Getting Started with Turbine Studio
Turbine Studio turns a sentence like "fade BTC momentum on 15-minute Kalshi markets" into a live trading bot.
You describe what you want. Studio writes the code, runs a backtest on the last 30 days, and deploys to cloud infrastructure you don't have to manage. The bot trades with your keys, through your wallet, on your account. You watch, iterate, retire, repeat.
This post walks through what that looks like in practice.
Signing In
Turbine uses email for sign-in. Enter your email, confirm a code, and you're in. No seed phrase, no MetaMask popup, no wallet extension to install.
Under the hood, signing in provisions a smart wallet for you through Locus, a Y Combinator company. The wallet is yours. Turbine never has access to the keys. You don't need to think about any of that on day one.
What $49 Buys You
Turbine is one price: $49 per month. That includes:
- Cloud infrastructure to run your bots 24/7, under your own account, in an isolated sandbox
- Unlimited strategies (you decide how many bots to run against your capital)
- Every supported venue (Kalshi live today; more coming)
- Every
edge.*data field (Coinbase spot data, NWS weather observations, derived indicators) - The AI strategy builder itself, including the credit budget for model calls
- Backtest replays against historical orderbook data
The first seven days are free. Cancel any time.
Describing a Strategy
Open Studio and type what you want. A few things work well as starting prompts:
- "Market-make on Kalshi weather markets with a 3-cent spread"
- "Buy YES on any political market that drops below 20 cents"
- "Fade the consensus, sell YES when a market goes above 85 cents, buy YES when it drops below 15 cents"
- "Build a 15-minute Bitcoin mean-reversion bot"
Studio reads your description, generates the bot code, shows it to you, and offers a backtest. You can iterate as much as you want before committing: change parameters, add conditions, cap position sizes, constrain which markets the bot touches.
The point of this step is to see the code Studio wrote, in plain Python, and know exactly what the bot will do. It's not a black box.
Running the Backtest
Before deploying, Studio runs the strategy against the last 30 days of Kalshi orderbook data. You get a return number, a win rate, a Sharpe ratio, a max drawdown, and an equity curve.
Read the backtest honestly. A high return number on a strategy with a 38% win rate is not the same as a modest return on a 62% win rate. The first is fragile. The second is durable.
If the backtest looks wrong (negative return, unstable equity curve), iterate on the strategy in chat. Studio will regenerate the code and re-backtest.
Deploying
When the backtest looks right, you deploy. One click.
The bot goes live on your isolated runner, managed by Locus. It trades through your Kalshi API key (you paste it once; the key stays encrypted inside your runner, scoped to your trading account). Turbine never touches your funds.
Your bot runs 24/7. You can close your laptop, go to sleep, take a vacation. It keeps trading until you stop it.
What Happens After
Most people iterate after the first deploy. They watch logs for a day, notice something the bot is doing that they didn't expect, and go back to Studio. "Make the spread wider." "Add a max daily loss of $100." "Skip the first hour after resolution." Each change is a conversation with Studio, a re-backtest, a redeploy.
That loop is the whole point. Before Turbine, a trader with a new idea had a choice: learn Python and build the bot yourself over two weekends, or hire someone to build it and wait a month. Either way, by the time the strategy was live, the idea had cooled.
Turbine compresses that loop to minutes. Idea, backtest, deploy. Watch. Iterate. Retire when it stops working.
Start your seven-day free trial.
This post is for informational purposes only and does not constitute financial advice. All strategies described carry risk of loss. Never trade with money you can't afford to lose.