Riddler - Solutions to Castles Puzzle: castle-solutions-3.csv
Data license: CC Attribution 4.0 License · Data source: fivethirtyeight/data on GitHub · About: simonw/fivethirtyeight-datasette
153 rows where Castle 2 = 3
This data as json, copyable, CSV (advanced)
Suggested facets: Castle 1, Castle 3, Castle 4, Castle 5, Castle 6
Link | rowid ▼ | Castle 1 | Castle 2 | Castle 3 | Castle 4 | Castle 5 | Castle 6 | Castle 7 | Castle 8 | Castle 9 | Castle 10 | Why did you choose your troop deployment? |
---|---|---|---|---|---|---|---|---|---|---|---|---|
3 | 3 | 2 | 3 | 4 | 5 | 6 | 22 | 6 | 22 | 22 | 8 | Based on previous results, I focussed on castles 6, 8 and 9 and left myself a healthy backup in each of the others |
21 | 21 | 3 | 3 | 3 | 3 | 3 | 10 | 15 | 20 | 30 | 10 | Just guessing based on the previous two events. 678 heavy vs 459,10 heavy, sort of a mix. |
24 | 24 | 2 | 3 | 4 | 5 | 7 | 9 | 26 | 33 | 6 | 5 | It just felt *right* |
32 | 32 | 1 | 3 | 5 | 7 | 9 | 11 | 13 | 15 | 17 | 19 | Linear |
40 | 40 | 2 | 3 | 4 | 6 | 8 | 9 | 18 | 20 | 12 | 18 | I am uncertain as to how people will adjust to two contests worth of results, so I've taken a slightly more balanced approach that targets higher value castles more proportionately to their values, while still leaving enough troops to pick up the low and mid value castles that others may defend lightly. |
43 | 43 | 2 | 3 | 3 | 12 | 15 | 7 | 14 | 14 | 17 | 13 | I looked at the historical success strategies of the first and second FiveThirtyEight crusades. It looked like people in the second war adjusted their strategy away from what won in the first war. So I took the top 5 from each war and took the average number of troops per castle. I picked numbers close to the average to deploy my troops for the Third FiveThirtyEight crusade. And once I take over the world, I'll change the name of your website to FiveThirtyNine. |
45 | 45 | 1 | 3 | 5 | 5 | 5 | 5 | 5 | 32 | 5 | 34 | The most prominent strategies that have been winning have been strategies that have had the "four castle" strategy which would win the slight majority of the points (28). Assuming this is the strategy most people seek to optimize on I wanted to build a strategy that would beat these strategies. Every four base must win either castle 10 or castle 8 to reach this 28 point threshold (which is the primary way they win). After that the number of troops sent to the other castles should be greater than with a four castle strategy that you win the rest of the needed points on the castles that others gave over for free. I would like to test it with 30 in bases 8,10 and 5 troops in 1 and 3 as well but I think you need to make sure you juice your troop count in the bases you are going for because if you don't win at least one of those you are going to be in trouble. You will also lose to a split evenly strategy but I don't think that will be popular as most people will look at the data and realize you probably want to have a win condition. |
61 | 61 | 3 | 3 | 1 | 1 | 1 | 1 | 10 | 35 | 44 | 1 | focus on castle 8 and 9 with the assumption that castle 10 is likely going to be taken and castle 1 and 2 will have 1 soldier brought to them |
69 | 69 | 3 | 3 | 3 | 3 | 13 | 18 | 21 | 32 | 2 | 2 | I put at least two people in each castle so I could beat the 0s and 1s in each castle. And then I tried to mimic the winners from the first time, thinking that the winning strategy would revert back to the first game. |
79 | 79 | 3 | 3 | 4 | 6 | 7 | 9 | 15 | 25 | 27 | 1 | I've never actually participated in something like this before. I assumed most people would attempt to capture the castle worth the most points (10). I felt if I essentially sacrificed that castle and then stuck to a rather linear distribution of soldiers increasing from 1-9 I stood a greater chance of capturing those castles and thus winning the Game. I guess we'll see. |
86 | 86 | 1 | 3 | 5 | 7 | 9 | 11 | 13 | 15 | 17 | 19 | Trying to be competitive at every single castle, without wasting too many soldiers. |
101 | 101 | 2 | 3 | 3 | 3 | 21 | 17 | 2 | 3 | 24 | 22 | Last times winner but more even alignment |
105 | 105 | 3 | 3 | 14 | 4 | 18 | 15 | 3 | 15 | 4 | 21 | Randomish |
106 | 106 | 2 | 3 | 1 | 5 | 16 | 28 | 6 | 9 | 18 | 12 | Troop deployments to low point castles are just enough to tie up enemy troops while focusing on the mid to upper range castles that are worth the most. Don't over dedicate to 10 as people are drawn to the easy number. |
111 | 111 | 2 | 3 | 4 | 4 | 21 | 21 | 21 | 22 | 1 | 1 | I sacrificed 9 and 10 hoping that my enemy would focus a lot of soldiers on them and instead tried to capture a lot of of the mid value castles. |
120 | 120 | 3 | 3 | 5 | 5 | 3 | 16 | 17 | 16 | 16 | 16 | you need 28 points to win. I maximize my chances of winning 10 points 100% of the time in castles 1-4, concede castle 5, then hope even distribution wins me 3 of 5 in castles 6-10 versus a field that allocates 30 plus to a single castle. |
130 | 130 | 1 | 3 | 1 | 7 | 4 | 12 | 32 | 3 | 34 | 3 | Lots of folk went for 7-8 or 9-10 previously. I figure few will go for 7-9. With those in the bag, I need another 12 points. I'm hoping for 2-4-6, but also spreading out my options to get lucky against a poorly defended 8, 10, and 5. |
135 | 135 | 2 | 3 | 0 | 5 | 7 | 12 | 16 | 18 | 18 | 19 | Idk let's see if I win |
136 | 136 | 1 | 3 | 5 | 7 | 9 | 11 | 13 | 15 | 17 | 19 | Need 28 pts to win, expected value of n pts/ 55 total its per castle. Rounded up higher pt castles. |
158 | 158 | 3 | 3 | 11 | 11 | 4 | 4 | 19 | 20 | 21 | 4 | The past winners placed 2-3 troops at each of their worst bases, by placing 4 I could acquire those bases at a lower marginal cost of entry. I wanted to try and take 5 bases total, and wanted to make sure that each of those 5 bases had more than 10 so that I could beat out the average person who just runs 10's across the board. I avoided the 10 spot because I think the average person will overplace value on that and overallocate their troops there. |
162 | 162 | 3 | 3 | 6 | 13 | 15 | 17 | 4 | 15 | 15 | 9 | Sort of a smooth mound shape but I pulled back on #7 to boost the tails |
166 | 166 | 3 | 3 | 3 | 6 | 6 | 25 | 25 | 27 | 1 | 1 | |
167 | 167 | 2 | 3 | 5 | 7 | 12 | 17 | 20 | 25 | 5 | 4 | Sounded good to me |
169 | 169 | 2 | 3 | 4 | 0 | 6 | 15 | 10 | 26 | 34 | 0 | Clustered to win as many points against last time's winners. |
170 | 170 | 1 | 3 | 3 | 4 | 4 | 7 | 8 | 13 | 20 | 37 | Roughly exponential increase for each next castle |
185 | 185 | 2 | 3 | 4 | 12 | 1 | 24 | 4 | 26 | 2 | 22 | Pretty random, some psychology |
203 | 203 | 3 | 3 | 3 | 3 | 12 | 12 | 3 | 29 | 29 | 3 | I choose to concentrate on towers 8 and 9, hopefully winning them almost all the time. I should also win towers 5 and 6 much of the time making 28 points for a victory. If I miss one or both of 5 and 6, I hope to make it up with scouting forces of 3 soldiers which may be more than most scouts. |
223 | 223 | 2 | 3 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 11 | random assignment |
229 | 229 | 1 | 3 | 5 | 7 | 9 | 11 | 13 | 15 | 17 | 19 | I reinforced the higher value castles with 1 army from each less-valued castle in the hopes that I could both win some high-value battles against warlords trying to win a greater number of low-value castles and some (more?) low-value battles against top-heavy warlords. |
253 | 253 | 2 | 3 | 5 | 8 | 2 | 22 | 23 | 4 | 27 | 4 | Well, I didn't use *actual* game theory, that's for sure! |
257 | 257 | 2 | 3 | 4 | 5 | 6 | 6 | 32 | 31 | 5 | 6 | Focused on 2 in the middle, never lower than 2 to beat the 1s deployed and heavier on two important |
266 | 266 | 1 | 3 | 5 | 7 | 10 | 12 | 0 | 19 | 23 | 20 | Slight tweak on EV 1, 3, 5 etc. deployment |
269 | 269 | 3 | 3 | 3 | 3 | 3 | 20 | 20 | 21 | 21 | 3 | |
280 | 280 | 1 | 3 | 3 | 3 | 3 | 26 | 4 | 26 | 27 | 4 | Winning 6, 8 and 9 will all but assure me victory. If I lose one of them, I hope I have enough at castle 7 or 10 to pick up one of those instead |
284 | 284 | 3 | 3 | 11 | 15 | 18 | 11 | 22 | 6 | 4 | 7 | Random numbers with the majority of troops deployed to castles with medium values (4-7). |
298 | 298 | 1 | 3 | 7 | 17 | 17 | 18 | 22 | 5 | 6 | 4 | I kept enough in the top three to catch any that decided to sluff those, then loaded up on 4-7. 10 is just not SO much more than 6 or 7 that it justifies a huge commitment. |
310 | 310 | 1 | 3 | 4 | 8 | 10 | 13 | 16 | 1 | 34 | 10 | I assigned troops proportional to castle value, then sacrificed castle 8 and a bit of castle 10 to target castle 9. Just to change it up. |
316 | 316 | 1 | 3 | 5 | 7 | 9 | 13 | 16 | 18 | 15 | 13 | I took last round's averages and shaved the lower half to give more juice to the top castles. |
360 | 360 | 2 | 3 | 4 | 20 | 26 | 15 | 10 | 10 | 5 | 5 | There are 55 points available for capture. The first to 28 points wins. No one can win unless they capture AT LEAST 4 castles. Most people would likely try to capture the most valuable castles first and weight their troops towards those objectives. But those who spend 50ish troops on castles 9/10 only have 50ish troops to spend on the remaining 8 castles, needing to win at least 9 points, between those 8 castles. I could see a 2, 7, 9, 10, strategy working well enough compared to last year's 4, 5, 9, 10, meta. By all but abandoning 9 and 10, i should like take the other 8 castles in most scenarios mainly due to the fact that the enemy had no more troops to spend. I put substantial enough troops in each castle that no one can steal cheap points without investing a fair amount into those castles in the first place. Against last year's winners, I would have won: Vatter: 36-19 Winder: 33.5-21.5 Shafer: 36-19 Schmidt: 35-20 Trick: 36-19 |
363 | 363 | 6 | 3 | 3 | 16 | 3 | 22 | 31 | 4 | 4 | 8 | I found that having more troops at castles 1, 4, 6, 7, and 10 would be enough to win, so I focused on those. Also, those castles were not as heavily contested last time. I did just enough in those castles to win most games last time then allocated the rest of the troops to the other castles. |
366 | 366 | 3 | 3 | 4 | 6 | 6 | 3 | 3 | 34 | 4 | 34 | |
387 | 387 | 3 | 3 | 3 | 3 | 11 | 11 | 16 | 21 | 26 | 3 | You’ll never know |
398 | 398 | 2 | 3 | 3 | 5 | 17 | 19 | 19 | 2 | 2 | 28 | Try and win 5, 6, 7, and 10 against most people, which would give me the 28 points needed to win. |
401 | 401 | 1 | 3 | 5 | 7 | 9 | 11 | 13 | 15 | 17 | 19 | I figure everyone else is goint to overthink it, so I just went with a basic strategy. Since every castly is worth progressively more, I decided to put progressively more troops in each castle |
403 | 403 | 1 | 3 | 4 | 15 | 15 | 16 | 16 | 26 | 2 | 2 | Two basic "coalitions" can get to 27 points. The first is 8+9+10 (or mods of 9+10 lower numbers). The second is 8+7+6+(7--either 5+2, 3+4, 4+2+1, 5+3, 5+4). Because the winners all took the first last time, I'm focusing on the second. I give extra protection to 8 because it is most likely to be challenged by an 8+9+10 strategy. I need to win all of 8, 7, 6 and at least one of 5, 4, with 3,2,1 insuring against the loss of either 5 or 4. The oddity of my approach is that it would lose to the past winning strategy, but I expect that the _reason_ that strategy won is that most people attacked the 8 rather than devote so many resources to the 4 and 5, and that people will shift toward 8,7,6 and away from 4 and 5 this time. I keep a few guys on 9 and 10 as insurance against similar strategies that are more purist. |
429 | 429 | 0 | 3 | 3 | 13 | 15 | 16 | 17 | 17 | 10 | 6 | The lower numbers are obviously less valuable. 10 and 9 I armed moderately, so that they could take a small force, but I didn't want to waste forces that could be used on the medium-high numbers. Those are the meat, and if past trends prevail, 10 and 6 may very well be good enough to beat many people anyway (for 9 and 10) |
434 | 434 | 3 | 3 | 8 | 3 | 21 | 5 | 26 | 10 | 10 | 11 | I wanted to defeat the previous champions. The first round winners won by going heavy in 4,5,9,10. The 2nd round they went heavy in some combination that didn't include 9,10. I went for go for 7, 5 and 3. With average values in 8,9,10 in hoping to get one or two of these. |
444 | 444 | 0 | 3 | 5 | 4 | 10 | 17 | 0 | 0 | 29 | 32 | I assumed everyone would group-think back to the round before the last one (focusing on 7 and 8). Given that, I mostly copied the strategies of the last round , assuming that everyone else is "too smart" to try it. |
460 | 460 | 2 | 3 | 3 | 15 | 18 | 23 | 24 | 1 | 1 | 10 | |
476 | 476 | 1 | 3 | 6 | 8 | 10 | 12 | 14 | 16 | 15 | 15 | I split the difference between the average soldiers per castle from the previous iteration vs. roughly proportional #s of soldiers per castle value. |
481 | 481 | 1 | 3 | 5 | 7 | 9 | 10 | 13 | 15 | 17 | 20 | Roughly their percentage value of 55 total available points. |
482 | 482 | 0 | 3 | 4 | 5 | 1 | 8 | 15 | 18 | 22 | 24 | Looks good to me! |
488 | 488 | 3 | 3 | 4 | 18 | 18 | 3 | 6 | 11 | 17 | 17 | I looked at the top deployments from the previous rounds and looked at how they fared against each other. I then chose the best one and manipulated it until it beat all the others. |
490 | 490 | 0 | 3 | 8 | 9 | 13 | 5 | 28 | 30 | 2 | 2 | |
501 | 501 | 3 | 3 | 3 | 1 | 1 | 15 | 3 | 22 | 27 | 22 | This combination had a good performance in tests against the data from past competitions |
502 | 502 | 1 | 3 | 4 | 7 | 13 | 20 | 24 | 28 | 0 | 0 | I figured most people would choose increasing sequences, which means a lower numbers on 1-8 and more on 9 and 10. So if I put all my solders on 1-8 and beat them, maybe I'd have a better chance! :) |
553 | 553 | 1 | 3 | 5 | 10 | 16 | 26 | 20 | 11 | 4 | 4 | Why wouldn't you choose this troop deployment? |
557 | 557 | 2 | 3 | 4 | 6 | 10 | 13 | 14 | 19 | 15 | 14 | Average of prior deployment data with small adjustments. |
564 | 564 | 1 | 3 | 0 | 4 | 15 | 0 | 21 | 16 | 9 | 31 | I generated it randomly. I multiplied the value of each castle by a random real number selected from a Poisson distribution with rate=2, rounded down to the nearest integer, then gave any remaining soldiers to castle 10. I generated a few allotments this way, picked one that looked nice, and checked it against the top five from the past two iterations. I had a decent record against past winners so I went for it! |
572 | 572 | 0 | 3 | 6 | 8 | 9 | 11 | 12 | 14 | 17 | 20 | Using a base-10 logarithmic scale to determine base troop deployment for each castle (base troop deployment = log(castle#) * 10). Deduct each base number of troops deployed at each castle from 10, and send those troops to each castle in reverse order. E.g. spare troops from #1 go to #10, spares from #2 to #9, and so on until spares from #10 go to #1. I end up not sending any to #1 because log(1) = 0 and log(10) = 1. |
581 | 581 | 1 | 3 | 5 | 7 | 9 | 11 | 13 | 15 | 17 | 19 | I just distributed troops proportionally to the value of the castle. I very strongly doubt that this will be successful. |
583 | 583 | 1 | 3 | 6 | 8 | 13 | 14 | 15 | 16 | 1 | 23 | This feels like what Nate Silver's mom would do. |
593 | 593 | 3 | 3 | 6 | 13 | 6 | 18 | 9 | 11 | 14 | 17 | I generated some random troop deployments, had them all battle each other, and this was the best one. |
595 | 595 | 3 | 3 | 1 | 2 | 3 | 6 | 5 | 20 | 12 | 45 | Made a non linear shot for two big numbers and hope to get a couple of lower castles. |
597 | 597 | 0 | 3 | 5 | 6 | 15 | 5 | 6 | 10 | 24 | 26 | Basically a half-baked revision of the winner of the last time (trying not to duplicate exactly or respond too directly) |
607 | 607 | 0 | 3 | 3 | 5 | 10 | 21 | 21 | 21 | 10 | 6 | Castle 1 is basically worthless, and as for the rest I just have to beat the most people, not the best people. So I'm assuming most people who do this didn't read and react the previous results and will therefore lose to a similar strategy as before just with minor tweaks. |
608 | 608 | 3 | 3 | 3 | 3 | 3 | 17 | 17 | 17 | 17 | 17 | |
609 | 609 | 3 | 3 | 3 | 4 | 4 | 5 | 5 | 5 | 34 | 34 | Slanging it |
612 | 612 | 3 | 3 | 9 | 2 | 3 | 14 | 21 | 5 | 17 | 23 | There are 7 strategies I'm trying to beat, 4 historical and 3 forecasts. The 4 historical strategies are the February Average, the May rematch Average, and the two champions Vince Vatter and Cyrus Hettle. The 3 forecasts are what I call the "Forecast Average," and Copycat 1 and Copycat 2. The Forecast Average is what I expect the average castle distribution to be based on the last two battles: 3,4,8,9,11,11,14,15,12,13. The Copycats are players who are trying to synthesize the strategies of the last two winners. Copycat 1 focuses troops on castles 5, 8 and 9 (distribution: 1,3,5,8,12,2,3,31,33,2). Copycat 2 focuses troops on castles 4, 6, 7, and 10 (distribution: 2,2,6,12,2,17,22,2,3,32). My distribution scores very well against the 3 historical averages, which I hope will represent the majority of players and get my win rate above 50%. And hopefully it narrowly defeats most of the elite players who are trying to copy previous champions, putting me in the upper echelon. |
628 | 628 | 2 | 3 | 4 | 6 | 7 | 11 | 12 | 14 | 16 | 25 | Trying to adhere to the 2 troops for 1 vp but with some skew to capture 10 based on last time around. |
630 | 630 | 1 | 3 | 7 | 9 | 12 | 8 | 24 | 30 | 3 | 3 | |
643 | 643 | 1 | 3 | 5 | 7 | 9 | 10 | 12 | 14 | 16 | 23 | Trying to maintain approximately the same troop-to-score ratio for each castle (1.8 soldiers per point, rounded down), then threw my 5 left over soldiers into castle 10 to try and win the highest scoring castle. |
649 | 649 | 2 | 3 | 4 | 20 | 23 | 13 | 4 | 7 | 0 | 24 | Counter Strategy |
655 | 655 | 2 | 3 | 5 | 6 | 7 | 15 | 14 | 15 | 16 | 17 | Tried to win castle 6, plus 2 of 7, 8, 9, 10 assuming that most contestants will go for 2 of 7, 8, 9, 10. Then scatter enough on 1-5 to pick up some points there. |
676 | 676 | 1 | 3 | 3 | 3 | 5 | 5 | 10 | 10 | 30 | 30 | I distributed them based on how important the castle was. |
682 | 682 | 1 | 3 | 4 | 9 | 6 | 18 | 8 | 18 | 6 | 27 | Base: Assign soldier number equal to castle number using 55. Do it again using castle #-1 using the other 45. Adjust: disfavor odd # castles trying for wins in #4, 6, 8, and 10. |
688 | 688 | 1 | 3 | 6 | 7 | 9 | 10 | 13 | 15 | 17 | 19 | If F is a fraction of the troops, 1F+2F+...+9F+10F should equal 100. F is 100/55, or 1.81818...As there are no fractional people, I wanted to allocate the closest whole-number equivalents to 1F, 2F, etc. to the various castles, to minimize my ‘shortfall fraction’. So because some castles have an extra fractional person, the castles I chose to have a ‘shortfall’ were 1, 2, 4, 5 & 6. |
699 | 699 | 1 | 3 | 5 | 7 | 9 | 11 | 13 | 15 | 17 | 19 | 2V-1. Assets (troops) distributed in proportion to cattle point values. |
701 | 701 | 2 | 3 | 4 | 5 | 5 | 15 | 15 | 25 | 25 | 1 | I’m feeling lucky. |
715 | 715 | 1 | 3 | 5 | 7 | 11 | 11 | 13 | 15 | 16 | 18 | Allocated the same proportion of troops equal to the proportion of total points the castle represents. |
716 | 716 | 1 | 3 | 5 | 7 | 9 | 11 | 13 | 15 | 17 | 19 | I split all my troops up equally based on each castles point value. Since there were a total of 55 points between all ten castles and I was given 100 troops there was no way to split up 100/55 straight up. Instead, I went with the equation 2(points)-1= soldiers. This leads to having exactly 100 troops distributed among the ten castles while assigning troops equally among each point value. |
724 | 724 | 3 | 3 | 3 | 7 | 7 | 6 | 6 | 15 | 30 | 20 | My goal was to fight for every castle. A sizable investment in castle “9” and “10” was meant to punish any player who got too cheeky while also remaining competitive in the middle values. No castles for free to the opponent. |
741 | 741 | 3 | 3 | 3 | 3 | 3 | 11 | 24 | 35 | 8 | 7 | Last time but winning those. |
749 | 749 | 2 | 3 | 4 | 6 | 9 | 14 | 21 | 17 | 12 | 12 | Focus on the valuable middle to high castles |
753 | 753 | 2 | 3 | 4 | 6 | 10 | 18 | 24 | 1 | 31 | 1 | I wanted to obviously weigh the greater castles with more troops. I didn’t want to dump a lot of resources into 10 because people would target it. I also chose 9 instead of 8 due to previous results (in case that influenced other people’s picks) |
764 | 764 | 2 | 3 | 4 | 5 | 9 | 9 | 11 | 19 | 19 | 19 | Fibbinochi sequence |
767 | 767 | 1 | 3 | 4 | 13 | 15 | 18 | 1 | 20 | 22 | 3 | Go hard on 4, 5, 6, 8, and 9. |
778 | 778 | 2 | 3 | 2 | 2 | 3 | 8 | 22 | 22 | 15 | 21 | |
782 | 782 | 2 | 3 | 8 | 10 | 14 | 7 | 6 | 5 | 21 | 24 | If I don't know what I'm doing than certainly no one else will |
797 | 797 | 2 | 3 | 3 | 4 | 5 | 10 | 18 | 22 | 18 | 15 | I tried to ride the wave from earlier deployments and emphasize the trough in the middle. |
799 | 799 | 0 | 3 | 0 | 18 | 0 | 17 | 9 | 15 | 5 | 33 | The winning strategy in round 2 was primarily to take castles 4, 5, 9, and 10. I'm largely trying to disrupt that by using more force at 10 and 4. At the same time I'm trying to take 4, 6, 8, and 10 to get myself to 28. |
809 | 809 | 3 | 3 | 6 | 11 | 14 | 2 | 27 | 27 | 2 | 5 | |
810 | 810 | 3 | 3 | 4 | 15 | 16 | 15 | 18 | 20 | 3 | 3 | Not really sure |
812 | 812 | 0 | 3 | 6 | 1 | 12 | 20 | 24 | 23 | 6 | 5 | Following the logic of last games winner, trying to optimize against those who optimize. Without too much thought. |
814 | 814 | 3 | 3 | 6 | 15 | 20 | 25 | 25 | 1 | 1 | 1 | to attack the other guys |
827 | 827 | 2 | 3 | 3 | 7 | 10 | 14 | 18 | 21 | 18 | 4 | I figured I'd look at what strategy riddlers used last time. I looked at both the mean and the median. I started with the median set and increased most of the numbers 1. I also compared this number set to the mean. It won 35 of the 55 points. So, why not go with that? |
841 | 841 | 1 | 3 | 4 | 1 | 1 | 4 | 6 | 8 | 38 | 34 | A few simulations to find good strategies, and then searching for one that would perform well against those. |
846 | 846 | 2 | 3 | 4 | 6 | 8 | 10 | 13 | 15 | 19 | 20 | I started from simulating a tournament of 500 random players, that is, each players distribution of soldiers over their castles was uniformly sampled from all possible soldier configurations (at least I hope it was uniformly sampling from that). Then the top 5 players were taken and put aside. I then repeated this random tournament 99 more times to obtain 500 top 5 players. These players then competed in another tournament and I took out the top 5 players (top of the top players as I call them). Then I repeated this whole thing 99 more times to get 500 top of the top players. From these 500 top of the top players, I calculated the median placements for each battlefield. Then I repeated the above until I had a 10 medians for each battlefield. I took the mean of each battlefields medians and used the 10 means to calculate my strategy. I begin by allocating 1 soldier to each battlefield and set this as my starting configuration. Then I calculated the points per median soldier allocation for each battlefield. This would give me a way to rank which battlefields should be allocated to first. Going according to the highest points per median soldier allocation battlefield, I added to the battlefield the floor of the respective battlefields mean of medians. I went down the rankings until I ran out of soldiers or finished allocating to the last battlefield. If there were any remaining soldiers, I allocated one by one to the battlefield that had the highest points per soldier if adding one more soldier meant I won that battlefield. |
847 | 847 | 1 | 3 | 5 | 7 | 11 | 15 | 17 | 19 | 21 | 1 | Sacrifice the king, win the rest, and maybe sneak the 10 if someone sacrifices harder. |
849 | 849 | 1 | 3 | 5 | 13 | 17 | 2 | 14 | 16 | 17 | 12 | I just took an average of the distributions of the previous two winners |
Advanced export
JSON shape: default, array, newline-delimited
CREATE TABLE "riddler-castles/castle-solutions-3" ( "Castle 1" INTEGER, "Castle 2" INTEGER, "Castle 3" INTEGER, "Castle 4" INTEGER, "Castle 5" INTEGER, "Castle 6" INTEGER, "Castle 7" INTEGER, "Castle 8" INTEGER, "Castle 9" INTEGER, "Castle 10" INTEGER, "Why did you choose your troop deployment?" TEXT );