Quick Answer
Claude Cowork is a desktop app from Anthropic (also known as the Claude desktop app) that operates within a local folder on your computer. To use it for TradeStation EasyLanguage coding:
- Create a local folder containing the three official TradeStation EasyLanguage PDF manuals, plus your reference indicator code
- Point Claude Cowork at that folder
- Use this system prompt: “You are my TradeStation EasyLanguage coding assistant. Use the EasyLanguage manuals and my example code in this folder as the source of truth for syntax and style.”
- Describe the indicator you want in plain English
- Paste the generated code into TradeStation’s Development Environment and verify
The result: clean, verified EasyLanguage code on the first attempt – no hybrid Python syntax, no debugging loop. I built three indicators live in the video above. All three verified first try.
Table of Contents
Use the links above to jump to the Claude Cowork EasyLanguage topics that interest you.
Why Every Other AI Tool Fails at EasyLanguage
Over the last nine months, every time a new AI model dropped – ChatGPT, Claude, Grok, Gemini, Perplexity – I tested it on TradeStation EasyLanguage code. The results were always the same: reasonably promising, but never clean. You’d end up with hybrid code that looked like EasyLanguage but had Python syntax baked in. Verification errors. Four to six rounds of debugging before you finally hit the green tick.
The reason is straightforward. When you ask any AI model in a standard chat interface to write EasyLanguage, it’s working from general training data. It has seen some EasyLanguage. It knows Python. It knows C++. So it blends them. It guesses at syntax it doesn’t know and fills the gaps with what it does know.
Claude Cowork solves this by giving the model the actual documentation to reference. When it writes a function call, it checks the EasyLanguage Functions manual. When it writes array syntax, it checks the Reference Guide. When it calls Better Momentum, it checks my working indicator code. It’s not guessing – it’s referencing.
That’s the entire trick. And it changes everything.
What Is Claude Cowork?
Claude Cowork is part of the Claude desktop app from Anthropic, separate from the Claude web browser interface and Claude Code. It’s installed as a standalone application on your Mac computer (or Windows machine).
There are three modes in the Claude desktop app:
- Standard chat – same as the web version, nothing special for coding purposes
- Claude Code – powerful, but context windows are too small for large projects in my experience; things get stuck
- Claude Cowork – the one that works for this use case. You point it at a local folder, it reads everything in that folder, and uses it as the source of truth for every task you give it
People searching for the “Claude desktop app” or “Claude desktop” are often looking for exactly this. Cowork is the mode within the desktop app that enables folder-based coding workflows. Get Claude Cowork here.
Setting Up Your Folder
Create a local folder on your desktop – I called mine “TradeStation EasyLanguage.”
Critical: this must be a local folder, not a cloud folder. If you try pointing Claude Cowork at a Google Drive folder, the back-and-forth between the cloud and the app breaks down completely. It needs to read from your local drive.
Into this folder, load four files:
The three official TradeStation EasyLanguage documentation PDFs:
- EasyLanguage Reference Guide (~8MB)
- EasyLanguage Essentials (~7MB)
- EasyLanguage Functions (~3MB)
Download these from TradeStation’s developer resources section. These are the most comprehensive documentation files TradeStation has released, covering syntax, functions, reserved words, and examples in full.
Your reference indicator code:
If you’re a subscriber to the TradeStation version of my Better Indicators, you have access to a Google Doc showing exactly how the functions are called, with the full code for Better Momentum, Better Pro Am, and Better Sine Wave. Download that as a PDF from Google Docs and add it to the folder.
This is the piece that makes the setup specific to my indicators. By including real working code alongside the manuals, Claude learns the exact function signatures, input parameters, and coding style – so when you ask for something new using those functions, it already knows how they work.
Connecting the Folder to Claude Cowork
Open Claude Cowork. Go to Work in a Folder > Choose a Different Folder, navigate to your desktop, select the TradeStation EasyLanguage folder, and click Open.
Claude will scan everything in the folder and confirm what it found. From this point on, every piece of code it writes is grounded in your documentation.
The System Prompt
Before asking for any code, send this prompt:
Prompt: You are my TradeStation EasyLanguage coding assistant. Use the EasyLanguage manuals and my example code in this folder as the source of truth for syntax and style. When writing EasyLanguage code, add plenty of comments to explain how the code works. Can you help me write EasyLanguage code?
Select the most capable model available – currently Claude Opus 4.6.
Claude’s first response will be to scan the folder and confirm what it’s working with. After that, you’re ready to build.
Three Indicators Built Live
Indicator 1: Show Me Exhaustion Momentum
Prompt: I want to plot a large dot when you see an exhaustion momentum pattern using the Better Momentum indicator.
Claude read the Better Momentum function signature from the reference PDF, verified the syntax against the EasyLanguage manuals, and generated the indicator:
- Calls the Better Momentum function and checks for exhaustion bull and exhaustion bear output flags
- Plots a green dot below the bar for bullish exhaustion
- Plots a red dot above the bar for bearish exhaustion
- Uses a 20-period average range from the reference code to scale the dot offset with volatility
Pasted into TradeStation Development Environment. Hit Verify. Saved and verified, first try.
In previous attempts with other AI tools, reaching a verified indicator required four to six rounds of debugging. This was one shot.
To apply to a chart: select the indicator, go to Inputs, untick “Update Value Intrabar” (TradeStation enables this by default – always turn it off), set Style to Points, set point size large, scaling to Price Bars.
Indicator 2: Show Me Exhaustion + Big Pro Combo
Prompt: Add another condition: plot a large dot only when Better Momentum shows exhaustion AND Better Pro Am shows Big Pro bars within ten bars of each other. Color bullish exhaustion red and bearish exhaustion white.
Before writing a line of code, Claude pulled the Better Pro Am function signature from the reference PDF to confirm the exact call. It then explained the logic:
“Every bar, both functions run. When a Big Pro bar or exhaustion signal fires, it records the bar number. The combo triggers only on the bar where one signal just fired AND the other occurred within the ten-bar lookback. The dot appears on the bar that completes the pair.”
That’s exactly the logic I asked for. Code generated. Saved and verified, first try.
Applied to a gold chart with 1,000-tick bars: correctly filtered out exhaustion signals with no nearby Big Pro bar activity, and only plotted where both conditions aligned.
Indicator 3: Better Momentum Divergence
This one came from what I was seeing in the market. Equity prices had been weakening for two weeks, but Better Momentum was showing an unusually strong pattern – a divergence. I wanted to quantify it and find historical instances of the same setup.
Prompt: Code a new indicator measuring the divergence between price and the Better Momentum output. Use the Better_Momentum function output called ‘momentum’ and a ten-bar lookback of momentum compared to closing price. Output a line plot – line red, zero line white.
Claude flagged a design decision before generating code:
“Close price and momentum are on completely different scales – you can’t just subtract them. I’ll normalize both as z-scores over the lookback, then subtract. When momentum leads price, the line goes positive. When price leads momentum, it goes negative. Zero is alignment.”
There are different approaches to doing this, and the one Claude chose is definitely valid. But with more time, I definitely would refine the approach and try different divergence measurement algorithms.
Converting both series to z-scores (measuring each change in units of its own standard deviation) makes the comparison meaningful. The code verified all seventeen Better Momentum parameters in the correct order and included division-by-zero guards in the z-score math.
Saved and verified, first try.
Applied to a daily chart: currently reading an outlier signal – price moving down while momentum is moving up. Looking back at historical instances, similar spike readings preceded significant moves. That’s the edge I was looking for.
What’s in Your Folder After a Session
Every indicator Claude generates gets saved as a text file in your working folder automatically. After this session, the folder contained:
ShowMe_Exhaustion_Momentum.txtShowMe_Exhaustion_BigPro_Combo.txtIndicator_Momentum_Price_Divergence.txt
You have a permanent record of every indicator built, along with the documentation that informed it. That’s a growing asset – the longer you use the setup, the richer the reference library becomes.
Setup Summary
| Step | Action |
|---|---|
| 1 | Create a local folder on your desktop |
| 2 | Add the three TradeStation EasyLanguage PDFs |
| 3 | Add your reference indicator code as PDF |
| 4 | Open Claude Cowork > Work in a Folder > Select your folder |
| 5 | Paste the system prompt, select Claude Opus 4.6 |
| 6 | Describe the indicator you want in plain English |
| 7 | Copy the generated code into TradeStation Development Environment |
| 8 | Verify – expect green ticks first try |
Frequently Asked Questions
Does this work on Windows as well as Mac?
Yes. Claude Cowork is now available for both Mac and Windows. The setup process is identical – create a local folder, load your documentation files, and point Cowork at it.
Which Claude model should I use?
Use the most powerful model available in Cowork. At the time of writing, that’s Claude Opus 4.6. More capable models produce cleaner code and handle complex multi-function indicators better.
Do I need to be a Better Indicators subscriber to use this?
No. The folder setup works with any EasyLanguage reference code you have. The Better Indicators PDF just gives Claude a working example of my specific function calls. If you have other indicator code you want Claude to learn from, add that instead.
What if Claude generates code that doesn’t verify?
Paste the error message from the TradeStation Development Environment back into Claude Cowork and ask it to fix the specific error. With the documentation loaded, it will reference the correct syntax and resolve it quickly. This is significantly faster than debugging manually.
Can I use this for TradeStation strategies, not just indicators?
Yes. The EasyLanguage documentation covers strategies as well as indicators. Describe the entry and exit logic you want, and Claude will apply the same process – reference the manuals, write clean code, and output a text file ready to paste into the Development Environment.
Final Thoughts
I’ve been trading futures and building TradeStation indicators since the early 2000s. Getting clean EasyLanguage code from AI has been a problem I’ve been working on for the better part of a year. This is the solution.
The trick is simple: don’t ask AI to guess at EasyLanguage syntax. Give it the manuals. Give it working examples. Let it reference those instead of its training data. Claude Cowork makes that workflow seamless because the folder environment is built for exactly this kind of reference-based task.
Three indicators built live in the video. All verified. None required editing.
If you’ve found other ways to code TradeStation EasyLanguage with AI, I’d like to hear about them. And if you want to try this yourself, the Better Indicators reference code is available to TradeStation subscribers – details here.
Related: TradeStation EasyLanguage and AI: How to Build Better Indicators Faster
