 200%) is applied to the bonus hack, so we can think of P1-7 being three hacks in one (normal, bonus, bonus) and P8s being four hacks in one (normal, bonus, bonus, bonus) - only if the total bonuses exceed 100% or 200% respectively. Therefore all glyphs need to be correct otherwise the 38/60/85/120/162 percentages are nil, so the only bonus you are getting is the speed bonus which can never reach 100% so a failed glyph would produce two hacks in one (normal, bonus). I think things like Power Cubes only have a drop rate of 1 per hack (not sure about keys), therefore I'd expect 4 power cubes max at P8 and 3 power cubes max at P1-7.

    I'm not sure if the 100/200% are hard limits with respect to the amount of hacks in one, but for example if I have failed a P8 glyph I don't recall me ever getting more than 2 keys or power cubes (as the total bonus is under 100% so it's just 2 hacks in one), whereas if I get them all correct and total bonus exceeds 200% I think I've seen 4x power cubes from a P8 before (1 normal and 3x in the bonus).

    The two paragraphs above are just my observations of gameplay and isn't based off any studies, so it's probably not right. It's also what I've noticed more with power cubes and not keys. Oh, and when I'm hacking P8s I'm normally drunk.

    What I'm trying to say is maybe disregard this comment.

  • DrThodDrThod ✭✭✭

    Bonuses don't work that way !!

    You might want to read the following: https://drthod.com/glyph-hacking-how-many-more-items-do-you-get/

    This is the key graph:

    This was prior to my analysis for keys. 100% bonus doesn't mean double the number of items I'm afraid. I don't have yet such a graph for keys - but for P1 you get 0.2 keys more if you increase your speed bonus from 15% to 95%. According to your maths you would expect three times as many extra keys (0.75 * 0.8 = 0.6).

    You also don't have multiple hacks. How do you explain 3 keys (1 normal, 2 bonus) on a P2. I would have to go deep into binominal distributions to show that this can't be explained by multiple dice rolls/coin flips but rather akin to the probability of a single spinner with different areas for zero, one and two.

  • Thanks for the explanation, I knew what I said wasn't 100% correct but it sort of matched what I saw (I don't glyph under P4) without doing an actual study which I'm definitely not qualified to do!

  • DrThodDrThod ✭✭✭

    This leaves a very interesting question: what does the reduction of 33% really mean. I don't think it truly reduces the number of items by 33%.

    If they use the maths of a spinner then the easiest way to represent this is by a table with weights. Let us assume that the weight for nothing is a constant 100. A weight of 1 would give us approx. 1% (1/101). A weight of 25 would give us 20% (25/125) and so on. Doing this would allow you very simply to scale the weight with any arbitrary factor. If that assumption is correct, then reduction by 33% just is a scaling of the weight by 2/3.

    For small chances this is close enough to a reduction of 33%. For higher chances this would differ. A spinner like implementation would easily allow multiple weights on the same spinner. For example weights of 100 (none) 10 (CS) 3 (RS) and 1 (VRS) would pretty well explain the observed drop rates for shields at normal hacks while also explaining that you never get 2 shields in a single (normal) hack.

    You can even include separate weights for double (triple) items. I'm currently at an utter loss to describe this in any different way:

    Out of 364 glyph hacks P2 115 gave PC as a result but only this single one resulted in 2 PC. A single draw can never get you to 2 PC. The speed bonus wasn't even spectacular but just an average 78%.

    On the other hand a table like the following scaled by 238% would explain the data I see (only the weights after the first get scaled up/down)

    100 0 PC

    30 1 PC

    0.3 2 PC

    Draw back - it would mean each level has separate tables.

  • GoblinGranateGoblinGranate ✭✭✭✭✭

    @DrThod The table per level theory makes sense to me since experience has shown me that some portal levels tend to give out more stuff on glyph hack than others.

    Perhaps this helps you current investigation :)

  • vidiconvidicon ✭✭✭✭✭


    @DrThod It appears that the glyph bonus itself has been reduced. See an example from a level 5, 3-glyph sequence here. Ordinarily that would be a 85% hacking bonus, but instead it's 60% (which is actually a 30% reduction, not 33%). There appears to be no reduction in the time bonus (14.71/18 = 81.7). So overall you're getting a 16% reduction.

    Similarly, a correct level 6 4-glyph sequence yields a hacking bonus of 84% (ordinarily 120%) and no apparent reduction in time bonus (in my case, 9.02/17 =53%). This means overall you're getting 137% instead of 173%, or an overall reduction of 21%.


    Missing a glyph on a 4-glyph sequence yields a 21% hacking bonus, which would ordinarily be 30%, so this is a true 30% reduction (still not 33%).

    This then, I believe, gets put into the output formula that you study in your article.

  • DrThodDrThod ✭✭✭
    edited August 2019

    @vidicon thanks - I already noticed myself.

    It is an ideal time from a data miner perspective to gather some unique data now to check the hypothesis how this reduction affects the numbers at the same level.

    But i have to admit - even myself I’m exhausted after the challenge and was looking for a week of reduced play.

  • vidiconvidicon ✭✭✭✭✭

    @DrThod I glyph all the time and don't intend to reduce play this week (honestly I'm happy to get back to glyphing higher level portals for gear). But cataloging output is way too tedious for me to do with a large enough dataset to be useful.

  • TinderbiscuitTinderbiscuit ✭✭✭
    edited May 2020

    I’ve removed the player’s name and any portal info that might give away their location but I got sent this on messenger just now. It was from a P8. I’ve never known anyone be able to get 4 keys except with a fracker. Is This some weirdness to do with A higher P8 bonus probability?

  • oscarc1oscarc1 ✭✭✭✭✭

    Yeah I've had x4 keys before without a fracker as well. I see no mention of x3 in the document either, or P8's.

    That's certainly some interesting insight on your statistics DrThod, thanks for sharing!

  • MirthmakerMirthmaker ✭✭✭✭

    1. Impressed with the math.

    2. Convince me that glyph hacking is NOT overpowered.

    3. Make tap hacking special again, (e.g. particular mods are easier to get tap hacking than glyph hacking).

    There should be a price to pay for trying to cover and own 500 portals at the same time.

  • This is the announcement I was thinking of.

    https://twitter.com/ingress/status/801563523394523136?s=20

    I've never seen an official announcement that stated you could get 4, so I'm not sure if it's intended or just an unexpected feature. :P

    It may be that they have changed it since then. I don't think I've ever seen 4 keys myself but then I'm lucky if I ever get to see a P8.

  • I’ve got 4 keys yesterday from a portal. It doesnt happen so often, but it has to be a portal of higher level. Probably a p6 or a p8 where maximum of bonus can be reached.

  • ClumsyNinjaClumsyNinja ✭✭✭
    edited June 2020

    @Tinderbiscuit If you read my comment above, that fits in with what I think.

    My comment is only applicable to power cubes and keys though, so I would expect a max of 4 power cubes and 4 keys from a P8, and 3 power cubes and 3 keys from P1-7. No studies, just fits with what I've observed (although I've never seen 4 keys on a P8, only 4 cubes).


    EDIT: Would also expect 4 keys/cubes from a P6/7 if you got over 80% speed bonus as well, but this is probably hard to prove?

  • Is there a follow up to that 2019 research?

Sign In or Register to comment.