ORATS - Driven By Data | Episode 130 Aired: May 05 2026 Title: Using The New ORATS CLI To Analyze Data With AI Video: https://www.youtube.com/watch?v=8W6lvn78WmQ ============================================================ Hi everyone and welcome back to another show. My name is Tyler and I'm with Matt, the founder of Orats. Orats is an options analytics platform built on top of over 15 years of historical options data. Our dashboard is integrated with Tradier, meaning you can research and discover trades on our platform and then seamlessly send them through to your Tradier brokerage account. Today on the show, we're going to be announcing and demoing our new Orats CLI for AI agent integration. I'll actually be sharing my screen today and doing the demo. What does a CLI stand for, Tyler? So, a CLI is a command line interface, meaning if you are a developer or even if you're not a developer, you can use the CLI to in the terminal. It's a kind of a terminal-based integration. You can ask questions, you can write commands to get data. It's sort of like using the API inside the terminal. So, you're kind of bypassing a web browser, right? And you Exactly. And you're searching don't need to Yeah. Yeah, you don't need to write a program or go in the web browser or do anything like that. You can just do it straight from the terminal. Yeah, well, I'm I'm curious to learn it. I'm going to be learning along with our other clients. This is a new product for us old fossils, chunks of coals. Well, we'll be learning how to do this. And I'll be learning with you. All right, great. Yeah, so we built this as an AI agent-first product. So, you can integrate it into your cloud or Codex environment. So, I have my terminal open here and this is where I'll be doing the demo. Um, I'm using Claude code, but you can easily use this with Codex or any other sort of open source agent. You can also use the CLI just on your own. Um, without Claude or Codex and you can just type in the commands and and get the data. Uh, but today I'll be demoing it with Claude. Um, there are three steps to install it. Um, so you install it from the NPM package with this command. Uh, you you do have to have a API token, uh, which you'll put, um, in your home directory, uh, with this command. And then we also have a skills package that you'll install. Um, and this helps the CLI uh, and your AI sort of understand what's going on behind the surface. So, you can kind of think of the skill as uh, something that we taught Now, we have the O'Reilly University. Um, we basically taught uh, uh, but put this all into one document and taught the AI how to use the CLI um, with all this methodology and stuff like that. So, that's kind of what the skill uh, the skill does. It's sort of a reference to our documentation, our data definitions, um, how we uh, do our calculations and our methodology and our history and all that kind of stuff. So, it provides the AI agent the context to be able to make uh, good decisions. Does that make sense, Matt? Yeah, but let me just ask stupid questions. Uh, on the install, I always get that like everything you said makes sense, but like where I am I? Like would I go to the command line to do an install or do I have to do things before the install? Yeah, so this is, um, before the install you'd get your API token from O'Reilly. So, Okay, I that. So, but then like is that Okay, you have That's awesome that you have that copy thing. So, do you open this command like CMD thing or Yeah, so you'd open your terminal. You can use any sort of terminal app on your machine. I'm on Mac and I'm using iTerm, but you can use any sort of terminal app. And then you'd go into your home directory. Home directory meaning like like the slash one or something or Exactly, yeah. So, this is this is my home directory. Tyler's MacBook Air. And I'm there in the slash and I just paste um I paste this command. What if it says it's not recognized as an internal or external command? Um then uh not recognized as an internal or external command. Um Oh, that means you have to install NPM. Yeah, what is that? Let's talk about that. How do you install NPM? Yeah, so NPM stands for Node Package Manager. So, you do have to have Node.js installed on your machine, which some developers will already have installed, but if you don't have that installed you can just go to download Node.js and you go here and you can you can either type this or or download the pre-built installer um right here. Choose Windows or Mac and download the installer. And this should be this should be everything that you need. You just have to install this from the Node Package Manager. So, that's why you need Node. Does that make sense? Yeah. So, I'm I'm doing it along with you. So, as we go along So, then it goes to program file Node.js. Is that where we'll be running it from or does it go everywhere? No JS will just install onto your machine. Okay, so you I could do that still from the home page once you get Yeah, I recommend installing this from the home directory. The uh dash G uh parameter right here will install it um globally. Okay. Yeah. And so then you'll be able to kind of Cloud will be able to access this from from anywhere essentially. Okay. Um should I get into demoing some of the features? Um if you want to like I've still not in but that you know it's um You could you could do that. So I'm I'm still on the home page. So I have a like it's not I just installed it and it's not getting it but like um if you want to go and then come back and help me out with that but it's program files No JS. Were you able to get No JS installed in um Yeah, I see it. Verified. Yeah. Okay, then So if if you do npm -v it it should print 11.12.1 or something like that. Does does it do that if you do that in your terminal? Uh so I'm seeing like So No No JS is there. I want to Do I need to go to Well, we let Why don't we Why don't you go through your demo and then after the demo you can come back and deal with my and you know kind of personal issues here. Okay, sounds good. Yeah. Um you know, maybe um I I I plan to write a blog post here shortly with instruction like a deeper dive into the instructions and the downloads. So um maybe I'll I'll put it in there kind of like a walk through of that. Okay. Um Okay, so there's some examples here. Uh so, I have my Claude code open. Um in order to sort of invoke the uh the skill, uh I guess is what what you're invoking here. You just type {slash} Orats. Uh and that calls that essentially tells Claude, "Hey, use use the Orats skill. Use the Orats CLI." And you can type in kind of whatever you need. Um you can start just by asking, uh "What can the CLI uh help me do?" Um I guess this is a this is kind of a great a great start to figuring out what it can what it can do. So, um it has access to live data, um delayed cores, and um IV rank, and some of the delayed endpoints. It has access to historical uh end of day and intraday data. Um and it's also familiar with common workflows like earnings analysis and skew analysis, um and also some um gotchas and stuff that uh users have ran into. So, it's familiar with that. So, um it kind of knows a lot of those nuances. Um yeah, so you can ask it kind of what what you can do. Um You can ask it anything about methodology or data definitions, but my favorite thing to do is just ask it um you know, questions about historical data. Um cuz it it can really dive in and and get some some good information. So, I'm going to ask it to replay the VIX spike on August 5th, 2024, um and then what the SPY zero DTE at the money straddle was intraday uh during that during that VIX spike. So, you'll see what it's sort of doing. Um this is sort of the CLI portion of it. So um we we taught Claude how to use the CLI with the skill. That's what the skill is doing. Um and the CLI is just calling this is basically like calling the API. You can think of that as what you would do in a in a URL. You you'd type in um the URL for the API endpoint and then pass in the parameters like the trade date and the ticker. Um and then with the CLI you can also filter and do some things like that. So uh once it was able to uh do that it gave us our output here and and told us the uh at-the-money strike and um the prices of that and everything. Pretty cool, huh? Yeah, that is sweet. Yeah. Yeah, so we have a lot of um a lot more websites a a lot more examples on the website. Um let's do another one here. Um this is a good one. Uh is Nvidia's implied earnings move rich versus last eight quarterly earnings moves. So it can do earnings analysis. Uh it can do historical earnings analysis. Uh it's really good at earnings analysis as uh some of you may know Orats has really great earnings indicators. We have ex-earnings volatility, we have historical earnings moves. We can compare earnings to um uh the sector implied move and and a lot of cool stuff that we've gone over in prior shows on the dashboard. So this can do some of the stuff you can do in the dashboard. Uh not quite all of it. Uh maybe eventually that that'll that'll be the case, but right now it uh just kind of looks at data that you can find in our public APIs. It says current implied earnings move is 6.28%. Now let me grab the realized moves for last eight quarters. Sometimes it actually writes bash commands, which is is really interesting. So it's kind of like writing code on your machine to to get the data. If it's if the command is not explicitly available in the API, it'll it'll write code to do it, which is pretty neat. And the capabilities here are really quite deep in in breadth. So you can have it make you an entire dashboard of analysis. You could say like and I want you to analyze Nvidia stock and make me a dashboard with everything I need to know. And it would essentially it could code up an entire website for you on the fly and show you these things like in in graph form and stuff like that. Like right now it's just kind of in the terminal format, but you could do it in a graph form or or chart and make it look really cool on a website and the possibilities are really quite endless here. Yeah, we've been getting some already some feedback. Some like a guy said he made a complete dashboard and he's using it to get all information and he's paper trading. So it's it is it is pretty crazy. I mean I'm I am still blown away on how how it all works. Yeah, and it's really giving the you know, this is the same data that we have exposed in our APIs that you know, our clients are already using. Uh we're just kind of giving it now we're kind of unlocking the potential with an AI agent. Uh to sort of reason through the data and uh make decisions on the data and present it to you uh in in a very intelligent way. And that's kind of the big the big unlock here. So, the this is a list of all the endpoints that it has access to. So, the live summaries, the live implied and forecast monies, live strike chain. Um it has access to the cores and the IV rank, uh as well as all the historical end-of-day uh and historical intraday data. So, that's why it was able to go and grab the VIX spike on that day and and things like that. So, it can really it can look at the whole market. It can basically it can do stock scans uh because it has access to the the cores endpoint. So, it can do a stock scan using these fields. Um One of my favorite commands uh Oh, sorry, Matt. Do you have a any questions? way, I did get it going. I just needed to restart the command uh program. Like once I got it. So, and then I have it installed and then, you know, we'll write blogs on on I'm still coming up with like see like, you know, changing directories. I don't you know, I find like the command line is very difficult to use for for non uh non uh whatever I'm I'm called a non-technical person. Yeah. Yeah, I mean, it's not for everybody, right? Um it's you know, that's why we have our our different products. We have the the dashboard suite of trading tools where you can do all of this and more in a UI. We have the API if you're uh a a hardcore developer and really into writing programs for this. We have and we also just have announced the hard drive delivery. So, if you're a researcher or you want the data in a hard drive format, um where you don't want to pay the egress fees for an S3 delivery, we can deliver you intraday data going back to 2015 with the hard drive. And then could the Clyde work on that? Um, you know, it it's uh it it can't work on that cuz it's integrated with the API, but I'm sure you could you could kind of write your own program to have Clyde go through the data, um So, what do you think? I mean, like, okay, so we have a dashboard. So, what why would Clyde be better than like using the OATS dashboard? Yeah, that's a great question. So, um you know, the dashboard is kind of a point-and-click interface with lots of charts and graphs and uh visuals to sort of show you uh the things you want to look for. And we have a lot of features in the dashboard. We have a backtester, uh scanner, paper trader. We have an optimizer, which is very advanced charting platform. Um, so we have all these tools. The CLI is a built for a kind of a specific use case where um you have an OATS API token um already or or you you're new and you're you're getting one. Uh and you want to interact with the API through an AI agent like Clyde or Codex. And this workflow is uh really great for some people who um want like kind of like a chat-based interface. You know, I I've been using the CLI a lot. I'll I'll wake up in the morning and I'll just be like, "Hey, uh you know, you give me a summary of the market today. Uh what what are some of the top movers? Um what uh what what's the vol surface looking like on, you know, Tesla?" Uh I'll I'll ask it, you know, these types of questions that I would otherwise have to kind of go into the dashboard and and look for. Does it remember? Does it have a memory like Claude has a memory of what you've used before? Yeah, so that's that's sort of Claude functionality that's built in. So, Claude has memory files. Um Claude obviously has chat history, so um it it does have stuff like that built in. A little bit of that deter is determined by how your Claude is set up or your Codex. But yeah, a lot of the AI agents these days have pretty advanced memory capabilities. Um so, it it can know what your what your goals are and Yeah, so it seems to me like if you're doing the same thing every day, right? You could you could just have a like you could write this program out and then just say just give me the same thing and and like like if you you could get, you know, your alerts, you could get all the things that you normally look at and and program it in. So, that, you know, that seems like a great use case. Yeah, that's actually funny you mentioned that. Uh it's it's something I've been setting up. So, I was going to go over it on on this call. So, the loop command in Claude code, I'm not sure what it is for Codex, but you can run a prompt on a schedule. So, you can have it you can just type {slash} loop and type in the interval, let's say 1 day. And now every day it will run whatever prompt I do next. Um so, I can type the O ruts skill and vote command and say summarize the market, show me the stocks making largest moves, stocks with upcoming earnings, and mispriced wall services. And it'll actually just schedule this and run it uh every day. Um I'm not going to schedule it now, but let's see what uh let's see what it let's see what it outputs. So, yeah, I think that's a really cool feature and and it's definitely something a great use of the CLI Matt, like you like you said. Yeah, I mean like you know, for again the non-technical like me like what you know, would I get out of the CLI and what it is is you can customize it for your your needs. Like and you can make it make exactly like a lot of people like well, you know, we have this dashboard product people say could I customize this this this this I'm like no, you can't you know, like we have a public dashboard that we have thousands of of subscribers and we can't just like uh customize it but you know, it sounds like with CLI you're going to be able to do your customizations. Yeah, that's exactly right. Yeah, this is very personalized very custom, you know, it's it starts with an open chat. So you can you can ask it anything and Claude will find a way to to present it to exactly how you want. So that's how I like to use it sometimes when I'm like how this is going to be too too much to do on the dashboard or too too hard to use or um it's going to take too much time. I'll just go in here and kind of let it work in the background go back to what I was doing and um and just kind of wait for it. So it's it's great for people who are I guess short on time or or don't want to go through the dashboard um you know, the dashboard has some things that you know, you have to learn and there are lots of tabs in there and it's it's kind of a power user type of type of interface um but you you can get to uh a lot of the same answers in the CLI. Nice. Yeah, so yeah, like you said there's there's a lot of Claude or code specific features that allow you to do really custom things um and I love watching it sort of work through because uh you can tell when it runs into issues and then it just kind of solves them. So that's what the skill helps it do it helps it understand um the data and understand uh how to how to parse it basically and so, Claude's familiar with writing code and programs, it's essentially writing these code and programs on the spot. So, here's our market snapshot. Yeah, that's that's cool. I think it it is going to be helpful just to show the exact steps for uh someone like me that actually gets set up using command line that, you know, I always thinks like a dumb way to do it. Things that Yeah. you know, but like, you know, it's if we could if we could get that done, I think it'd be really cool. Yeah, no, I think I think that'd be a great blog post to write, so yeah, we'll do it. Um we are planning on coming out with more capabilities on this at at some point. We're going to see kind of what the uptake is and we're going to listen to you guys. If you guys have any features that you'd like to add, we did integrate all of the APIs that we could in the uh data API and the intraday API, but if you guys have anything particular that you'd like to see added in here, like paper trading or something like that, please let us know and we'll we'll see what we can do. Um Yeah, that's going to be nice, you know, like when we have other things like, you know, like we have an alerts endpoint, right? Like that we could, you know, like potentially call as an API, right? And in you could probably do that in native CLI, but, you know, we if we have endpoints, that just makes it a lot easier. You know, we've already designed it. Like we already have a lot of the like for paper trading, for example, we have a lot of the methodologies in there. For example, like when do you assume a fill? Like how wide is the market? Where's the SMB versus the market? We we work on the percentage chance of a fill to make it more realistic to the market. I mean, that's you couldn't I mean, that'd be really hard for you to do. You could do it, I guess, but, you know, Orats has already thought of that. You you I'm a ex-market maker. We have you know, 50 years of experience at Orats right now. So, I mean, we have a lot of things that the API endpoints could really help you use, and you know, by calling it, you get all that information and IP in a single call. Mhm. No, yeah, I I think that's a great example. We know we're kind of blending the power of what Orats can do, right, in the paper trader and in all of our other APIs that we offer, and kind of giving them giving that power to an AI agent that makes it super easy for you guys as the clients of the data to interact with and understand and really make a lot of sense of the market in in a way that I was is is very powerful. Do you I mean, what's the difference between like auto up there and a CLI? Yeah, that's a great question. So, auto does have some of the same capabilities, especially auto in the dashboard can query some of the data. But, it's definitely it's a lot slower, and it's it's not as it's not going to be as skillful, I guess, cuz we have a skill file in here and that's very robust, and kind of catches a lot of errors and mishaps that you could do with with the data and understanding the data. Auto also doesn't write custom code. So, auto will just try and and find the endpoint. But, if the endpoint isn't there, it's not going to be able to do your request. So, Claude can convert your request into a very a very nice snapshot, and it can Claude can write code. You guys talk about skills a lot. What is I mean, like what does that mean? Yeah, so a skill is basically the bridge between an API and an AI agent's understanding of the API. So, if we were to just feed uh Claude the documentation, um it it wouldn't be able to get very far because it doesn't have any context on what um put value means, right? So, we basically have to teach it everything we know about ORATS and how how ORATS works, how we calculate things, what the SMV is, and and those types of questions. Uh so, it it needs to have the context of ORATS in order to deliver you a really great understanding and interpretation of the data. But, I mean, Auto has that, right? A context, but a skill like what you get just get a little bit more nuance on on on that. Yeah, so a skill is is a AI agent-specific um like I guess a terminal AI agent-specific thing. So, it's it can be invoked with what's called a slash command. So, you type slash and then you type the name of the skill that you're uh that you want to invoke. So, if I want to invoke the ORATS skill, I would type slash ORATS. it, so you got to move it up a little bit. Okay. Um I type the type the ORATS skill, um the Do you see it now? Yeah. Um it type the ORATS skill and then uh it that basically tells Claude to uh to read the ORATS methodology. It kind of um takes you through what It it it's kind of um putting the ORATS context in front of Claude before you ask it a question. Uh so, it avoids any potential um errors in API calling and errors in understanding of the data. Um it teaches it what all the data definitions are. Uh and a skill is just it's just markdown files. Uh it's just telling Claude to, "Hey, read these markdown files uh before you start writing code and executing commands. Okay, so the the big difference is that the CLI could write code, whereas the auto just returns the the information. Yeah, that's that's a very big difference. Yeah, the the CLI is on it's on your own machine, so it's using your cloud or codex account and then writing code and and displaying very custom stuff to you. This is a lot faster than auto and a lot more custom cuz it can write code and and kind of execute code. This can make you a whole a whole dashboard and you just open it up in a web browser and and look at it. Like you you can't do that with auto. So, what what are what if you run out of things that that do you have any ideas like for you know, for people is there like on that we have a page, right? A web page dedicated to the CLI. Does it have ideas for what you know, what we want, you know, what what some ideas are for them to ask? Yeah, yeah, right here. So, you can ask it miss price wall, ask it to replay replay days. You can ask it about earnings. You can ask it to analyze the skew of a certain stock. Is that actually working or is that just like a This is just a replay, but yeah, the the actual command would yield something similar to this. Yeah. All right, yeah, that's Yeah, you can ask it to compare IV versus history or versus the forecast. But yeah, I mean what I like to do is I like to just I like to just ask what what are some things you can help me with. And you know, that that's you know, that you could just do that and it's it's super fast and um it it can do market scans, uh, earnings analytics, the deep dives into single name tickers. Um, yeah, I mean it can it can even make you a backtest if if you want. Um so you know, that these these are really kind of advanced things that you could do. Uh, but I'm I'm sure you know, what I what I'm excited about is the community kind of using this and really pushing it to its limits. Yeah. Uh, cuz AI is almost infinitely capable uh, and so you can you can do a a bunch of stuff with this. Yeah, let let me just read something. Hopefully this guy doesn't He goes, "What I've accomplished with Orats is remarkable. I'm in my second OPEX option expiration month. My paper trading results have been beyond my expectations. My intent is to paper trade for another four cycles before committing real money. Uh, we were pulling data from various sources and getting open interest and and other things. The CBOE wanted 8300 and these other places like massive uh said, "Nope." He He goes, uh "Some of these are unbelievable, but now I have Orats in my brokerage and for the current day option chain." So, this guy's really happy so that's really cool. You know, I mean You know, I'm a fossil like it's incredible that it's come this far. Uh, you know, what what we're able to do now. So, that's really cool. Yeah, and you can just get go to the orats.com/ai agents page and these are the download instructions. We'll write a blog post on how to do this in more detail. Yeah. But yeah, it works with your existing API token. If you don't have one, um, you can get one here. Um, and that's Yeah, leave a like comment below and and say what you want to see cuz you know, we're we're always trying to think about what is best for all you guys to to see. So, let us know. Give us comment. That would be That would be fantastic. Yeah, I'd I'd love to see the comments on this. So, all right. Yeah, it looks like we're we're about out of time. So, thanks Matt and I hope you guys enjoyed the demo. If you guys have any feedback, yeah, comment email support@orats.com. And and we'll we'll have a blog post up shortly. So, thank you guys. Thank you tastytrade for hosting. Thank you Matt and we'll see you again next week. Nice job, Tyler. Thanks. You know how the financial world goes. Here comes the buzzkill. Please make sure to take some time to read this boring disclaimer. We will owe you one.