JudgeWargrave Posted February 17, 2019 Share Posted February 17, 2019 (edited) Hey, it's been about 6 months since the last time and I got inspired while learning Python to remake the auctioneer program as an exercise, so I felt like doing another race. Here's the last one for context. For those unfamiliar, the way this works is rather than selecting units one by one, we all bid privately on each character in terms of turn-count penalty, then they get assigned to us along with turn-count handicaps by a program. I've attached a PDF detailing the rationale and operation of the auction. I'll also attach the Python files. Create a text file with your bids each on a separate line (plus an extra 0 for the blank slot) and it will automatically create some random values for the other players and run the auction. I'd delete the line that does rotations in Main.py to save some time while you experiment. Here's the rules, I've bolded some parts that are different from usual. If anyone has strong opinions I'll consider adjustments. I kept the bit about survive chapters becoming rout chapters, the optional Gaiden turn-counts are reduced, Marcus is no longer split into two units, early game penalty is reduced, and what you can do with others' units is generally better defined and restricted. Team structure 1a. There will be five players. 1b. There are 34 units. One team will have 6 units; the others will have 7. 2. All teams automatically include: a. Hector, b. Merlinus, Ninian/Nils, and Athos due to their unique nature, c. Wallace/Gietz and Harken/Karel as they would only create dead team slots. 3. Bartre and Karla come as a pair. 4. If you get Marcus, he may only be used freely starting chapter 17x. Rules of play 1. The game will be played on Hector Normal Mode, skipping Lyn Mode. 2. Restricted actions a. You may only deploy other player's units on chapters with units recruitable by a specific non-Hector unit or desert items, specifically 14, 17, 19, 20, 22, 23, 24, 27, and 28. b. Another player's units under your control may ONLY move, recruit units (any), rescue/drop other players' units (not yours or NPCs), trade between chapters, and dig up items in the desert. c. They may not do anything else, including but not limited to: enter combat, use items/staves, trade/merch mid-chapter, visit (any building), rescue/drop your units or NPCs, steal, or build supports. d. Your units may do as you please without penalty. e. Mine/enemy control glitch is banned. 3. All Gaiden Chapters except 19xx, 23x and 28x must be visited. 23x and 28x are optional, and don't count towards the total turn-count up to 15 turns taken per chapter. The value is shortened from 20 which feels like too long. The last bridges needed to reach Sonia on 28x open turn 9 so even without water transport you can seize in 15. 4a. Chapters 13x, 15, 28, and 31 count the last Player Phase for which enemies are present for turns (or last PP if enemy not routed). 4b. Chapter 26 counts as 11 turns and is excluded from 4a so the player does not need to kill Vaida, and so that Vaida is actually usable. Penalties 1a. You may use an another player's unit as yours with a 4 turn penalty, per unit per chapter. 1b. The penalty is instead 3 turns for chapters 12, 13, and 13x due to forced deployment and to mitigate unfun early game. Exceptions 1. Matthew is free for Chapter 11. 2. Unarmed units may be attacked in Chapter 13x. 3. Lyn, Kent, Sain, and Wil may act freely in the southeast 5x5 box for Chapter 16. 4. Undrafted Thieves may be used to obtain necessary Promotion items. PM me your bids if interested, and post below so others can see how many slots remain. Please format your bids in the following order to make it easy for me to accurately create the bid file (example values given are the medians of the previous FE7 auction): Spoiler Matthew 3.10 Serra 4.00 Oswin 3.51 Eliwood 5.00 Lowen 11.80 Rebecca 1.00 Dorcas 4.00 Bartre+Karla 3.02 Guy 2.00 Erk 8.00 Priscilla 5.00 Florina 12.00 Lyn 2.00 Sain 8.30 Kent 8.20 Wil 0.26 Raven 3.76 Lucius 6.00 Marcus>=17x ? Canas 5.00 Dart 1.76 Fiora 8.60 Legault 1.00 Isadora 4.20 Heath 8.00 Rath 2.70 Hawkeye 2.76 Farina 1.00 Pent 5.00 Louise 1.00 Nino 0.26 Jaffar 0.76 Vaida 2.01 Renault 0.50 The SHA-256 hash for my own bids is a47d4291ad131a370119711f848893ef99cc909c90ecd120be5b8f63c4377b51. feAuctionDoc.pdf AuctionState.py GameData.py Main.py Pricing.py Edited March 11, 2019 by JudgeWargrave Mine glitch Quote Link to comment Share on other sites More sharing options...
JudgeWargrave Posted March 11, 2019 Author Share Posted March 11, 2019 Got a bit impatient so I just generated teams using my own bids for each player: A B C D E Lowen Eliwood Matthew Serra Dorcas Wil Rebecca Bartre&Karla Oswin Erk Lucius Guy Priscilla Florina Lyn Fiora Kent Sain Raven Marcus>=17x Hawkeye Canas Dart Isadora Legault Louise Heath Rath Farina Pent Renault Jaffar Vaida Nino --none-- 0.83 0.59 0.43 0.31 0 They are all within 1 turn with the handicaps so if my bids were reasonable these actually are all equally balanced teams, arranged in tiebreak order (E wins ties). Looking at them I'm not so sure, but I guess I'll find out. If people still want to submit bids that's cool and we can run the auction, but for now I'll just start LTCing with team A. Feel free to pick another team and race me too. I might stream this on Twitch rather than record videos to see if it's any fun, you might see me around there. I don't think I'll be adding anything here for my progress until I finish. Oh, and I remembered the mine glitch so I suppose to be comprehensive I'll add it to the rules section in the OP. Quote Link to comment Share on other sites More sharing options...
JudgeWargrave Posted September 3, 2019 Author Share Posted September 3, 2019 I've finished A and B, here's the comparison. B turned out to be 5 turns faster. Spoiler A B Another Journey 11 6 6 Birds of a Feather 12 5 6 In Search of Truth 13 7 7 The Peddler Merlinus 13x 7 6 False Friends 14 6 5 Talons Alight 15 4 2 Noble Lady of Caelin 16 7 8 Whereabouts Unknown 17 9 9 The Port of Badon 17x 3 3 Pirate Ship 18 3 3 The Dread Isle 19 5 4 Imprisoner of Magic 19x 4 4 Dragon's Gate 20 4 4 New Resolve 21 2 2 Kinship's Bond 22 3 3 Living Legend 23 3 3 Genesis 23x 0 0 Four-Fanged Offense 24 1 1 Crazed Beast 25 3 3 Unfulfilled Heart 26 11 11 Pale Flower of Darkness 27 6 6 Battle Before Dawn 28 5 5 Night of Farewells 28x 0 0 Cog of Destiny 29 4 4 The Berserker 30 4 3 Sands of Time 31 2 2 Battle Preparations 31x 5 5 Victory or Death 32 4 3 The Value of Life 32x 1 1 Light 33 3 3 127 122 Graphical comparison: I revised my values so that team B is worth exactly 5 more than team A: Spoiler 12.5 Lowen 8 Kent 7.5 Heath 7 Fiora 5.5 Eliwood 5.5 Canas 5 Lucius 3 Rebecca 3 Guy 1.5 Wil 1.2 Hawkeye 0.3 Louise 0.1 Jaffar 0.1 Renault Rebecca proved surprisingly useful in the early chapters. She contributed to the gains in 13x, 14, and 15, and while most of the early game units can do that, she definitely is substantially more valuable than Wil just for her join time. Her stats were also high enough to one round snags on Ch19 thereby opening passage through the center and contributing to the 4 turn clear. Heath's ability to carry Hector after promotion definitely saved a turn on the two chapters it was needed, 30 and 32, compared to team A with Fiora did, I think this is worth enough to rate him higher despite the later join time and poor base speed. In an average growth run his speed might have been more of a problem. Canas was actually better than Lucius. Reaching A rank staves was no trouble, particularly since I got 23x for team B. Canas's better magic cap probably wouldn't matter in average growths but I was able to get to 13 range by 30 which saved a turn. Also he is more durable but Lucius didn't have trouble surviving. Again, average growths might change that. Also Eclipse was actually useful on two occasions. A couple fun things I did with Canas was set up a torch square on 28 that allowed Eliwood to canto after attacking, which actually saved a turn, as well as warp Rebecca while riding a balista on 32 to increase reliability of killing Limstella on turn 3. Having two bow locked units was definitely a problem for team A and two sword locked units was similarly annoying for team B. Light Brand is just a bad weapon and its terrible damage output at range made Ch31 harder to 2 turn. I forgot Jaffar could use lockpicks until after Ch29 so I could have gotten more loot on the earlier chapters by using more chest keys as well as having Jaffar open the 3 island chests in 28x but it wouldn't have made any difference in turns. I didn't get Jaffar to S rank swords but maybe I could have with the Silver Blade from 23x. Also I tried to get Eliwood to S lances but failed, next time I'll just get S swords for him. Ch26 Jerme may actually be preferable to Ch26 Kenneth. Yes Kenneth is a seize map, but it snows for the first 4 turns; Jerme's only snows turns 1 and 2 (and 7+, but I finished both in 6). Jerme's map also contains better loot. 10 turns for Ch23x and 15 for 28x were more than enough, but I don't think I will lower it more as some teams might struggle to casually hit 9 or 14 turns and I would prefer players not hesitate to accept the side chapters if they have the option. Overall I enjoyed playing more FE7 but I don't plan to finish the other 3 teams at the moment. When I revisit FE7 it will probably either be with an average growth patch or on Hard Mode to increase the challenge. I tracked enemy attack speed and most are abysmally slow. Ch25 is particularly sad as 8AS doubles over 80% of the enemies. Next will probably be FE8. I want to try the QoL patches, and I have ideas for how to address the route split while simplifying the auction process. Rather than doing any automatic adjustments for join time, promotion times, redundancy etc, simply have the players submit synergy values for pairs of units (which can be positive and negative and adjust team values accordingly). This would mean slightly more work (don't need to submit values for every pair naturally, default value of 0) but also more control and transparency for the players. By having negative synergy values for units that prefer different routes, the system would assign teams more intelligently, and the same method covers limited promotion items as well as redundant duplicate roles e.g. flier, warper, bow user. Alternately the players could bid on the units sequentially but that adds logistical overhead. I also want to try adding a portion of the Lagdou Ruins to extend the game a bit. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.