New: Bot Check. $500 for the first 5, then $750.
Send me your Polymarket or Kalshi bot. In 48 hours you get a written report on what can sink your account:
- stale prices
- no size caps
- paying fees twice
- P&L that doesn't match the exchange
Credited toward a build. DMs open.
@JeffWhiting8338 Real fills for me: Polymarket US, 31 days live. Different horizon (month and year-end BTC, priced off Deribit), but one thing carries over. Resting maker orders fill most when you are wrong. My paper bot looked great until I measured markouts after each fill. Do you track those?
New high: +$70.28.
Day 31 of my Bitcoin bot trading real money on Polymarket. That's +76% on the $92 it started with.
It now picks up new markets on its own: 58 two days ago, 71 today.
Still tiny on purpose.
@DanayiCapital Agreed, that's the honest read. Losing ~$5 on $294 once you cap at what was resting says the paper EV was mostly book depth the sim didn't have. I'd keep logging depth on every call. Then you know the real size ceiling before a dollar goes in.
@DanayiCapital That's the right way to look at it. +0.2% without the one hit is basically flat, and paper never tells you the book was 32 cents deep. Next step I'd take: run it live at tiny size and score every call at the price you actually got. What size are you thinking?
@_21Nad Smart to price it out first. At $5 the fees and the spread are most of the game, so I'd build the exposure cap and kill switch before the strategy. Every bot I run has both. What are you planning to trade, the 5-min crypto markets?
@0xRekless Nice start. On 5-min markets I'd watch fills more than win rate. My NBA bot in paper flipped sides 355 times in one close game until I capped it per game. Is this live money or paper, and does it cap how much it can hold per window?
@ValueManIsrael Paper fills are the usual liar. One I found: Polymarket's REST book ran ~10s behind the live stream. A bot reading it "wins" on prices that are already gone. Was your paper filling off the REST book or the stream, and at the ask or the last trade?
@polytutor Good questions. It never crosses the spread, it only posts quotes, so the number I watch is markout: where the price sits 2 minutes after each fill.
And yes, it's lumpy. About half the +71% came from 2 days. Agree 2-3x is the more honest test than 10x.
30 days of my Bitcoin bot trading real money on Polymarket.
+71%. That's +$65.59 on a $92 bankroll. Flat the last 2 days.
It prices Bitcoin markets off the options market. Hard loss stop at -$15.
Won't hold at 10x size, the order flow is too thin. Proving it works first.
I'm on Polymarket US, so there's no chain to check. The exchange API is the only source.
So the bot keeps its own fill ledger and the loss stop runs off that. A trip writes a stop file that survives restarts. A weekly job reconciles the ledger against the exchange's trades, fill by fill. Last check: 5 cents off.
@RoundtableSpace 38 trades an hour for 202 days is ~184,000 trades. $334,927 over that is about $1.80 a trade.
Other versions of this story say $96 a trade, which would be ~3,500 trades. Both can't be true.
On 5-min BTC markets the fill price matters more than the signal anyway.
Not 100 agents.
One bot. One ledger. Every fill checked against Polymarket's own records, every week.
The viral version: AI agents shipping while you sleep. Mine trades while I sleep too.
The difference: when my ledger and the exchange are a few cents apart, I find out why.
@cobra_wise1 I build these. Picks feed in, orders out on Polymarket, with size caps and a kill switch so one bad feed can't drain the account. I do it as a fixed-price build on your own account. Which Polymarket are you on, US or international? DMs open.
My NBA bot went 39-13 and nearly gave it all back in one night.
Heat at Clippers, 117-116. The lead kept flipping, and so did my bot: 355 trades, -$364.
The fix: max 600 shares a game. Same 52 games, replayed: +$107 → +$542.
Paper trading, simulated fills.
3 rules my trading bot can't break:
1. Never trade on a price that's 10 seconds old
2. Never trust its own P&L until the exchange agrees
3. Never forget a loss limit after a restart
Each one exists because I found that bug the hard way.
What's on your list?