Avoiding Akikawa's Phone Calls is the Key to the Umamusume Open League - A reflection blog
Umamusume is a speed game? Umamusume is a wit game? Or... is it an everything game? Nah, Umamusume is an 8199 skill points game! There is something extremely satisfying with manuevering builds every step of the way to min max the Umas, to make a composition that will achieve the max win rate against the perceived competition. It’s a bit funny at times, frustrating at times, intriguing at times, and now I want to share everything I experienced so far with the Gemini Cup from preparation to race.
Umamusume Open League Experience
With my best setup, I believe it should achieve a 98% win rate against the current player base, losing 2% of the matches against probably other well-studied min-max freaks, which I have unfortunately, yet to encounter in my 60 matches so far.
I went 18/20 with a suboptimal Mejiro McQueen / Nice Nature on day 1. 20/20 on day 2 after improving Mejiro and Nattie with only one close call, 17/20 on day 3 with 3 losses coming from an experimentation with a 3 front runners Ace composition. I was 5/5 with Mejiro / Goldship / Nice Nature and 10/10 with the same comp except Silence Suzuka instead of Mejiro McQueen. Totalling 92% win rate so far.
Here are my setups and win rates. Silence Suzuka is only 40% due to the 3 front runner experiment, and her only taking 1 out of 5 races there.
When it comes to races itself, there were few observations:
Victoria Frontier’s guide is misleading: He advised 600 SPD and 1200 Stam + 3 gold recoveries. In reality, based on Uma Musume: Pretty Derby race simulator, stamina doesn’t run out with 975 stam + 1 gold recovery or 800 stam + 2 gold recoveries. Of course, with debuffers/rush mechanic taken into account, it would be optimal to go for ~1050 stam + 2 gold recoveries in the Graded League. In the open league, overly sacrificing speed for stamina is not the way to go. Ideally, the umas should be at least 700/900/500/350/350 as an acceptable baseline. I ran into a few Umas that may have taken his advice (1100 stam + 2 Gold Recoveries, but only 500 ish speed and low guts/wit) and they were blown out badly.
Assuming there are two “optimal” debuffers in game in total… Four Pace Chaser stamina debuffing skills + Two stamina eaters would reduce 5% of the Uma’s HP, but one Gold stamina recovery is 5.5%. With All Seeing Eyes lacking the FOV to cast, there aren’t that many effective stamina debuffing skills. Having 1 extra gold recovery is enough to protect oneself against even the most optimized debuffers, e.g Mejiro McQueen at 975 stam with Gourmand and Swinging Maestro. If skill activation rate is 75%, in 56% of the games she will get both recoveries off and be untouchable from two debuffers, that is if they don’t fail and skill checks themselves.
Running into two players running two debuffers one ace comps? Technically possible, but players running two debuffers will naturally have a lower win rate anyways due to variances with Uma mood swings, or bad luck with race mechanics. They shouldn’t be incentivized to do this even in an environment where the mood swing mechanic is removed. Players shouldn’t prepare against strategies that doesn’t exist often enough to matter.
Pace Chaser Block: My Mejiro McQueen with 561 Power keeps on getting Body blocked consistently, lots of ppl running Pace Chasers with no stamina, a shit ton of power, and velocity skills such as Professor of Curvature for some reason. This led to some weird games where McQueen gets sandwiched and she is not being able to break out of the pack. My Goldship on the other hand, never seems to have this as an issue even though technically she needs to run pass 6-7 other Umas to get 1st.
There were some funny games where my Mejiro McQueen was stuck at 6th before the final corner behind Goldship, but somehow she manages to catchup and win the race with her speed being about 120 higher and both having the S aptitude.
Front Runner is kinda bad:The three Front Runner composition’s goal is to create the “Front Runner battle to take the lead” scenarios, which ideally would lead to them being too ahead for the other runners to catch up. What actually happened was A) Suboptimal players also ran Front Runners with high speed and high power that none of my three Front Runners can overtake. B) Front Runner speed up pack scenario didn’t occur and they just end up getting overtaken by other Pace Chasers / Goldship instead, whose builds were actually a bit under the basic stamina requirement. Two Front Runners + One Debuffer should be much better in this CM, but it is untested.
In the Silence Suzuka / Goldship / Nice Nature experimentation for 10 games. Silence Suzuka won 5/10, which impressed me quite a bit. However, her wins were a little ugly.
1 - Opponent doesn’t have Front Runner but a lot of Pace Chasers. In this scenario, usually all the Pace Chasers would be able to overtake Silence Suzuka before the final corner. Which lead to Silence Suzuka positioned around 3rd-5th. For Silence Suzuka to come back into the game, it would require Pace Chaser players to have stamina problems, which did occur every game as the opponents didn’t build optimally. Still, this is undesirable as it would lead to Silence Suzuka not being able to Ult.
2 - Opponent ran one Front Runner Ace. This is ideal because it generate the Front Runner competition scenario. Silence Suzuka being a bit more optimally built, woud eventually overtake the other Front Runner. If she is still first at the final corner and is able to activate her ult, the game is won usually. There were games where Silence Suzuka was about 12L ahead of my Goldship thanks to this race mechanic.
Therefore, what running just one Front Runner Ace achieve is that it removes Lucky Pace Chaser, be potentially OP if opp runs one semi-competent Front Runner, and would be more likely to avoid debuffers due to Front being a weak strategy. But it is far inferior in speed / pacing compared to Pace Chasers, leading to its ult being disabled against well-built Pacers if it is the Graded League. However, what actually happens is that since many Pacers in this CM doesn’t have enough stamina + my Nice Nattie is kinda GOAT’d, my Silence Suzuka manages to overtake the Pacers as a Front Runner quite often.
A few Tips and Thoughts on Run Optimization
The Ult Level Up Avoidance Game: Everytime you level up your ultimate, you raise its effectiveness by about 2-3%. If your Uma’s unique allowed her to gain 0.35 speed over 5 seconds, now it would gain 0.357 speed over 5 seconds after you level it up. This effect is so minimal that it is not even taken into account in Uma Musume: Pretty Derby race simulator. Therefore, it is always ideal to avoid levelling up your Uma as much as possible in the Open League.
To succeed in the Open League, the Uma star level must be 3 or maybe 1. Anything that is a 2 star or 4 stars automatically means inefficiency. For level 4, it would lead to the Uma’s base ult improving from level 1 to level 2, which takes 170 points away from your Uma at the start of the game for maybe about 5 points of value (RIP my Rice Shower). For level 2, Uma’s base ult is level 2 and it is worth 240 points, compared to a 3 stars level 1 of 170 points.
Therefore, characters whose runs allows you to easily avoid levelling the ultimate are the strongest. These characters needs a name, only thing that pops up is “Peasants” for now. Mejiro McQueen is the one long racer who is a wonderful Peasant, because she would always perfectly miss the fan requirements as long as you don’t enroll her on any races yourself. Silence Suzuka and most short distance runners can achieve the same feat as well.
Although something that makes Mejiro difficult is that she would need 2 Gold Recoveries badly due to her being a Pace Chaser, and it is hard to RNG the Swinging Maestro + another decent gold stamina recovery in a run.
While Goldship is the most OP character in medium and long races, her trait of being a massive Bright Light Merchant is her downfall. Thanks to her many extravagant planned races, her ultimate will be extremely difficult to avoid level up as it would take a perfectly mediocre build to narrowly avoid the 60K fan level up. The only method to at least keep the ultimate level at 3 is to become the #1 Director Akikawa Hater, as Green friendship bond (3/5 bar) is needed for Fan Fest to lvl up the ultimate. Couldn't find a guide on the exact math, I would say train with Akikawa just once or twice at max to avoid the green bond, and always avoid additional training as that provides 5 friendship points with her.
This does spice the Open League a little, because by the very virtue of picking Goldship as a character, she has ~330 skill points fewer compared to Mejiro McQueen / Silence Suzuka. This is partly how my McQueen/Silence have pretty much a 50/50 win rate split with Goldship, since they were able to allocate those additional points into other stats/skills instead.
Higher Score is not 1:1 with Effectiveness: This is very important for both long races and upcoming sprint heavy CMs. One may assume that a 1200 Stam Uma has 33% more HP than a 900 Stam Uma, but in reality it is only about 5% starting HP. For every 100 stamina points you raise, your Uma gains about 74 effective HP if it is a Pace Chaser, 79 for Front, 82.75 for a Late Surger and End Closer, starting from 3200.
If the Final Score reflected actual power level change, then Score gain would be something like 3 points in final score for every 1 Stamina point you gained. The way it actually works is that you start with 1 point score for every 1 stamina point gained, and it gradually increases whenever you reach a milestone of every 50 stamina. From 900 Stamina to 950 Stamina, every Stamina points gained is worth 4.3 points, it is 5.2 from 950 to 1000, and 6.6 from 1050 to 1100.
Let’s say you are playing Mejiro McQueen. When you raise your stamina from 950 to 1100, you lose out on 930 skill points to gain 222 effective max HP. However, if stamina stayed at 950 and a Gold Recovery was invested into instead, it would be 214 effective HP when the skill activates and it only cost 508 skill points for a Swinging Maestro. The other 408 skill points can be used to add ~100 points to speed or improve wit from 240 to 400.
Double circle green skill is worth 174 points, and it would provide you 60 base stat. 174/60 = 2.9. Therefore, once you reach 500 stat points on any attribute, it is always worth buying green skills. If you have high level support cards, it may be worth skipping training to buy Green skills instead, e.g when you have 900 stamina already, Standard Distance would do the job.
Characters Training Pros and Cons
Mejiro McQueen: Good at not levelling up her ultimate. But it took 11 runs for me to get 1 run where I get both Gourmand and Swinging Maestro. I only have one Super Creek SSR, which made her run harder. The run would be a bit easier with level 5 skill McQueen or a Vodka SSR power card, which would have offered decent stam recovery skills.
Mayano Top Gun: In between McQueen and Goldship. It is hard to not level up the Christmas ult, but she can easily avoid the Valentine and Fan Fest ult. Built in BOFA at lvl 3 and free Mejiro McQueen SSR makes her easy to build for any players. Her ultimate triggers too inconsistently though for her to be a good runner imo. She feels somehow worse than Silence Suzuka, who is a definitive worse version of Mejiro McQueen.
Silence Suzuka: Long E can easily be improved to Long A using Umamusume DB and follow a trainer with good Long sparks. She has no built in stam skills but that doesn’t prevent her from achieving 900-1K stam + 1 or 2 Gold Recovery easily. Her main issue is that if she lacks speed, she cannot win the URA finale as a front runner. I was able to win URA finale with 840 speed but consistently achieved just 3rd place with 780 speed. A few Yellow skills would have pushed her to win the URA finale, but that would result in her being unoptimal for this champion’s meet. Imo vanilla build with Green Skills + Gold Recovery seems to be the way to go.
She is a weaker version of Mejiro McQueen, but it is kinda funny watching her win a race that she doesn’t belong in. My one regret with her build is that I took Professor of Curvature over Cooldown because I thought she is already slightly over debuff range, Front has more HP, and Front debuff is rare. But Rush mechanic is real and some players are weird, so I think Cooldown would really solidify her as a consistent Ace.
Goldship: Golshi’s issue is that she is so op, when I tried matchfixing with her by putting her as a Front runner, she manages to win them all anyways… While avoiding Fan Fest is possible, this also leads to difficulty in having to skip out on good stats gains or Hishi Amazon’s Straightaway Spurt hint. Good Golshi is therefore kind of hard but a lot easier if you already have lvl 5 skill, and a “Great” Golshi is a gigantic leap to take because I am suffering from success with her as 20% stam is too helpful for winning races.
Nice Nature: Some guides suggested her to be an easy character to build but… She is not easy at all. Her runs are always on the verge of ending due to bad stats and 0% in speed and stam. She must take most skill hint training she can even if they are Guts, but the amount of skill points are also too limited to get everything I want in a debuffer. If Nice Nature had every good debuffing skills but also 700 speed and stam, then she would likely be able to actually activate All-Seeing Eyes that would hit at least 1 half-decent End Closer. Nice Nature being in the open league does help her a bit in that aspect due to limited speed of the other Umas, but usually she is still too far behind.
Since All-Seeing Eyes is not good, Air Groove should be better if you have her lvl 4 Flustered End Closers, this would mean that the Air Groove SR Gut card is not needed in that case, which brings so much more flexibility for the build.
All-Seeing Eyes is also useless to level up in the run itself because the condition to cast this skill is when Nice Nattie is already hard losing the race. Learning this early hurts the chance of passing runs quite a bit.
Lastly, Tools Used - Umamusume Rating Calculator Eng is a must use tool to ensure that your Uma doesn’t go over the 8200 points limit. One habit I have is pre-emptively calculating how many points I could end up with if I win the URA finale, and if I have too many points already, skipping training or matchfixing Ura Finale can both be viable strategies.
Uma Musume: Pretty Derby race simulator can be helpful when it comes to evaluating the needed stamina requirements. I also use it to get a basic idea of which skills bring the most value than others.
And then just YouTube stuff. Although most guides assume one is playing in Graded League, the concepts they teach are still helpful, especially when it comes to character evaluations and how the acceleration mechanic work on different tracks. 71M3 is the one I watched to understand the Gemini Cup.
Shower Thought - I think if a player has a really OP deck to the point where they can afford to skip training turns just to make sure that the stat doesn’t exceed the 8200 mark, one possible thing to try is highrolling fail unrelated races. E.g getting a “Summer Runner x” is irrelevant when the Champion’s Meet is in Spring. Negative/Purple skills reduce the total final score by 129, which is exactly the same score as a level 1 positive green skill.