Bots that show you a screenshot ledger.
I build automated trading and market-making systems for your own account: live order-book streams, a fill ledger reconciled to the exchange to the penny, and circuit breakers that stop the bot before you have to. Every strategy gets recorded, replayed and paper traded before it touches real money.
Most bots are sold on a P&L screenshot.
Mine are delivered with a ledger.
A screenshot can't tell you whether the profit is real, or if it's luck, stale prices or a bug quietly ignoring positions. The only way to know is to measure every fill against the exchange and kill the ideas that don't hold up.
The usual "trading bot"
- Backtest on prices you could never have traded
- Reads the public REST book, seconds behind the market
- P&L from its own marks, never checked against the exchange
- One kill switch, if any, that resets on restart
- Black box you rent; the seller keeps the code
What you get here
- Record the real market first, then replay against it
- Live WebSocket order book and trade stream
- Fill-by-fill reconciliation with the exchange's own records
- Drawdown, exposure and stale-feed breakers that survive restarts
- Your account, your keys, your code, all handed over
Start small, or build the whole thing
Fixed scope, fixed price. Builds are 50% to start, 25% at a milestone and 25% on delivery. An Edge Study's fee is credited toward any build that follows it.
Edge Study
Before you build anything: is there money in the idea? I record the market, replay it, and tell you in writing whether it holds up.
- Recorder on your target markets, live stream
- Replay of your strategy against real fills, queue and delay
- Markouts: where price went 30 s, 2 min, 5 min after each fill
- Written findings, including what didn't work
- Fee credited toward a build
Starter Bot
A proven resting-order engine, set up on your own account for one venue and tuned to your markets and your limits. Live in weeks, not months.
- Polymarket US or Kalshi, one strategy
- The same engine as the reference build: live stream, fill ledger
- Loss stop, exposure cap and stale-feed breakers
- Two weeks of paper trading before any real money
- Daily P&L summary; typically 2–3 weeks to live
Trading System Build
A production trading or market-making system built around your own pricing model, any supported venue.
- Exchange API + live order-book stream integration
- Fill ledger reconciled against exchange records
- Drawdown, exposure and stale-feed circuit breakers
- Paper mode, replay testing and a findings log before real money
- Cloud deployment, CI/CD, monitoring and runbooks
Keep it running
Hosting, monitoring, exchange reconciliation and fixes when APIs or markets change. Cancel any month.
The numbers, including the boring ones
What my own systems are doing right now. Real money and paper are labeled, and the ideas that failed stay on the list.
Bitcoin market maker
- Ledger vs exchangeΔ $0.00
- Fills855
- Markets quoted58 / 58 matched
- Bankrollsmall, under $250
NBA one-sided maker
- Fills / shares655 / 7,734
- Price move 2 min after fills−0.07¢
- Readmostly favorites that won
- Statuspaper until 30+ games
Ideas that didn't hold up
- MLB speed on scoring playsmarket repriced before the public feed showed the run
- Baseball model between playsedge of ~0.2¢, smaller than the fee
- NBA score modellost to the market after fills
- Two-sided NBA quoting−$170 to −$200 on replay
Last updated Nov 2, 2026. Research and engineering results from my own accounts, not trading advice or a forecast of future results. Weekly recaps on LinkedIn.
Real money is the last gate, not the first
Every strategy has to pass each gate on measured data. Most ideas die at gate 2 or 3, which is far cheaper than finding out live.
Scope
30-minute call, then a written scope: venue, strategy, risk limits, price.
exit: signed scopeRecord
Capture the live book and every trade on your markets. No assumptions about the data.
exit: clean dataReplay
Run the strategy against the recording with queue position and reaction delay.
exit: markout ≥ 0Paper
Live market, simulated fills, the same code that will trade.
exit: replay ≈ paperSmall live
Small size behind hard caps. The ledger has to match the exchange before size goes up.
exit: Δ $0.00A market maker on Polymarket US, running with real money
My own system, built the same way a client build is. It quotes 58 Bitcoin price markets around the clock, prices them from options data, and is checked against the exchange fill by fill. Alongside it, a research stack tested sports strategies. Here's the log, including the parts that failed.
Findings log (excerpt)
Architecture
Stack: Python, WebSockets, SQLite, Azure VM, Azure DevOps CI/CD. Your build can run on your own cloud account.
Does your idea survive the fee?
Many "edges" are smaller than the cost of trading them. Put in your number.
Taker fee uses Polymarket US's published formula, 0.0695 × p × (1 − p) per contract (about 1.7¢ at 50¢); resting (maker) orders pay none. "Adverse selection" is how far price tends to move against you right after a fill: in my sports research it was about 0.5–2¢ within two minutes. Illustrative only; not trading advice.
Questions people ask first
Will you trade my money or sell me signals?
No. You hold the account and the API keys; the system trades your strategy within the limits you set. I build it, hand it over, and can keep it running, but I don't manage money, take a share of profits, or sell trade signals.
Can you guarantee it will make money?
No one honestly can. What I guarantee is measurement: you'll know from the ledger and markouts whether it does, and the Edge Study exists so you can find out before paying for a full build.
Which venues do you support?
Polymarket US and Kalshi for event contracts, and major crypto exchanges with REST and WebSocket APIs. If the venue has a documented API, it's usually buildable; we confirm in scoping.
Who owns the code?
You do. The repository, deployment scripts and runbooks are transferred on delivery.
How long does a build take?
A Starter Bot is typically 2–3 weeks to live. A custom build is typically 4–8 weeks from deposit to small-live, depending on venue and strategy complexity. Paper trading time is part of the schedule, not an extra.
Do I need a strategy already?
For a build, yes, at least a clear idea of how you'd price the market. If you're not sure there's an edge, start with an Edge Study.
Bring the idea. Leave with a scope and a price.
30 minutes: what you want to trade, where, and how you'd price it. You get a written scope and fixed price within a few days.