Square has released horse racing prediction software! ? Do you know "Power Stakes"? [Feature]

These days, the name of a racehorse is running around on social media. There are various impressions of horses and interesting episodes introduced, and some readers may have become interested in "horse racing" along with the momentum.

By the way, did you know that Square Enix actually released software dealing with "horse racing"? A lot of people would probably say, "Square Enix horse racing game? Ah, 'Chocobo Stallion'!" However, this time we're not talking about chocobos. The theme is real racehorses.

In fact, that software was included in the game launch confirmation information on the retro game compatible "POLYMEGA" sample machine that was previously posted on Game*Spark. The software in question, "Power Stakes," is a slightly different genre of "horse racing prediction software" released by Square (at that time) in 1997. In this article, let's approach this "Power Stakes".

What is “Power Stakes”?

Horse racing prediction software produced in cooperation with Square and the horse racing newspaper "Keiba Book". It was released on April 11, 1997, and mainly contains racehorse data and race records around 1996.

According to the manual, the feature of this work is that it reflects "prediction data set by the player" unlike conventional horse racing prediction software. By freely manipulating the items of "condition", "leg quality", "development" and "estimated time", a big change will be born in the race result. It seems that players can enjoy predicting virtual races by assuming various situations.

After that, "Power Stakes Grade 1 (1997)" and "Power Stakes 2 (1998)" with additional input data items were also released. By the way, in 1997 Square released works such as "Final Fantasy 7", "Saga Frontier" and "Front Mission Second".

Hard interface! Data entry is surprisingly easy

There are three main modes in this game. It is divided into "registration of racecards" for entering race and racehorse data, "running" for simulating input data, and "movie" for looking back on G1 in the spring of 1996. The movies are from Oka Sho (Katsuma: Fight Gulliver) to Takarazuka Kinen (Katsuma: Mayano Top Gun).

In "Registration of the race card", first enter the expected race information. The race name (you can leave it blank) and the distance, as well as the conditions of the track and the estimated time, will greatly affect the results, so be careful. In this work, there is a manual saying "Be sure to prepare a horse racing newspaper". After creating the race data, it is time to enter the data for the racehorses that will run.

Racehorse data can be created by searching the racehorse data (approximately 6,000) recorded in this work, or by entering it yourself. In the case of new data, "Name", "Affiliation (Kanto/Kansai/Regional/Overseas)", "Classification (father's domestic horse, etc. See JRA official website for details)", "Age", "Class (from new horse to opening)". Enter "gender" and "hair color" to register.

In this game, the ability of a horse is determined by entering data from up to three races that horse has run in the past. The previous run data can be entered quite finely. After collecting the horse data and past race results, enter the "jockey", "weighing weight", "leg quality (e.g. running away, running, etc.)" and "tone" on the race card. The condition will be the "player's input part" that affects the race result. It is a part that judges by comments such as horse racing newspapers.

After entering the data, return to the top screen and press "Start" to enter the virtual race. Now comes the moment of tension.

Race is more than just watching! Create an expansion!

On the screen during the race, you can switch cameras and select the horse you want to focus on. You can double the speed by pressing the x button, so if you just want to see the results, just keep pressing it. The same data always gives the same result.

During the virtual race, players can freely change the "final corner positioning" of each horse. By changing the positioning the race is re-simulated and the result is changed. In addition, the "leg quality" and "condition" on the race table screen are also greatly reflected in the results. It seems that these players put a lot of emphasis on the "prediction part".

However, since the development prediction is only for the "final corner", it is difficult to predict a large development. The system that tries to forcibly return to reality even in a fairly unreasonable development is excellent.

Let's reproduce it by actually inputting data!

Well then, let's check the power of this software. The sample race is the "February Stakes" held on February 21, 2021. In addition to the JRA official website, I referred to "horse racing book" and "netkeiba.com" for data entry. In addition, since the notation method for horse age has changed in 2001, it is entered in the notation at that time.

What I felt when entering racehorse data was that there were no jockeys. It's been 24 years since this work came out, so it's natural, but most of the people who were riding in the race the other day are not registered in the data. Fortunately, you can register in "Kanto", "Kansai", "Local", and "Overseas", so there is no problem with the input itself.

Also, even in the previous race data, there were problems such as "There is no race" and "The distance of the registered race has changed (Negishi Stakes)", but somehow I completed the input without any problems. He immediately tried simulating it, but Katsuma became 'Wonder Riedel' instead of 'Cafe Pharaoh'. After that, even if I tweaked the tone and development, the stronghold of "Wonder Riedel" did not collapse after all.

As a result of the final adjustments, the predictions of the horses themselves were correct for up to 5 of the top 6 finishers (although the order was inconsistent...). In that sense, the weight of "the part where the user inputs" in this work seems to be quite large.

Regarding the "Power Stakes" that I have introduced, I have the impression that the data is reflected more firmly than I expected. There were a lot of adjustments here, but I was surprised to see a decent result in the Virtual February Stakes in the end.

Of course, I don't think this work can be used for serious predictions even in modern times. Twenty-four years is cruel and too little data. There are too many jockeys and new races, and even horse age notation is different from now.

However, as long as I am happy to see the races of those days and the gallant figures of those days such as "Hishi Amazon", "Air Groove", and "Machikane Fuku Kitaru". Such a database of yesteryear may be fun for those who want to remember the past or want to know detailed data of racehorses whose names are recently seen on SNS.