Talk:Lucky Crimbo tiki necklace

From A KoL Wiki

My guess is that this works the same as The Marmot. We,ll have to see if their effects stack. --MaxVance 23:27, 24 December 2006 (CST)

  • I just spent a 160 adventures with it at the Giant Castle (lotsa non-combat adventures) and didn't get a single clover. --Misanthrope 05:44, 25 December 2006 (CST)

What I really want to know is if multiple necklaces stack! --Barstool 23:38, 24 December 2006 (CST)

  • Since the clovers come from a status effect the item causes rather than the item itself, it's safe to assume multiple necklaces do not stack. --Smello 12:56, 26 December 2006 (CST)
  • Um yeah. Didn't Jick confirm that while Marmot and Necklace stacks, multiple necklaces do not stack for that very reason. --CluckyB 22:36, 26 December 2006 (CST)
  • Is that a question. --Co678 14:24, 27 December 2006 (CST)
  • I reverted the note added by GRORG, "Seems to weaken enemy sometimes" -- what is the evidence for this? Is there combat text supporting this claim? Delevelling is incredibly tricky to spade, so I think there should be some solid research or an in-game hint before this is added back.... --DirkDiggler 05:25, 26 December 2006 (CST)
  • At around 1:15 into the Dec. 28 radio show, Jick estimated that if the neclaces did stack, and if someone with Marmot wore 3 of them, they would be getting a clover about once every 12.5 adventures (in other words 8%). So, I would theorize that necklaces and Marmot both give 2%. --Prestige 00:30, 29 December 2006 (CST)

spent 260 clover adventures at crimborg and got 6--Error1 00:12, 31 December 2006 (CST)

  • Seems like the probability is more like 3 percent, then... Anywhere from two to three percent seems reasonable... BTW, when I calculated it out, it turned out to be about 2.91%...--Pancor 01:58, 31 December 2006 (CST)

Can you gain 10-leaf clovers while resting at your campground with this? --ImperialBest 05:50, 31 December 2006 (CST)

Keeping the description, the tiki necklace gives an effect when surfing. The Tubular Vacation adventure at the shore gives the following message after the normal text:

Suddenly, your board gets caught in the undertow! You fall off and are sucked beneath the waves. In the water, you'd swear you saw an evil green glow coming from your tiki necklace. You surface and swim to shore, your heart pounding in your chest.

In addition, I gained muscle subpoints, 50% of the value of the myst subpoints given for the adventure. --Nekosoft 17:03, 2 January 2007 (CST)

I think that the lucky crimbo tiki necklaces stack with one another. I had 3 equiped while clovering in the gallery to get my level 30 trophy, and had about x3 amount of "luck" getting clovers then the 2 days that I only had one necklace equiped. I also was not under the marmot sign. --Nic 20:14, 6 January 2007 (CST)

just adventured 263 times in the eXtreme slope, had sonata of sneakiness on me. mafia said 55% chance of non-combat, scored a whopping 3 clovers =/ --Braindancer 12:10, 18 January 2007 (CST)

The item notes say this doesn't bring you luck while surfing. I dunno, but additional stat points as part of a noncombat adventure sounds lucky to me... --Missingno 12:23, 21 January 2007 (CST)

Does the Lucky Crimbo tiki necklace increase the number of 'worthless' items you get from the sewer? --UltimateRevolution 21:22, 31 January 2007 (CST)

Clover Adventure Test

It strikes me that the best way to test this is with clover adventures, since they're 100%. i spent 299 turn at the ballroom today with three necklaces on and got 15 clovers. perhaps a table is called for. --Evilkolbot 02:22, 24 January 2007 (CST)

  • oh-oh! smells like stacking to me. that, or the usual rng love. --Evilkolbot 16:29, 6 February 2007 (CST)
  • thinking about it, the "effect not stacking" doesn't seem to be entirely logical. one mechanic would be that you have a percentage chance of finding a clover on every adventure. this chance is generally zero. the marmot adds a non-zero element to this percentage, as do the tiki necklaces. thus stackage. --Evilkolbot 16:36, 6 February 2007 (CST)
  • or, more definitely: there is an M in N chance of finding a clover in each adventure. a tiki necklace adds X to M. M is zero. from the above, it would seem likely that N is fifty, and X is one. or multiples thereof. with such small datasets this is more than likely wrong, though. time will tell. --Evilkolbot 06:39, 7 February 2007 (CST)
But stacking the same effect thrice!
Though this is lovely. --Raijinili 22:35, 7 February 2007 (CST)
  • results for two don't, at first, disprove the hypothesis. yay! --Evilkolbot 06:25, 9 February 2007 (CST)
  • so, had i bothered to read the page, 2% would have been my answer. now we got something that doesn't disprove it. although jick's maths (cow/plate) isn't entirely perfect (which it doesn't have to be since he is providing mechanics for how the world is and not how it should be) so his estimate of 2% could be out by a little. --Evilkolbot 04:42, 11 February 2007 (CST)
  • This is strong work evilkolbot. I've been checking in on your numbers to see if I can figure out a pattern but the larger gap between 3 necklaces and 2 is pretty baffling with anything I can come up with. The ratios between the rates share a common factor: 6.64/3.81=1.74; 3.81/2.19=1.73 =~ sqrt(3), but I can't posit some underlying model to make this happen. Looking at the uncertainties (binomial uncertainty in a ratio p=A/N is sqrt( p*(1-p)/N ) -- see PDF with hairy math), the row that needs the most data is unfortunately the most costly: the 1-tiki-necklace case. 0.63% out of 2.19% is a 28% relative error, and makes a big swing in any model we can hypothesize.--DirkDiggler 20:45, 14 February 2007 (CST)
  • you say that the uncertainty is currently so large as to make any analysis moot, how many more turns would cut it down? do you need fixed numbers of turns (200, say) or is the current method acceptable? --Evilkolbot 11:45, 15 February 2007 (CST)
  • Your current method looks perfectly good to me, provided we are fairly sure that the clover rates have nothing to do with moonlight or other confounders. I would guess that 10000 adventures with each of 1, 2 and 3 necklaces would be well more than enough to give accurate drop rates. If I were doing this, I'd get about another thousand adventures with 1 necklace on to even things out and then be happy with the results - it looks like decent support for the +2% per tiki necklace hypothesis. This is all assuming that you don't have the marmot sign, of course, though I would love to see the same testing applied there. Just remember, the more turns you spend the lower the uncertainty of the results.--Yiab 10:38, 26 February 2007 (CST)
  • no marmot for me. later, maybe. much more data needed, then. oh well. three more trophies, should be three times as much data. we'll see then if it converges to 2% or gets even wilder, suggesting, perhaps, that moons or something else is having an effect. --Evilkolbot 11:54, 27 February 2007 (CST)
  • to neutralize moons, you should adventure with 1, 2, and 3 on same day. Say, 100 advs with 1, 50 with 2, 50 with 3 or something (want more samples at 1, but able to track moon trend).--DirkDiggler 21:37, 27 February 2007 (CST)
  • Just wanted to let you know I added some spading to your table. SSPD Stupor is also 100% noncombat, and you can get clovers after the special novelty button and glowstick adventures too. So if anyone else wants to wear some tiki necklaces while doing their SSPD adventuring, we can get some more data. Looking at my small sample of 400adv, looks like I was slightly lucky, but it's consistent with other "2 necklace" tests on that chart.--Whiskey Jack 16:07, 17 March 2007 (CDT)

Test Data

Date User Zone Necklaces Adv Clovers %age (+/- Unc'y)
24/Jan EvilKOLBot Ballroom 3 299 15 5.02%
25/Jan EvilKOLBot Ballroom 3 205 15 7.32%
26/Jan EvilKOLBot Ballroom 3 411 29 7.06%
06/Feb EvilKOLBot Ballroom 1 278 7 2.52%
07/Feb EvilKOLBot Bathroom 3 311 27 8.68%
08/Feb EvilKOLBot Bathroom 3 311 16 5.14%
09/Feb EvilKOLBot Bathroom 2 309 10 3.24%
10/Feb EvilKOLBot Bathroom 2 312 11 3.52%
11/Feb EvilKOLBot Bathroom 2 312 11 3.52%
12/Feb EvilKOLBot Bathroom 2 331 14 4.50%
13/Feb EvilKOLBot Bathroom 2 305 13 4.26%
14/Feb EvilKOLBot Bathroom 1 269 5 1.86%
27/Feb EvilKOLBot Bathroom 1 97 2 2.06%
28/Feb EvilKOLBot Bathroom 1 170 3 1.76%
1/Mar EvilKOLBot Bathroom 1 316 8 2.53%
2/Mar EvilKOLBot Bathroom 1 302 2 0.66% (ARGH!)
3/Mar EvilKOLBot Bathroom 1 182 3 1.65%
17/Mar WhiskeyJack SSPD Stupor 2 400 19 4.75%
18/Mar WhiskeyJack SSPD Stupor 2 375 19 5.07%
Totals     3 1537 102 6.63% (+/- 0.63%)
      2 1549 59 3.81% (+/- 0.49%)
      1 1614 30 1.85% (+/- 0.36%)
  • To calculate the uncertainty in a sampled ratio p=A/N (here, p=Clovers/Adventures), use the formula
 se_p = sqrt( p*(1-p)/N ) = sqrt( clovers * (advs - clovers) / (advs^3) )

(this is the standard error of sampling mean for a binomial distribution, if I recall the terminology right -- see this PDF for (slightly hairy math) background.

Test Data II

Date User Adv 1 2 3
4/Mar EvilKOLBot 133 5 3.76% ~1.65% 3 2.26% ~1.29% 4 3.01% ~1.48%
5/Mar EvilKOLBot 101 6 5.94% ~2.35% 4 3.96% ~1.94% 1 0.99% ~0.99%
13/Mar EvilKOLBot 99 3 3.03% ~1.72% 4 4.04% ~1.98% 1 1.01% ~1.00%
14-Mar EvilKOLBot 101 8 7.92% ~2.69% 5 4.95% ~2.16% 3 2.97% ~1.69%
15-Mar EvilKOLBot 101 6 5.94% ~2.35% 2 1.98% ~1.39% 3 2.97% ~1.69%
16-Mar EvilKOLBot 100 7 7.00% ~2.55% 5 5.00% ~2.18% 3 3.00% ~1.71%
19-Mar EvilKOLBot 91 6 6.59% ~2.60% 5 5.49% ~2.39% 1 1.10% ~1.09%
20-Mar EvilKOLBot 99 5 5.05% ~2.20% 4 4.04% ~1.98% 1 1.01% ~1.00%
21-Mar EvilKOLBot 67 4 5.97% ~2.89% 3 4.48% ~2.53% 1 1.49% ~1.48%
Total 892 50 5.61% ~0.77% 35 3.92% ~0.65% 18 2.02% ~0.47%

I just got 16 clovers in 307 clover-adventures in the Haunted Bathroom with 3 lucky crimbo tiki necklaces in the Platypus sign. I'll be doing some more powerleveling on stat days, so hopefully I can give some more data. --Dalgar 19:48, 14 October 2007 (CDT)

Two thoughts:

Here's what I know:

First, I adventured in the Haiku Dungeon for 100 turns - 100% non-combat adventures. I got a single clover from a single Tiki necklace.

Next, I adventured in the Haiku Dungeon for 100 turns with TWO necklaces, and I got 4 clovers. This is not enough to prove anything by itself, but I do offer the Haiku Dungeon (or the companion Dungeon, the Limerick) as an ideal place to test drop rates. Also, you could get drunk.

Finally, it is my theory that each individual necklace is tested independently, so each necklace event has a separate trigger. Jick likely coded the event to not drop more than one clover at a time - this means: if the base percentage chance of a clover drop is 2 out of 100, then wearing two necklaces would not bump your chance to 4%, but to 3.96% (the difference is the chance that both events trigger, but only dropping a single clover). Adding a third necklace would bump your chance to 5.8808%

This would be interesting to test, but we would need several more 100-turn tests. I suggest keeping to 100-turn tests to make the math easier. I'll calculate how many tests we would need to get a reasonable degree of accuracy...

--Nyl 186 16:25, 15 February 2007 (CST)

  • Sorry if that came across overstated: you've done excellent work here, and I bet you're fairly close to being able to indentify that underlying mechanic.

You've certainly demonstrated that the necklaces stack -- the rates for 1,2, and 3 necklaces are several standard deviations separated over a convincingly large sample. The numbers you have are also plenty good enough for telling an adventurer about how many clovers she should expect for a given # of necklaces. But I can think of several models that are not-inconsistent with the data now:

    • If we said each necklace gives an additional x% chance a least-squares fit to the data gives 2.123% : 4.25% : 6.37%.
    • If the necklaces cascade we'd have p1=r : p2=p1+((1-p1)*r) : p3=p2((1-p2)*r). A least-squares fit gives 2.158% : 4.27% : 6.34%.
    • If N necklaces gives (baserate * 3^(N/2) a least-squares fit gives 2.208%: 3.83% : 6.62%

All of these models fit within uncertainty, mostly because of the large error bars on the 1-necklace base case. At near 2% clover rate for 1 necklace, 1250 adventures would give you uncertainty of about 0.4% (~20% rel. unc'y); 3400 adventures would give you uncertainty of 0.25% (~10% rel. unc'y). I think 1250 or more adv's worth with one necklace would make the base rate more precise and let you rule out some of the models. Also, having different-sized samples shouldn't make any difference. It's worth seeing if any moon effect or clover day effect shows up in your data, though.--DirkDiggler 17:12, 15 February 2007 (CST)

  • the data seems to be settling down to within half a variance of the assumed percentage. except for three clovers. when i've got one necklace up to 1500 i'll start splitting the turns into three lots of 100 and seeing if moons have any effect. small numbers though, and with such big viariances it will be a looooooong time before there is enough data to tell (1000 adv with each, 100/day = 10 * moon cycle = 1760 days, 5 years!) this unsigned comment was added at 12:59 GMT, on 1 March 2007 by Evilkolbot
Retrieved from "https://wiki.kingdomofloathing.com/index.php?title=Talk:Lucky_Crimbo_tiki_necklace&oldid=484310"