Transcript
Speaker: I think this transparency is a big one. So highlighting because of the lack of centralized nationwide reporting requirement, Japan's off listing market is larger. Agents sometimes avoid listing property owner properties on range to score double commissions by representing both the buyer and the seller. The US MLS has strict rules and massive transparency mandating the participating agents to share listings to promote cooperation. i think that's a really big difference.
Speaker: Welcome back to the buying a house in Japan podcast. Today, we are going to be talking about something very, very interesting. This has been in the works for, I think like six months, maybe a year, right, Jerry? A year ago, we started discussing this seriously.
Speaker: i started work on it six months ago. Okay. So today we're going to discuss the new changes to Acumart, what we call internally as the V4 update version four. oh boy, we've come a long way, but before jumping into it Joey, how you been, where are you at in the world? What's going, what's been going on with you? I think this is our third podcast episode of film, uh, we've done since I've been back in Nova Scotia temperature has been up and down. It's a super El Nino year, which means California, your side is going a lot wetter and East coast is going to a lot drier. We got to bring on, your buddy, Chris in New Mexico. This is his like jam, right? He like understand these, these climate patterns.
Speaker: Yeah, he he analyzes a lot of climate patterns for people that like to speculate on, i guess, produce markets or like so people that are buying and selling coffee beans or like bananas or that kind of stuff. So people trade that like people trade stocks. And so he i don't know if he's doing that work right now, but he was kind of like a freelance consultant analyst for people that were making those trades.
Speaker: We just talked to someone actually, one of the podcast listeners, Richard hit us up. He's a coffee trader. Uh, he gave a couple of good recommendations of coffee shops. So shout out to Richard. If you are a big podcast listener, anytime, feel free to reach us out, reach out to us at podcast at Akia dash mark.com and say what's up.
Speaker: But onto our main topic of the day, Akumar V4, Joey, what, what do we got? What changes are happening on the website? So about a year ago, we were discussing a lot of people were reaching out. People would complain about, they would zoom in on a specific location and you would see, you know, maybe duplicate listings twice or three times or sometimes even four times. In fact, I found one property. It's, it was somewhere in, i think Shinjuku and it had 495 properties.
Speaker: duplicate listings. So people would rightly complain, you know, a lot of the times a little bit, the pin on the map would, if if the location was the same, the pin on the map would kind of like overlap and you wouldn't actually see it on the map, but you could see it on the kind of listings on the side.
Speaker: This was the problem we were trying to tackle. And really the reason why we had so many duplicate listings is just the nature of how the data is structured and where the the data comes from. So right off the bat,
Speaker: we scrape a few different websites, Japanese real estate websites out there. And so you may get the same listing twice if it's all, if it appears on two or more source Japanese websites that we're scraping listings in Japan are true to their name. It's more of a listing itself. It it doesn't represent a property. It's more like a Craigslist ad. So let's say you're trying to, you know, sell,
Speaker: your lawnmower, right? You could create one Craigslist ad. You could create two Craigslist ads. You could create three Craigslist ad, you know, and that's what Japanese real estate agents tend to do. Actually, sometimes it might be playing around with the price. Sometimes they might be creating more than one listing to get it in sort of like ah maybe there's like a hot fresh and new category on these Japanese websites. And they're trying to like keep bumping their property up to the top of like the new list. So it's less about like, Hey, this yeah URL and this listing rep is the representation of this property. And there's only
Speaker: this one listing, it's more like a Craigslist ad where agents are creating many of them. And that problem was compounded by the fact that we scraped several websites. So that problem just gets, you know, multiplied by the number of websites we're scraping. So that's why we had so many listings. And to add a little bit more context to at least the U.S. s or Western kind of viewers here.
Speaker: On the underlying kind of like real estate database backend of Japan, it's, it's, it's the U S has the MLS, which is pretty, pretty good. Actually. It does a very good job of identifying parcels, sales history. it has all that data in MLS.
Speaker: Japan has the equivalent called reigns. Reigns is much more archaic to our understanding. Joey correct me if I'm wrong, but they seems a lot more disorganized. and that everyone has their own pdfs of their properties that are unformatted unformatted i feel like the data structure is just a lot messier and harder to parse which is kind of why you know when when we started this site it it is kind of building from scratch like there hasn't been a good data source for us to pull parcel data, land size, sales history, all stuff like that. So it's like over the past couple years, we've been gathering that data.
Speaker: Some of these real estate websites might have it on the backend, but it's not public public government data. So when you look at the US websites like Redfin, homes.com, Zillow, they're all kind of pulling from the same clean data source, which is kind of one of the big problems. So Joey mentioned here, like an agent could post their Craigslist type real estate listing like a hundred times just to grab attention. But it's not like regulated by, i think, a government source at this point.
Speaker: and MLS in the United States, multiple what is it? Multiple listings service. So I believe it is based on listings as well, similar to like what I'm complaining about just now about Japanese real estate listings. And I bet Reigns is the same thing. It might be fundamentally on that listings level. The difference is companies like Zillow, Redfin, et cetera, have gone the extra mile to consolidate all that data and try to form these you know properties and then display those properties to the users. Um, whereas I don't think any company in Japan up until now has attempted to collapse all these listings down into a specific property and then show that property to the end user, instead of showing, you know, the multiple listings for that specific property. And I think that the problem is worse in Japan and harder in Japan, because I think that fundamentally those listings are messier.
Speaker: So even if reigns and MLS both work on this concept of a listing, and not on the concept of a property. i think the data on the Japanese side of those listings is much messier. So one, I think it's disorganized in terms of like free form text input instead of like, you know, a well thought out database schema. Like for example, the layout of a house in Japanese is is frequently free form text and it's super messy. And, you know, part of that problem is cleaning up that data and collapsing it down to like a three LDK or a four LDK. When a real estate agent has written a sentence instead of just writing three LDK. So that's part of the messiness of the Japanese data. But also they don't, I think the the main difference between the US MLS and the Japanese MLS is that Japanese real estate agents obfuscate certain things about the property listings on purpose. For example, the address is never fully there. You never know the exact location of these listings. The the Latin longitude pin, again, obfuscated much of the time. So it's very difficult and and given the Japanese data to collapse these listings down to a specific property, given all the data messiness that we have on the Japanese side that I would guess is much less of a problem on the American side. I just pulled this Google search up and I think it does a good job of what you know range versus MLS here. One MLS is direct public access. So portals like RedZillow, Redfin. So we don't get that public access to RAINS. RAINS is strictly closed brokers only database. So our agents, we have asked them to show us through RAINS. And it's not ah an enticing UI to say the least, but we have seen it and it's a little bit more closed off. Like, do you think on the RAINS side, they have duplicate listings or do you think all those listings are
Speaker: tied together somehow. I think they have duplicate listings too. I think this transparency is a big one. So highlighting because the lack of centralized nationwide reporting requirement, Japan's off-listing market is larger. Agents sometimes avoid listing property owner properties on range to score double commissions by representing both the buyer and the seller. The U.S. MLS has strict rules and massive transparency mandating the participating agents to share listings to promote cooperation. i think that's a really big difference. yeah And you'll see, as we explain more about what this V4 system is, you'll see transparency is going to be a theme, uh, running through this whole thing. So basically the challenge that we were faced with was we have all these listings that exist and they're not tied together to represent any specific property, right? Two listings may be the same house. They may not be the same house. And we're trying to figure that out. We're trying to I've described as collapsing. So we try to collapse listings together, aggregate them together, um, to form the, this idea of a property concept that I don't believe is done anywhere else in Japan. And I don't even believe that reigns is doing this. Sumo's not doing it at homes is not doing it. I know some of our competitors out there, they do things like D duplication, which is removing listings that are clear duplicates of, um, another one. That is not what, that is not the path that I decided to choose here. I wanted to go a little bit, one step. more and create this idea of a property concept. So instead of just removing listings that are duplicates, we're going to aggregate them together and form a the idea of a property concept. So going to take all the data from all those, um, associated listings and then create a property concept and any new listings that come on the market, they're going to be tacked on. and then the property data is going to be updated. Yeah, shout out to Joey. He really did not take a shortcut here. When we were getting a lot of complaints about the deduplication, like, Joey, can we just like do like square foot area size year and just eliminate listings? He was like, no, that's ah that's ah that's a bandaid. I want a solution. So hats off to you, Joey, for getting taking the harder route.
Speaker: And so the real reason is I think that this path opens up a ton of possibilities. And I think that's the real, um, I think exciting thing here. Um, but let's just score some quick wins before we talk about, uh, the, the, the wonderful world of upcoming possibilities where I think the real meat is, but what this actually, what grouping all the listings together to form a property does is we can see, um, when a property price has changed over time. So that's the gray line here. So that property had no changes in its listing price. So as listings come and leave the market, I'm checking what prices they're at. we Yeah, so here's a good example. And I'm checking the price of every listing every day as well. So you can see that for most of this history, there were two live listings here. some time there was a price change down from 24 million to 23 million. And we can now track that on this property concept here and we can see um a lot about how the listings price has changed over time, which I think is really cool because You know, I'm looking at properties mostly out in the middle of nowhere. And you can sort of see, I would describe it as like seller confidence that they're going to be able to sell their house. And you can see them slowly dropping the price of the home. We had one of our, actually someone has been on the podcast before, Nin, who's a property investor across Japan and America. And he was talking about the price increasing in places in Tokyo. So I thought that was interesting to see because mostly my experience has been price prices decreasing in different places, but um Nin was saying prices increasing. So so a lot of this V4 collapsing listings into a property concept, that's going on under the hood. It's as if you know you you got your engine swapped out on your car. You're not going to see any differences on the car itself when you're just looking at the car, but it really comes out you know how the car works internally. And that's that's really what V4 is about here. So there's not too many large changes that you're going to notice on the website right away other than um you know the price history, for example, and a few smaller things. But it's really the engine of the Akiyamart website now is completely redesigned around the idea of a property and not on the idea of a listing anymore. So so for you guys who are just looking at V4 now, um let me show you how to get to associated listings. So you're on a listing right here. Go ahead and go down. You'll see these associated listings right here. You click this and these are basically all the Craigslist ads that Joey was kind of mentioning. but You can see these are the multiple listings. So if you will are curious to see them, click here.
Speaker: Again, under links and resources, this is where Joey has compiled all these listings. With that, Joey, we've heard feedback that we have a ton of photos.
Speaker: So, Joey, at this point, you're kind of aggregating all the photos into one listing, right? So there might be a lot of photos, but we think this is better for the buyer. I'd rather see more photos of place than less.
Speaker: This is a feature, not a bug, right? Take this property we're looking at right now. For example, there were five or six listings associated with this. So for every data point that we represent here on the property, i had five data points that I needed to pick. Well, which one do I show on the property when I have five different data points from those five different listings? Right? So depends on the data point, how I actually collapse that down and those five values down into one value. So with photos here, all I did was I just added them all together. So I had photos from five different listings. I just took all the photos together. So there will be some duplicates for the price. I took the cheapest price at any given time for something like the Midori or the layout.
Speaker: I believe I took the mode. So the one that was shown the most amount of times I put a lot of thought into what the data is, how the data is likely to be represented and then what's the best aggregation function to take that, you know, those five data points across five listings and then aggregate them into one data point for the property. So the aggregation method depends on the actual data point.
Speaker: So let's take a look at this, this house that we walk by. So if you've listened to our podcast, we actually filmed a podcast with Tomo, like right next to here. There's like a little river. So I'm I've, uh, if you've never tried the satellite mode, try satellite.
Speaker: This house is 60 K in that boot and it near the Haruki. I would say if I had a hundred K more, I think this could be a really, really dope property. He would need a lot of renovation, but. And just to show you again, we have the associated listings here. We have, as Joey mentioned, the price history. A lot of this price history is also the yen to dollar. So keep that in mind. Like it's not the price actually fluctuating, but if you see a huge drop or huge raise, those are probably it being relisted. Yeah, I wanted to mention that story. So there's two lines here. So in your selected currency, so I'm guessing Taki is looking in US dollars here. So the green line is going to be multiplied by the exchange rate on that day. The gray line going horizontally through. that's the yen price so this house has not been relisted at any other price other than 9.7 million yen so the gray line is flat and then you basically just see what the exchange rate has been um with japan when multiplied by the property price of the house but if you do go to a a property that as we saw earlier where the price was lowering down um you will see the exchange rate or basically the dollar valuation tracking that um those drops or raises in the yen price so green line your preferred currency gray line is the yen denominated value of the house the blue part is when the property actually came on the market and you can see at the top zero live listings and then as soon as you enter the blue part you see two live listings three live listings etc so that's when the property is actually on the market before that the property was not on the market yeah this is awesome to see as a active buyer uh specifically investors who are looking for kind of
Speaker: not distressed properties, but like, Hey, someone's done a price job. Maybe this is a good time to come on in. Just as a reminder, low balling in Japan is not too welcome. We've burned some bridges ourselves. So typically this price is really what the seller expects to get, but sometimes maybe 10% negotiation is okay, but each case will vary. But this is actually another data point for you to kind of go in and understand historical context, which is like,
Speaker: Absolutely huge. I don't think any Japanese real consumer real estate website allows this at this point. Yeah. As Taki mentioned, there's no Japanese real estate website out there at this moment, maybe not even reigns where you can see, oh, this was, this probably was listed at these different prices, um, over time. And it's been going up or it's been going down. I think this Japan first data point. I think that's really cool, but there's more that we can do with, uh, this idea of the property concept instead of just removing duplicate listings. collapsing everything together into a property opens up a lot more doorways apart from just this price history. It's cool now that we have this data. So one of our goals here is to have a very clean data set where maybe we can predict and see these trends before the Japanese government can, you know, with with whether listings are
Speaker: are going off or coming on and at what price. i don't but My head can't wrap my ah wrap around all the data, but I think there's some pretty cool insights we can draw from these from this data.
Speaker: Yeah. And again, all these prices and even the price chart, this is the listed price. Like this is not the price that a house sold for. We don't have that data. Nobody has that data. Maybe Raines has some of that data, but it's incomplete at home at BB has it. There's a couple like ah licensed agent portals that have some data, but it's hard to verify. Like it's not a government.
Speaker: Actually, the government should know. I think, but I don't think they keep track of it in a navigatable way for or us regular folks. If you guys have suggestions, please let us know. Again, and we're a small team, so really appreciate everything. All the feedback, podcast.akia-mark.com.
Speaker: Joey, what are you excited to build once we have this? what's What's next? Now that we're moving more towards this kind of like parcel ah kind of mentality, what' what's on top of mind for you?
Speaker: Yeah, I think the cool project, um, that I'd like to build now that we have this property concept would be, um, I think that we could build a model that, um, provides valuation or like a kind of like a continuous valuation like Zillow does with their zestimate. products if you go on to any listing on Zillow. So these ones are all for sale. But if if you zoom in further, you'll start to see little gray bubbles come up or yeah see on those. So those are as these properties are not on the market. These are estimated property prices that Zillow has come up with. They've based it on whatever, maybe the property sold in the past or even if it didn't sell in the past, they know some specifics about the lot size, about what the property actually is. If it's like a two bedroom or whatever, and they provide a valuation. So a continuous valuation. I think they do this maybe daily or weekly. They will update the value of all these off-market properties that they have.
Speaker: And I think we could begin to get to a place where we can build this too. So Right now we don't have data about properties that aren't on the market. Um, but I think it is possible again, this is really stretching the limits of Japanese available data out there. There is some data on land parcels. It's gonna be really tough to connect that to properties and figure out more, um, information about what kind of structures on the property, if any. But this is, this is an absolutely gargantuan data project here. But again, I think it's so cool to be able to one as a buyer, get a second opinion on a property. If you see a house, for example, that house and hero with the chandelier and stuff like that, they want, or that, that broken down house, they wanted 2 million for in hero, you know, what does objective independent valuation service think that this property is priced? And then also, if you wanted to just kind of snoop around, I think one of the very cool things about Zillow is you can look at properties that aren't on the market, maybe the house next door, maybe the house down the street that you'd love to buy one day and get ah get an idea of what that property is worth to to dream, you know, to be like, oh, someday i would love to get 572 Palisades Avenue and, you know, right by that eucalyptus forest or whatever. And all I need is $5 million, you know.
Speaker: Bro, I need $1.3 million to buy this two-bedroom. $1.3 million. But, like, it's really cool for this, like, dreaming factor. I feel like that's what is really exciting, to be able to look at properties that aren't on the market and sort of, you know, dream about it. Like, okay, and yeah I need $1.3 million dollars for this place.
Speaker: I think you as the audience can have now seen some of the features, but yeah, very, very cool on what we can build from here. I think truly unnavigated, untracked Japanese real estate territory here.
Speaker: The real estate parcel game is is like an impossibly huge task. We've gone down a rabbit hole trying to figure out this data. The data just doesn't exist for a lot of Japan. So it is going to be like, Can we, how do we get this story? Like any, any ideas? Yeah. Land parcel data. So like geospatial data for about 50% of Japan is there. i think 30% is digitized and in the correct, correct, which probably is not that correct latitude and longitude. The other 20% really just the. the geospatial polygon, but it's not mapped to any, any like latitude and longitude on the map. And then the other 50% just doesn't exist. It's all paper maps. So this is a a very large, this this is like the forefront of Japanese, you know, geospatial mapping.
Speaker: But i think it would be really cool. And maybe we start out with specific areas of Japan first, you know? Yep. Well, it's got it, Joey. This has been long time in the making. For listeners who are, know, been longtime Akiemart users, please give us feedback here. If you're UI UX person, you think something would make your life easier. A lot of these features have actually been born because of customer feedback. So let us know what's working, what's not working. We all will try our best because ultimately our goal is to make the best tool possible, you know, and I think for us, it's
Speaker: Uh, we can navigate it one way, but you know, if you have a certain way you or feature you want, we'd love to hear it. Change can happen out there. Keep, keep bugging us. If there's really something that you think we need to do. Great work so far. Anything else top of mind on the website before we wrap the pod?
Speaker: This is like a big project that happened under the hood here. There's a few places where you can see the, the, the fruits of that labor, like the price history. But, um, yeah, this was ah a very large project, but like with a lot of backend software engineering that you don't see result. Like it's a, it's a total success when no one notices any change. That's when like, it's, it's a big success, success, kind of like your mechanic, you know, fixing something on your car when you don't notice any, or like, you know, when when it keeps running smoothly, that's a successful mechanic.
Speaker: Hell yeah. So I think mission success has been a week so far. By the time this pod comes out, it'll probably be a couple weeks. So again, keep the feedback coming. If you are new to Akimart and have not tried our pro subscription, we have a deal for you. We are going to be issuing ah a promo code. It's going to be V4baby.
Speaker: This will get you $50 off our annual subscription. This will be for first time subscribers. So I'm calling out if you've already subscribed. No, you cannot unsubscribe and resubscribe.
Speaker: This is for people who have not yet tried it out or are on our trial ah for this pro subscription. So promo code V4baby, give it a shot.
Speaker: But other than that, thank you guys for listening. We will have some more tech improvements hopefully soon.
Speaker: Thanks for listening to this week's episode of Buying a House in Japan. If you find the show helpful or at least entertaining, give us a five-star rating. Takes five seconds and really goes a long way. If you or someone you know is looking for a Japanese home, check out Akiyamart.com, our English-friendly site that lets you browse over half a million listings. That's A-K-I-Y-A-M-A-R-T.com to find your dream home in Japan. If you've got questions or feedback, drop us a line, email contact at akiamart.com. You can also find us on Instagram as well as TikTok, where we post properties, tips on the home buying process, and updates on our own purchases and renovations. Thanks again for listening, and we'll see you next week.
